Q2 2025 Broadcom Inc Earnings Call

Of Investor Relations of Broadcom, Inc.

Thank you operator, and good afternoon, everyone.

Hock Tan: Joining me on today's call are Hock Tan President and CEO.

Speaker Change: Houston, Spears, Chief Financial Officer, and Charlie cause President semiconductor solutions group.

Hock Tan: Broadcom distributed a press release and financial tables after the market closed describing our financial performance for the second quarter of fiscal year 2025.

Hock Tan: Now let me provide more color. Q2 Semiconductor Revenue was $8.4 billion, with growth accelerating to 17% year-on-year, up from 11% in Q1. and of course, driving this growth. was AI semiconductor revenue of over $4.4 billion, which is up 46% year-on-year and continues the trajectory of nine consecutive quarters of strong growth.

Hock Tan: You did not receive a copy you may obtain the information from the investors section of the Broadcom <unk> website at Broadcom Dot com.

Hock Tan: This conference call is being webcast live and an audio replay of the call can be accessed for one year through the investors section of Broadcom website.

Hawkins: During the prepared comments Hawkins care spin will be providing details of our second quarter fiscal year 2025 years old.

Hawkins: Guidance for our third quarter of fiscal year, 2025, as well as commentary regarding the business environment.

Hawkins: We will take questions. After the end of our prepared comments.

Hawkins: Please refer to our press release today and our recent filings with the SEC for information on the specific risk factors that could cause our actual results to differ materially from the forward looking statements made on this call.

Hock Tan: We didn't do. Custom AI accelerators grew double digits year-on-year, while AI networking grew over 170% year-on-year. AI Network which is based on Ethernet was robust and represented 40% of our AI revenue. As a standards based open protocol, Ethernet enables one single fabric for both scale out and scale up. and remains the preferred choice by our hyperscale customers. Our networking portfolio of Tomahawk Switchers Cherico Routers and Nick. is what's driving our success within AI clusters in hyperscale. And the momentum continues with our breakthrough Tomahawk 6 suite. just announced this week. This represents the next generation. 102.4 terabits per second switch capacity.

Hawkins: In addition to U S GAAP reporting Broadcom reports certain financial measures on a non-GAAP basis.

Hawkins: A reconciliation between GAAP and non-GAAP measures is included in the table attached to today's press release.

Hawkins: Comments made during today's call will primarily refer to are non-GAAP financial results.

Håkan: I will now turn the call over to Hakan.

Håkan: Thank you Qi <unk>.

Håkan: And thank you everyone for joining us today.

Håkan: In our fiscal Q2 2025.

Håkan: Total revenue was a record.

Håkan: $10 billion up 20% year on year.

Håkan: Yes.

Håkan: This 20% year on year growth was all organic.

Håkan: Q2 last year was the first full quarter with Vmware.

Håkan: No revenue was driven by continued strength in semiconductors.

Håkan: And the momentum we have achieve in Vmware.

Håkan: No, reflecting excellent all operating leverage Q.

Hock Tan: Tomahawk-6 enables clusters of more than 100,000 AI accelerators to be deployed in just two tiers instead of three. This flattening of the AI cluster is huge. because it enables much better performance in training next generation frontier models through a lower latency, higher bandwidth.

Håkan: Q2, consolidated adjusted EBITDA was $10 billion.

Håkan: 35% year on year.

Håkan: Now, let me provide more color.

Håkan: Q2 semiconductor revenue was $8 4 billion.

Welcome to Broadcom, Inc. Second quarter fiscal year 2025 financial results Conference call.

Operator: Welcome to Broadcom's Inc. Second Quarter Fiscal Year 2025 Financial Results Conference Call.

Håkan: With growth accelerating to 17% year on year up from 11% in Q1.

Ji Yoo: At this time, for opening remarks and introductions, I would like to turn the call over to Ji Yoo, Head of Investor Relations of Broadcom Inc. Thank you, operator and good afternoon, everyone. Joining me on today's call are Hock Tan, President and CEO. Kirsten Spears, Chief Financial Officer and Charlie Kawwas, President, Semiconductor Solutions.

At this time for opening remarks, and introductions I would like to turn the call over to Jim <unk> head of Investor Relations of Broadcom, Inc.

Hock Tan: and lower power. turning to XP use or customer salary We continue to make Excellent progress on the multi-year journey of enabling our three customers and four prospects to deploy custom AI accelerators. As we had articulated over six months ago... We eventually expect at least three customers to each deploy 1 million AI accelerated clusters in 2027. largely for straining their frontier model. and we forecast and continue to do so, a significant percentage of these deployments to be custom XPU. These partners are still unwavering in their plan to invest. Despite this certain economic environment, in fact, What we've seen recently is that they are doubling down.

Håkan: And of course driving this growth.

Speaker Change: Thank you operator, and good afternoon, everyone.

Håkan: Also AI semiconductor revenue of over $4 4 billion, which was up 46% year on year and.

Speaker Change: Joining me on today's call are Hock Tan President and CEO.

Speaker Change: You're still Spirit's, Chief financial Officer, and Charlie cause President semiconductor solutions group.

Håkan: And continues the trajectory of nine consecutive quarters of strong growth.

Ji Yoo: Broadcom distributed a press release and financial tables after the market closed describing our financial performance for the second quarter of fiscal year 2025. If you did not receive a copy, you may obtain the information from the investor section of the Broadcom's website at Broadcom.com. This conference call is being webcast live and an audio replay of the call can be accessed for one year through the investor section of Broadcom's website. During the prepared comments, Hock and Kirsten will be providing details of our second quarter fiscal year 2025 results. Guidance for our third quarter of fiscal year 2025, as well as commentary regarding the business environment.

Speaker Change: Broadcom distributed a press release and financial tables after the market closed describing our financial performance for the second quarter of fiscal year 2025.

Håkan: Within this customer.

Håkan: Custom accelerators grew double digits year on year, while AI networking grew over 170% year on year.

You did not receive a copy you may obtain the information from the investors section of the Broadcom website at Broadcom dotcom.

Håkan: AI networking, which is based on Ethernet was robust and represented 40% of our AI revenue.

Speaker Change: This conference call is being webcast live and an audio replay of the call can be accessed for one year to the investors section of Broadcom website.

Speaker Change: S. A standards based open protocols.

Hawkins: During the prepared comments Hawkins care spin will be providing details of our second quarter fiscal year 2025 results.

Speaker Change: <unk> enables one single fabric for both scale out and scale up.

Speaker Change: Guidance for our third quarter of fiscal year 2025.

Speaker Change: And remains the preferred choice by our Hyperscale customers.

Speaker Change: Well as commentary regarding the business environment.

Ji Yoo: We will take questions after the end of our prepared Please refer to our press release today and our recent filings with the SEC for information on the specific risk factors that could cause our actual results to differ materially from the forward-looking statements made on the website. In addition to U.S.

Speaker Change: We will take questions. After the end of our prepared comments.

Speaker Change: Our networking portfolio.

Speaker Change: Please refer to our press release today and our recent filings with the SEC for information on the specific risk factors that could cause our actual results to differ materially from the forward looking statements made on this call.

Speaker Change: Tomahawk switches.

Speaker Change: Sherri Coale routers.

Hock Tan: on inference in order to monetize their platform. and reflecting this, we may actually see an acceleration of XPU demand. into the back half of 2026 to meet urgent demand for inference on top of the demand we have indicated from training. And accordingly, we do anticipate now, our fiscal 2025 growth rate of AI Semiconductor Revenue to sustain. into fiscal 2026.

Speaker Change: And Nyx.

Speaker Change: It's what's driving our success within AI clusters in Hyperscale.

Speaker Change: In addition to U S GAAP reporting Broadcom reports certain financial measures on a non-GAAP basis.

Ji Yoo: GAAP reporting, Broadcom reports certain financial measures on a non-GAAP A Reconciliation Between Gap and Non-Gap Measures Included in the table attached to today's press...

Speaker Change: And the momentum continues with our breakthrough Tomahawk six switch.

Speaker Change: A reconciliation between GAAP and non-GAAP measures.

Speaker Change: In the table attached to today's press release.

Speaker Change: As announced this week.

Speaker Change: This represents.

Ji Yoo: Comments made during today's call will primarily refer to our non-GAAP financial I will now turn the call over to... Thank you.

Speaker Change: Comments made during today's call will primarily refer to are non-GAAP financial results.

Speaker Change: Next generation.

Speaker Change: 102.

Speaker Change: 0.4, Terabits per second switch capacity.

Hawk: I will now turn the call over to Hawk.

Speaker Change: Come on six enables clusters.

Hock Tan: Thank you, Ji. and thank you everyone for joining us today. in our fiscal Q2 2025. Total revenue was a record $15 billion, up 20% year-on-year. This 20% year-on-year growth was all organic. as Q2 last year was the first full quarter with VMware. Now revenue was driven by continued strength in AI semiconductors. and the momentum we have achieved in VMWare. Now reflecting excellence of operating leverage.

Hawk: Thank you Qi <unk>.

Speaker Change: More than 100000 accelerators to be deployed in just two tiers instead of three.

Speaker Change: And thank you everyone for joining us today.

Hawk: In our fiscal Q2 2025.

Hawk: Total revenue was a record $15 billion up 20% year on year.

Speaker Change: This flattening of the AI cluster is huge.

Hock Tan: Turning to our Q3 Outlook. As we continue our current trajectory of growth. We focus AI semiconductor revenue. to be $5.1 billion, up 60% year-on-year, which would be the 10th consecutive quarter of growth.

Speaker Change: Because it enables much better performance in training next generation from GM models through a lower latency and higher bandwidth.

Hawk: Yes.

Hawk: This 20% year on year growth was all organic.

Hawk: Q2 last year was our first full quarter with.

Speaker Change: And lower power.

Hawk: Vmware.

Speaker Change: Turning to <unk>.

Hawk: No revenue was driven by continued strength in semiconductors.

Speaker Change: XP use our customer accelerators.

Speaker Change: We continue to make.

Hawk: Semiconductors.

Hock Tan: Now turning to non AI semiconductors in Q2. Revenue of $4 billion was down 5% year-on-year. Non-AI Semiconductor Revenue is close to the bottom. has been relatively slow to recover. but there are bright spots in q2 broadband Enterprise Networking and Server Storage Revenues were up sequentially.

Speaker Change: Excellent progress.

Hawk: And the momentum we have achieve in Vmware.

Speaker Change: On the multi year journey of enabling our three customers.

Hawk: Now, reflecting excellent operating leverage.

Speaker Change: And for prospects to deploy custom AI accelerators.

Hock Tan: Q2 consolidated adjusted EBITDA was $10 billion, up 35% year-on-year. Now let me provide more color. Q2 Semiconductor Revenue was $8.4 billion, with growth accelerating to 17% year-on-year, up from 11% in Q1. and of course, driving this growth. was AI semiconductor revenue of over $4.4 billion, which is up 46% year-on-year and continues the trajectory of nine consecutive quarters of strong growth. We didn't do. Custom AI accelerators grew double digits year-on-year, while AI networking grew over 170% year-on-year. AI Network which is based on Ethernet was robust and represented 40% of our AI revenue. As a standards-based open protocol, Ethernet enables one single fabric for both scale out and scale up.

Hawk: Two consolidated adjusted EBITDA was $10 billion up 35% year on year.

Speaker Change: As we had articulated over six months ago.

Speaker Change: We eventually expect at least.

Hawk: Now, let me provide more color.

Speaker Change: Three customers to each deploy 1 million accelerated.

Hawk: Q2 semiconductor revenue was $8 4 billion.

Speaker Change: Accelerated clusters in 2027.

Hawk: With growth accelerating to 17% year on year up from 11% in Q1.

Hock Tan: However... Industrial was down and as expected, wireless was also down due to seasonality. In Q3, we expect enterprise networking and broadband to continue to grow sequentially. But server storage, wireless, and industrial are expected to be largely flat. And overall, we forecast non-AI semiconductor revenue to stay around $4 billion.

Speaker Change: Largely for screening their frontier models.

Speaker Change: And we forecast and continue to do so a significant percentage of these deployments.

Hawk: And of course driving this growth.

Hawk: So it's for me.

Speaker Change: To be custom XP use.

Hawk: Conductor revenue of over $4 4 billion.

Speaker Change: These partners.

Speaker Change: Still unwavering India plan to invest.

Hawk: Which was up 46% year on year.

Hawk: And continues the trajectory of nine consecutive quarters of strong growth.

Speaker Change: Despite these certain economic environment.

Speaker Change: In fact.

Speaker Change: What we've seen recently.

Hawk: Within this tough.

Hawk: Custom accelerators grew double digits year on year, while E. R M.

Speaker Change: Is that they are doubling down.

Speaker Change: On inference in order to monetize their platforms.

Hock Tan: Now let me talk about our infrastructure software segment. Q2 infrastructure software revenue of $6.6 billion was up 25% year-on-year, above our outlook of $6.5 billion. As we have said before, This growth reflects our success in converting our enterprise customers from perpetual vSphere to the full VCF software stack subscription. Customers are increasingly turning to VCF to create a modernized private cloud on-prem, which will enable them... to repatriate workloads from public clubs. while being able to run modern container-based applications and AI applications. Of our 10,000 largest customers, over 87% have now adopted VCF. The momentum from strong VCF sales over the past 18 months Since the acquisition of VMware, has created annual recurring revenue, or otherwise known as ARR, growth of double digits in our core infrastructure software.

Hawk: Networking grew over 170% year on year.

Speaker Change: And reflecting this we may actually see an acceleration of <unk> demand.

Hawk: AI networking.

Hawk: His base on Ethernet.

Speaker Change: Into the back half of 'twenty 'twenty six to meet urgent demand for inference on top of the demand we are indicated for <unk>.

Hawk: <unk> robust and represented 40% of our AI revenue.

Hawk: S. A standards based open protocols Ethan.

Speaker Change: From training.

Hawk: <unk> enables one single fed break for both scale out and scale up.

Speaker Change: And accordingly, we do anticipate now our fiscal 2025 growth rate.

Hock Tan: and remains the preferred choice by our hyperscale customers. Our Networking Portfolio of Tomahawk Switchers Cherico Routers and Nick. is what's driving our success within AI clusters in hyperscale. And the momentum continues with our breakthrough tomorrow, six weeks. just announced this week. This represents the next generation. 102.4 terabits per second switch capacity. Tomahawk-6 enables clusters of more than 100,000 AI accelerators to be deployed in just two tiers instead of three. The flattening of the AI cluster is huge. because it enables much better performance in training next generation frontier models through a lower latency, higher bandwidth. and lower power.

Hawk: <unk> remains the preferred choice by our Hyperscale customers.

Speaker Change: AI semiconductor revenue to sustain it.

Speaker Change: Into fiscal 2026.

Hawk: Our networking portfolio.

Hawk: Tomahawk switches.

Speaker Change: Turning to Q2, our Q3 outlook.

Hawk: Jerry call routers.

Speaker Change: We continue our current trajectory of growth.

Hawk: And Nick.

Hawk: It's what's driving our success within AI clusters in Hyperscale.

Speaker Change: We forecast a semiconductor revenue.

Hawk: Okay.

Speaker Change: To be $5 one.

Hawk: The momentum continues.

Speaker Change: $1 billion.

Speaker Change: Up 60% year on year.

Hawk: Our breakthrough Tomahawk six suites.

Speaker Change: Which would be the 10th consecutive quarter of growth.

Hawk: Just announced this week.

Hawk: This represents.

Hawk: Next generation.

Speaker Change: Now turning to non AI semiconductors in Q2 revenue of $4 billion was down 5% year on year.

Hawk: 102 point.

Hawk: 0.4, Terabits per second switch capacity.

Hawk: Come on six enables clusters of more than 100000 accelerators to be deployed in just two T is instead of three.

Speaker Change: Non semiconductor revenue is close to the bottom.

Speaker Change: And it's been relatively slow to recover.

Hock Tan: in Q3. We expect infrastructure software revenue to be approximately $6.7 billion, up 16% year-on-year. So in total, we're getting Q3 consolidated revenue to be approximately $15.8 billion, up 21% year-on-year. We expect Q3 adjusted EBITDA to be at least 66%.

Speaker Change: But there are bright spots in Q2 broadband.

Hawk: This flattening of the AI cluster is huge.

Speaker Change: Enterprise networking and server storage revenues were up sequentially.

Hawk: Because it enables much better performance in training next generation frontier models through.

Speaker Change: However.

Speaker Change: Industrial was down and is a N S. Expectant wireless was also down due to seasonality.

Hawk: Lower latency and higher bandwidth.

Hawk: And lower power.

Hock Tan: turning to Expeos, or Custom Accelerator. We continue to make Excellent progress on the multi-year journey of enabling our three customers and four prospects to deploy custom AI accelerators. As we had articulated over six months ago... We eventually expect at least three customers to each deploy 1 million AI accelerated clusters in 2027. largely for framing their frontier model. And we forecast, and continue to do so, a significant percentage of these deployments to be custom XPEs. These partners are still unwavering in their plan to invest. despite this certain economic environment. In fact. What we've seen recently is that they are doubling down.

Hawk: Turning to <unk>.

Speaker Change: In Q3.

Hawk: XP use our customer accelerators.

Speaker Change: We expect enterprise networking and broadband to continue to grow sequentially.

Hawk: We continue to make.

Hawk: Excellent progress on the multiyear journey of enabling our three customers.

Kirsten Spears: With that, let me turn the call over to Kirsten. Thank you, Hock. Let me now provide additional detail on our Q2 financial performance. Consolidated revenue was a record $15 billion for the quarter, up 20% from a year ago. Gross margins was 79.4% of revenue in the quarter, better than we originally guided on product mix. Consolidated operating expenses were $2.1 billion, of which $1.5 billion was related to R&D. Q2 operating income of $9.8 billion was up 37% from a year ago with operating margin at 65% of revenue. Adjusted EBITDA was $10 billion or 67% of revenue, above our guidance of 66%.

Speaker Change: But server storage wireless and industrial.

Speaker Change: <unk> to be largely flat.

Hawk: And fall prospects to deploy custom AI accelerators.

Speaker Change: And overall, we forecast non semiconductor revenue to stay around $4 billion.

Hawk: As we had articulated over six months ago.

Speaker Change: Now, let me talk about our infrastructure software segment.

Hawk: We eventually expect at least three.

Hawk: Three customers to each deploy 1 million accelerated.

Speaker Change: Q2 infrastructure software revenue of $6 6 billion was up 25% year on year above our outlook of $6 5 billion.

Hawk: Accelerated clusters in 2027.

Hawk: Largely for screening their frontier models.

Hawk: And we forecast and continues to do sold a significant percentage of these deployments.

Speaker Change: As we have said before.

Speaker Change: This growth reflects our success in converting our enterprise customers from perpetual vis fear.

Hawk: To be custom XP use.

Hawk: These partners.

Speaker Change: The full Vcs software state subscription.

Hawk: Steel unwavering in their plan to invest.

Speaker Change: Customers are increasingly turning to <unk> to create a modernized private cloud on Prem.

Hawk: Despite these certain economic environment.

Kirsten Spears: This figure excludes $142 million of depreciation.

Hawk: In fact.

Hawk: What we've seen recently.

Speaker Change: Which will enable men.

Hawk: Is that they are doubling down.

Kirsten Spears: Now a review of the P&L for Archie's segment. Starting with semiconductors. Revenue for our Semiconductor Solutions segment was $8.4 billion, with growth accelerating to 17% year-on-year driven by AI. Semiconductor revenue represented 56% of total revenue in the quarter. Gross Margin for our Semiconductor Solutions Segment was approximately 69%, up 140 basis points year-on-year, driven by product quality. Operating expenses increased 12% year-on-year to $971 million on increased investment in R&D for leading-edge AI semiconductors. Semiconductor Operating Margin of 57% was up 200 basis points year on year.

Speaker Change: To repatriate workloads from public clouds.

Hock Tan: on Inference in order to monetize their platform. and reflecting this, we may actually see an acceleration of XPU demand. into the back half of 2026 to meet urgent demand for inference on top of the demand we have indicated from training. And accordingly, we do anticipate now, our fiscal 2025 growth rate of AI Semiconductor Revenue to sustain. into fiscal 2026.

Hawk: On inference in order to monetize their platforms.

Speaker Change: <unk> been able to run modern container based applications.

Hawk: And reflecting this we may actually see it.

Speaker Change: AI applications.

Speaker Change: Of our 10000 largest customers over 87% have now adopted this year.

Hawk: Deceleration of export demand.

Hawk: Into the back half of 2026 to meet urgent demand for inference on top of the demand we have indicated.

Speaker Change: The momentum from strong DCF sales over the past 18 months since the acquisition of Vmware.

Speaker Change: Hence, creating annual recurring revenue or otherwise.

Hawk: From training.

Hawk: And accordingly, we do anticipate now our fiscal 2025 growth rate.

Speaker Change: Otherwise known as our growth of double digits in our core infrastructure software.

Hawk: Semiconductor revenue to sustain.

Speaker Change: In Q3.

Hawk: Into fiscal 2026.

Speaker Change: We expect infrastructure software revenue to be approximately $6 7 billion.

Hock Tan: Turning to our Q3 out... as we continue our current trajectory of growth. We focus AI semiconductor revenue. to be $5.1 billion, up 60% year-on-year, which would be the 10th consecutive quarter of growth. Now turning to non-AI semiconductors in Q2. Revenue of $4 billion was down 5% year-on-year. Non-AI Semiconductor Revenue is close to the bottom. has been relatively slow to recover. but they are bright spots in Q2 broadband. Enterprise Networking and Server Storage Revenues were up sequentially.

Hawk: Turning to Q2, our Q3 outlook.

Hawk: We continue our current trajectory of growth.

Speaker Change: Up 16% year on year.

Speaker Change: So in total we're guiding Q3 consolidated revenue to be approximately $15 8 billion up 21% year on year.

Hawk: We forecast a semiconductor revenue.

Kirsten Spears: Now moving on to infrastructure software. Revenue for infrastructure software of $6.6 billion was up 25% year-on-year and represented 44% of total revenue. gross margin for infrastructure software was 93% in the quarter compared to 88% a year ago. Operating expenses were $1.1 billion in the quarter, resulting in infrastructure software operating margin of approximately 76%. This compares to operating margin of 60% a year ago. This year on year improvement reflects our disciplined integration of VMware.

Hawk: To be $5 $1 billion.

Hawk: Up 60% year on year.

Hawk: Which would be the 10th consecutive quarter of growth.

Speaker Change: We expect Q3, adjusted EBITDA to be at least 66%.

Hawk: Yes.

Hawk: Now turning to non AI semiconductors in Q2 revenue of $4 billion was down 5% year on year.

Speaker Change: With that.

Kirsten: Let me turn the call over to Kirsten.

Kirsten: Thank you Hock, let me.

Kirsten: We now provide additional detail on our Q2 financial performance.

Hawk: Non semiconductor revenue is close to the bottom.

Hawk: And it's been relatively slow to recover.

Kirsten: Consolidated revenue was a record 15 billion for the quarter up 20% from a year ago.

Hawk: But there are bright spots in Q2 broadband.

Speaker Change: Gross margin was 79, 4% of revenue in the quarter better than we originally guided on product mix.

Hawk: Enterprise networking.

Hawk: Server storage revenues were up sequentially.

Kirsten Spears: Moving on to cash flow. Free cash flow in the quarter was $6.4 billion and represented 43% of revenue. Free Cash Flow as a Percentage of Revenue continues to be impacted by increased interest expense from debt related to the VMware acquisition and increased cash tax. We spent $144 million on capital expenses. Day sales outstanding were 34 days in the second quarter compared to 40 days a year ago. We ended the second quarter with inventory of $2 billion up 6% sequentially in anticipation of revenue growth in future quarters. Our days of inventory on hand were 69 days in Q2, as we continue to remain disciplined on how we manage inventory across the ecosystem.

Hock Tan: However... Industrial was down and as expected, wireless was also down due to seasonality. In Q3, we expect enterprise networking and broadband. continue to grow sequentially. But server storage, wireless, and industrial are expected to be largely flat. And overall, we forecast non-AI semiconductor revenue to stay around $4 billion.

Hawk: However.

Speaker Change: <unk> operating expenses were $2 1 billion of which one 5 billion was related to R&D queue.

Hawk: Industrial was down and this is a N S expected wireless was also down due to seasonality.

Speaker Change: Q2 operating income of $9 8 billion was up 37% from a year ago with operating margin at 65% of revenue.

Hawk: In Q3.

Hawk: We expect enterprise networking and broadband.

Hawk: To continue to grow sequentially.

Hawk: But server storage wireless and industrial.

Speaker Change: Adjusted EBITDA was $10 billion or 67% of revenue above our guidance of 66%. This figure excludes 142 million of depreciation.

Hawk: <unk> to be largely flat.

Hawk: Overall, we forecast non semiconductor revenue to stay around $4 billion.

Speaker Change: Now a review of the P&L for our two segments.

Hawk: Yes.

Hock Tan: Now let me talk about our infrastructure software. Q2 infrastructure software revenue of $6.6 billion was up 25% year-on-year, above our outlook of $6.5 billion. As we have said before, This growth reflects our success in converting our enterprise customers from perpetual vSphere to the full VCF software state subscription. Customers are increasingly turning to VCF to create a modernized private cloud on-prem which will enable them to repatriate workloads from public clubs. while being able to run modern container-based applications. and AI Applications. Of our 10,000 largest customers, over 87% have now adopted VCF. momentum from strong VCF sales over the past 18 months.

Hawk: Now, let me talk about our infrastructure software segment.

Speaker Change: Starting with semiconductors.

Hawk: Q2 infrastructure software revenue of $6 6 billion was up 25% year on year above our outlook of $6 5 billion.

Speaker Change: Revenue for our semiconductor solutions segment was $8 4 billion with growth accelerating to 17% year on year driven by AI.

Kirsten Spears: We ended the second quarter with $9.5 billion of cash and $69.4 billion of gross principal debt. Subsequent to quarter end, we repaid $1.6 billion of debt, resulting in gross principal debt of $67.8 billion. The weighted average coupon rate and years to maturity of our $59.8 billion in fixed-rate debt is 3.8% in seven years, respectively. The weighted average interest rate and years to maturity of our $8 billion in floating rate debt is 5.3% and 2.6 years respectively.

Speaker Change: Semiconductor revenue represented 56% of total revenue in the quarter.

Hawk: As we have said before this.

Hawk: This growth reflects our success in converting our enterprise customers from perpetual vis fear.

Speaker Change: Gross margin for our semiconductor solutions segment with approximately 69% up 140 basis points year on year driven by product mix.

Hawk: The full vcs software stack subscription.

Hawk: Customers are increasingly turning to vcs to create a modernized private cloud on Prem.

Speaker Change: Operating expenses increased 12% year on year to 971 million on increased investment in R&D for leading edge AI semiconductors.

Hawk: Which will enable Ned.

Hawk: To repatriate workloads from public clubs.

Speaker Change: Semiconductor operating margin of 57% was up 200 basis points year on year.

Hawk: <unk> been able to run modern container based applications.

Kirsten Spears: Turning to capital allocations. Q2 we paid stockholders $2.8 billion of cash dividends based on a quarterly common stock cash dividend of $0.59 per share. In Q2, we repurchased 4.2 billion, or approximately 25 million shares of common stock. In Q3, we expect the non-GAAP diluted share count to be approximately 4.97 billion shares, excluding the potential impact of any share repurchase.

Hawk: AI applications.

Speaker Change: Now moving on to infrastructure software.

Hawk: Of our 10000 largest customers over 87% have now adopted vcs.

Speaker Change: Revenue for infrastructure software of $6 6 billion was up 25% year on year and represented 44% of total revenue.

Hawk: The momentum from strong DCF sales over the past 18 months since the acquisition of Vmware.

Hock Tan: Since the acquisition of VMware, has created annual recurring revenue, or otherwise known as ARR, growth of double digits in our core infrastructure software. in Q3. We expect infrastructure software revenue to be approximately $6.7 billion, up 16% year-on-year. So in total, we're getting Q3 consolidated revenue to be approximately $15.8 billion, up 21% year-on-year. We expect Q3 adjusted EBITDA to be at least 66%.

Speaker Change: Gross margin for infrastructure software with 93% in the quarter compared to 88% a year ago.

Hawk: Has created annual recurring revenue or otherwise.

Hawk: Otherwise known as our growth of double digits cannot call infrastructure software.

Speaker Change: Operating expenses were $1 1 billion in the quarter, resulting in infrastructure software operating margin of approximately 76%.

Kirsten Spears: Now moving on to Guide. Our guidance for Q3 is for consolidated revenue of $15.8 billion, up 21% year-on-year. The forecast semiconductor revenue of approximately $9.1 billion, up 25% year-on-year. Within this, we expect Q3 AI semiconductor revenue of $5.1 billion, up 60% year-on-year. We expect infrastructure software revenue of approximately $6.7 billion, up 16% year-on-year. For modeling purposes, we expect Q3 consolidated gross margin to be down approximately 130 basis points sequentially, primarily reflecting a higher mix of XPUs within AI revenue. As a reminder, consolidated gross margins through the year will be impacted by the revenue mix of infrastructure, software, and semiconductors.

Hawk: In Q3.

Speaker Change: This compares to operating margin of 60% a year ago. This year on year improvement reflects our disciplined integration of Vmware.

Hawk: We expect infrastructure software revenue to be approximately $6 7 billion.

Hawk: Up 16% year on year.

Speaker Change: Moving onto cash flow.

Speaker Change: Free cash flow in the quarter was $6 4 billion and represented 43% of revenue.

Hawk: So in total we're guiding Q3 consolidated revenue to be approximately 15 $8 billion up 21% year on year.

Speaker Change: Free cash flow as a percentage of revenue continues to be impacted by increased interest expense from debt related to the Vmware acquisition and increased cash taxes.

Hawk: We expect Q3, adjusted EBITDA to be at least 66%.

Speaker Change: We spent 144 million on capital expenditures.

Hock Tan: With that, let me turn the call over to Kirsten.

Hawk: With that.

Kirsten: Let me turn the call over to Kirsten.

Speaker Change: Days sales outstanding were 34 days in the second quarter compared to 40 days a year ago.

Kirsten Spears: Thank you, Hock.

Kirsten: Thank you Hock, let me now provide additional detail on our Q2 financial performance.

Kirsten Spears: Let me now provide additional detail on our Q2 financial performance. Consolidated revenue was a record $15 billion for the quarter, up 20% from a year ago. Gross margins was 79.4% of revenue in the quarter, better than we originally guided on product Consolidated operating expenses were $2.1 billion, of which $1.5 billion was related to R&D. Q2 operating income of $9.8 billion was up 37% from a year ago, with operating margin at 65% of revenue. Adjusted EBITDA was $10 billion or 67% of revenue, above our guidance of 66%.

Speaker Change: We ended the second quarter with inventory of $2 billion up 6% sequentially and anticipation of revenue growth in future quarters.

Kirsten: Consolidated revenue was a record 15 billion for the quarter up 20% from a year ago.

Speaker Change: Our days of inventory on hand were 69 days in Q2 as we continue to remain disciplined on how we manage inventory across the ecosystem.

Kirsten Spears: We expect Q3 adjusted EBITDA to be at least 66%. We expect the non-GAAP tax rate for Q3 in fiscal year 2025 to remain at 14%.

Kirsten: Gross margin was 79, 4% of revenue in the quarter better than we originally guided on product mix.

Speaker Change: We ended the second quarter with $9 5 billion of cash and $69 4 billion of gross principal debt.

Kirsten: Consolidated operating expenses were $2 1 billion of which one 5 billion was related to R&D.

Kirsten Spears: And with this, that concludes my prepared remarks.

Ji Yoo: Operator, please open up the call for questions. Thank you. To ask a question, you will need to press star 11 on your telephone. To withdraw your question, please press star 11 again. Due to time restraints, we ask that you please limit yourself to one question. Please stand by while we compile the Q&A roster.

Speaker Change: Subsequent to quarter end, we repaid $1 6 billion of debt, resulting in gross principal debt of $67 8 billion.

Kirsten: Q2 operating income of $9 8 billion was up 37% from a year ago with operating margin at 65% of revenue.

Speaker Change: The weighted average coupon rate in years to maturity of our $59 8 billion in fixed rate debt is three 8% and seven years respectively.

Kirsten: Adjusted EBITDA was $10 billion or 67% of revenue above our guidance of 66%. This figure excludes 142 million of depreciation.

Kirsten Spears: This figure excludes $142 million of depreciation.

Speaker Change: The weighted average interest rate in years to maturity of our 8 billion in floating rate debt is five 3% and two six years respectively.

Ross Seymour: And our first question will come from the line of Ross Seymour with Deutsche Bank. Your line is open. Hi, guys. Thanks for letting me ask a question. Hock, I wanted to jump onto the AI side and specifically some of the commentary you had about next year. Can you just give a little bit more color on the inference commentary you gave? And is it more the XPU side, the connectivity side, or both that's giving you the confidence to talk about the growth rate that you have this year being matched next fiscal year? Thank you, Ross. Good question.

Kirsten Spears: Now a review of the P&L for R2-SEG. Starting with semiconductors. Revenue for our Semiconductor Solutions segment was $8.4 billion, with growth accelerating to 17% year-on-year driven by AI. Semiconductor revenue represented 56% of total revenue in the quarter. Gross Margin for our Semiconductor Solution Segment was approximately 69%, up 140 basis points year-on-year, driven by product. Operating expenses increased 12% year-on-year to $971 million on increased investment in R&D for leading-edge AI semiconductors. Semiconductor Operating Margin of 57% was up 200 basis points year on year.

Kirsten: Now a review of the P&L for our two segments.

Kirsten: Starting with semiconductors.

Speaker Change: Turning to capital allocation.

Kirsten: Revenue for our semiconductor solutions segment was $8 4 billion with growth accelerating to 17% year on year driven by AI.

Speaker Change: In Q2, we paid stockholders $2 8 billion of cash dividends based on our quarterly common stock cash dividend of 59 cents per share.

Kirsten: Semiconductor revenue represented 56% of total revenue in the quarter.

Speaker Change: In Q2, we repurchased $4 2 billion or approximately 25 million shares of common stock.

Kirsten: Gross margin for our semiconductor solutions segment with approximately 69% up 140 basis points year on year driven by product mix.

Hock Tan: I think we're indicating that what we are seeing and what we have quite a bit of visibility increasingly is increased Deployment of XPUs next year. and much more than we originally thought and hand-in-hand with it, of course, more and more networking. So it's a combination of both. and the imprint side of things. And yeah, we're seeing much more in France now. Thank you. One moment for our next question.

Speaker Change: In Q3, we expect non-GAAP diluted share count to be approximately $4 97 billion shares excluding the potential impact of any share repurchases.

Kirsten: Operating expenses increased 12% year on year to $971 million on increased investment in R&D for leading edge AI semiconductors.

Speaker Change: Now moving onto guidance.

Speaker Change: Our guidance for Q3 is for consolidated revenue of $15 8 billion up 21% year on year, we forecast semiconductor revenue of approximately $9 1 billion up 25% year on year.

Kirsten: Semiconductor operating margin of 57% was up 200 basis points year on year.

Speaker Change: Within this we expect Q3 AI semiconductor revenue of $5 1 billion up 60% year on year.

Kirsten Spears: Now moving on to infrastructure software. Revenue for infrastructure software of $6.6 billion was up 25% year-on-year and represented 44% of total revenue. Gross margin for infrastructure software was 93% in the quarter compared to 88% a year ago. Operating expenses were $1.1 billion in the quarter, resulting in infrastructure software operating margin of approximately 76%. This compares to operating margin of 60% a year ago.

Kirsten: Now moving on to infrastructure software.

Kirsten: Revenue for infrastructure software of $6 6 billion was up 25% year on year and represented 44% of total revenue.

Speaker Change: We expect infrastructure software revenue of approximately $6 7 billion up 16% year on year.

Harlan Sur: And that will come from the line of Harlan Sur with J.P. Morgan. Your line is open. All right, good afternoon. Thanks for taking my question and great job on the quarterly execution. Hock, you know, good to see the positive growth in suction, quarter quarter, year over year growth rates in your AI business. You have the team, as mentioned, right, the quarters can be a bit lumpy. So if I smooth out kind of first three quarters of this fiscal year, your AI business is up 60% year over year. It's kind of right in line with your three year kind of Sam growth cager, right?

Kirsten: Gross margin for infrastructure software with 93% in the quarter compared to 88% a year ago.

Speaker Change: For modeling purposes, we expect Q3 consolidated gross margin to be down approximately 130 basis points sequentially, primarily reflecting a higher mix of X P used within AI revenue.

Kirsten: Operating expenses were $1 1 billion in the quarter, resulting in infrastructure software operating margin of approximately 76%.

Speaker Change: As a reminder, our consolidated gross margins through the year will be impacted by the revenue mix of infrastructure software and semiconductors.

Kirsten: This compares to operating margin of 60% a year ago. This year on year improvement reflects our disciplined integration of Vmware.

Kirsten Spears: This year-on-year improvement reflects our disciplined integration of VM1.

Speaker Change: We expect Q3, adjusted EBITDA to be at least 66%.

Kirsten Spears: Moving on to cash flow. Free cash flow in the quarter was $6.4 billion and represented 43% of revenue. Free Cash Flow as a Percentage of Revenue continues to be impacted by increased interest expense from debt related to the VMware acquisition and increased cash tax.

Kirsten: Moving onto cash flow.

Speaker Change: We expect the non-GAAP tax rate for Q3 and fiscal year 2025 to remain at 14%.

Kirsten: Free cash flow in the quarter was $6 4 billion and represented 43% of revenue.

Hock Tan: Given your prepared remarks, and, you know, knowing that your lead times remain at 35 weeks or better, you see the Broadcom team sustaining the 60% year over year growth rate exiting this year. And I assume that that potentially implies that you see your AI business sustaining the 60% year over year growth rate into fiscal 26, again, based on your prepared commentary, which again is in line with your Sam growth cager. Is that kind of a fair way to think about the trajectory this year and next year?

Speaker Change: And with this that concludes my prepared remarks, operator, please open up the call for questions.

Kirsten: Free cash flow as a percentage of revenue continues to be impacted by increased interest expense from debt related to the Vmware acquisition and increased cash taxes.

Speaker Change: Thank you to ask a question you will need to press star one on your telephone to withdraw your question. Please press star one again due to time restraints. We ask that you. Please limit yourself to one question. Please standby, while we compile the Q&A roster.

Kirsten Spears: We spent $144 million on capital. Day sales outstanding were 34 days in the second quarter compared to 40 days a year ago. We ended the second quarter with inventory of $2 billion, up 6% sequentially in anticipation of revenue growth in future quarters. Our days of inventory on hand were 69 days in Q2, as we continue to remain disciplined on how we manage inventory across We ended the second quarter with $9.5 billion of cash and $69.4 billion of gross principal debt. Subsequent to quarter end, we repaid $1.6 billion of debt, resulting in gross principal debt of $67.8 billion.

Kirsten: We spent 144 million on capital expenditures.

Kirsten: Days sales outstanding were 34 days in the second quarter compared to 40 days a year ago.

Kirsten: We ended the second quarter with inventory of $2 billion up 6% sequentially and anticipation of revenue growth in future quarters.

Hock Tan: Harlan, that's a very insightful set of analysis here. And that's exactly what we're trying to do here. Because six over six months ago, we gave you guys a point a year 2027. As we come into the second now into the second half of 2025. And we've improved visibility and updates we're seeing in the way our hyperscale partners are deploying data centers, AI class We are providing you some level of guidance, visibility, what we are seeing, how the trajectory of 26 might look like. I'm not giving you any update on 27. We're just still establishing the update we have in 27, six months ago.

Speaker Change: And our first question will come from the line of Ross Seymore with Deutsche Bank. Your line is open.

Ross Seymore: Guys. Thanks for let me ask a question Hock I wanted to jump into the AI side and specifically some of the commentary you had about next year can you just give a little bit more color on the inference commentary you gave and then is it more of the SBU side, the connectivity side or both that's giving you the confidence to talk about the growth rate that you have this year being matched next fiscal year.

Kirsten: Our days of inventory on hand were 69 days in Q2 as we continue to remain disciplined on how we manage inventory across the ecosystem.

Kirsten: We ended the second quarter with $9 5 billion of cash and $69 4 billion of gross principal debt.

Kirsten: Subsequent to quarter end, we repaid $1 6 billion of debt, resulting in gross principal debt of $67 8 billion.

Speaker Change: Thank you Ross good question.

Speaker Change: I think we're indicating that what we are seeing and what we have quite a bit of visibility increasingly.

Kirsten Spears: The weighted average coupon rate and years to maturity of our $59.8 billion in fixed rate debt is 3.8% in seven years, respectively. The weighted average interest rate and years to maturity of our $8 billion in floating rate debt is 5.3% and 2.6 years respectively.

Kirsten: The weighted average coupon rate in years to maturity of our $59 8 billion in fixed rate debt is three 8% and seven years respectively.

Speaker Change: This increase deploy.

Speaker Change: Deployment of XP use next year.

Speaker Change: And much more than we originally thought and hand in hand with it of course.

Kirsten: The weighted average interest rate in years to maturity of our 8 billion in floating rate debt is five 3% and 2.6 years respectively.

Speaker Change: More and more networking so it's a combination of both.

Speaker Change: And then in print side of things.

Hock Tan: But what we're doing now is giving you more visibility into where we're seeing 26 headed. But is the framework that you laid out for us, like second half of last year, which implies 60% kind of growth kager in your SAM opportunity, is that kind of the right way to think about it as it relates to the profile of growth in your business this year and next year? Yes. Okay, thank you all. Thank you. One moment for our next question.

Kirsten Spears: Turning to capital allocations. At Q2, we paid stockholders $2.8 billion of cash dividends based on a quarterly common stock cash dividend of $0.59 per share. In Q2, we repurchased $4.2 billion, or approximately 25 million shares of common stock. In Q3, we expect the non-GAAP diluted share count to be approximately 4.97 billion shares excluding the potential impact of any share repurchase.

Kirsten: Turning to capital allocation.

Speaker Change: Yeah, we're seeing much morning for himself.

Kirsten: Q2, we paid stockholders $2 8 billion of cash dividends based on our quarterly common stock cash dividend of 59 cents per share.

Speaker Change: Thank you.

Speaker Change: Thank you one moment for our next question.

Kirsten: In Q2, we repurchased $4 2 billion or approximately 25 million shares of common stock in Q3, we expect non-GAAP diluted share count to be approximately $4 97 billion shares excluding the potential impact of any share repurchases.

Speaker Change: And that will come from the line of Harlan sur with Jpmorgan. Your line is open.

Harlan Sur: Hey, good afternoon. Thanks for taking my question and great job on the quarterly execution Hock good to see the positive growth inflection quarter over quarter year over year growth rates in your AI business do you have the team.

Kirsten Spears: Now moving on to Guide... Our guidance for Q3 is for consolidated revenue of $15.8 billion, up 21% year-on-year. The forecast semiconductor revenue of approximately $9.1 billion of 25%. Within this, we expect Q3 AI semiconductor revenue of 5.1 billion of 60% year on We expect infrastructure software revenue of approximately $6.7 billion, up 16% year-on-year. For modeling purposes, we expect Q3 consolidated gross margin to be down approximately 130 basis points sequentially, primarily reflecting a higher mix of XPUs within AI revenue. As a reminder, consolidated gross margins through the year will be impacted by the revenue mix of infrastructure, software, and semiconductors.

Kirsten: Now moving onto guidance.

Speaker Change: Mentioned right the quarters can be a bit lumpy so if I.

Ben Reitzes: And that will come from the line of Ben Reitzes with Mellius Research. Your line is open. Hey, how you doing? Thanks, guys. Hey, Hock, networking, AI networking was really strong in the quarter. And it's, it seemed like it must have beat expectations. I was wondering if you could just talk about the networking in particular, what caused that? And how much of that is your acceleration into next year? And when do you think you see Tomahawk kicking in as part of that acceleration? Thanks. Well, I think the network AI networking, as you probably would know, goes pretty hand in hand with deployment of AI accelerator clusters.

Kirsten: Our guidance for Q3 is for consolidated revenue of $15 8 billion up 21% year on year, we forecast semiconductor revenue of approximately $9 1 billion up 25% year on year.

Speaker Change: Mood, all kind of first three quarters of this fiscal year.

Speaker Change: Our business is up 60% year here, it's kind of right in line with your three year kind of Sam growth CAGR right given your prepared remarks.

Kirsten: Within this we expect Q3 AI semiconductor revenue of $5 1 billion up 60% year on year.

Speaker Change: Knowing that your lead times remain at 35 weeks of better do you see the broadcom team sustaining the 60%.

Kirsten: We expect infrastructure software revenue of approximately $6 7 billion up 16% year on year.

Speaker Change: <unk> growth rate exiting this year and I assume that potentially.

Speaker Change: Potentially implies that you see your AI business sustaining 60% year over year growth rate into fiscal 'twenty six again based on your prepared commentary, which again is in line with your Sam go speak or is that kind of a fair way to think about the trajectory this year and next year.

Kirsten: For modeling purposes, we expect Q3 consolidated gross margin to be down approximately 130 basis points sequentially, primarily reflecting a higher mix of X P used within AI revenue.

Hock Tan: Yeah, Harlan that's a very insightful set of analysis here and that's exactly what we're trying to do here because six over six months ago. We gave you guys.

Kirsten: As a reminder, our consolidated gross margins through the year will be impacted by the revenue mix of infrastructure software and semiconductors.

Hock Tan: It doesn't deploy on a timetable that's very different from the way that accelerators get deployed, whether they are XPUs or GPUs. It does happen, and they deploy a lot in scale out. Wei, Ethernet, of course. Choice or Protocol, but it's also increasingly moving into the space of what we all call scale up, within those data centers, where you have much higher, more than we originally thought, consumption or density of switches than you have in the scale out scenario. It's, in fact, increased density in scale up is. 5 to 10 times more than in scale-out. And that's the part that kind of pleasantly surprises, and which is why in this past quarter Q2, the AI networking portion continues at about 40% from when we reported a quarter ago for Q1.

Kirsten Spears: We expect Q3 adjusted EBITDA to be at least 66%. We expect the non-GAAP tax rate for Q3 in fiscal year 2025 to remain at 14%.

Kirsten: We expect Q3, adjusted EBITDA to be at least 66%.

Speaker Change: A point of year 2027.

Kirsten: We expect the non-GAAP tax rate for Q3 and fiscal year 2025 to remain at 14%.

Speaker Change: As we've come into the second and now into the second half of 2025.

Kirsten Spears: And with this, that concludes my prepared remarks.

Kirsten: And with this that concludes my prepared remarks, operator, please open up the call for questions.

Operator: Operator, please open up the call for questions. Thank you. To ask a question, you will need to press star 11 on your telephone. To withdraw your question, please press star 11 again. Due to time restraints, we ask that you please limit yourself to one question. Please stand by while we compile the Q&A roster.

Speaker Change: And we've improved visibility and updates we are seeing in the way.

Kirsten: Thank you to ask a question you will need to press star one on your telephone to withdraw your question. Please press star one again due to time restraints. We ask that you. Please limit yourself to one question. Please standby, while we compile the Q&A roster.

Speaker Change: Our Hyperscale partners are deploying data centers AI clusters.

Speaker Change: We are providing you more some level of.

Speaker Change: Our guidance visibility, what we are seeing how the trajectory of 26 might look like.

Ross Seymour: And our first question will come from the line of Ross Seymour with Deutsche Bank. Your line is open. Hi, guys. Thanks for letting me ask a question. Hock, I wanted to jump on to the AI side and specifically some of the commentary you had about next year. Can you just give a little bit more color on the inference commentary you gave?

Speaker Change: And our first question will come from the line of Ross Seymore with Deutsche Bank. Your line is open.

Speaker Change: I'm not giving you any update on the 27th we just still establishing the update we have been 27 six months ago, but what we're doing now is giving you more visibility.

Ross Seymore: I think certainly ask a question hock I wanted to jump into the AI side and specifically some of the commentary you had about next year can you just give a little bit more color on the inference commentary you gave and then is it more of the SBU side, the connectivity side or both that's giving you the confidence to talk about the growth rate that you have this year being matched next fiscal year.

Speaker Change: Into where we are seeing 26 head.

Hock Tan: And is it more the XPU side, the connectivity side, or both that's given you the confidence to talk about the growth rate that you have this year being matched next fiscal Thank you, Ross. Good question. I think we're indicating that what we are seeing and what we have quite a bit of visibility increasingly is increased Deployment of XPUs next year. and much more than we originally thought. And hand-in-hand with it, of course, more and more networking. So it's a combination of both. and the imprint side of things. Yeah, we're seeing much more inference now. Thank you.

Speaker Change: But as the framework that you laid out for us like second half of last year, which implies 60% kind of growth CAGR in the Sam opportunity is that kind of the right way to think about it as it relates to the profile of growth in your business. This year and next year, yes.

Ross Seymore: Thank you Ross good question.

Hock Tan: And at that time I said, I expect it to drop. It happens.

Ross Seymore: I think we're indicating that what we are seeing and what we have quite a bit of visibility increasingly.

Harlan Sur: Okay. Thank you hock.

Hock Tan: and and your thoughts on Tomahawk driving acceleration for next year and when it kicks in. Oh, Tomahawk 6. Oh, yeah, that's extremely strong interest. Now, we're not shipping big orders or any orders other than basic proof of concepts out to customers. But there is tremendous demand for this new 102 terabit per second Tomahawk Thanks, Hock. Thank you.

Ross Seymore: This increase deploy.

Speaker Change: Thank you one moment for our next question.

Ross Seymore: Deployment of XP use next year.

Ross Seymore: And much more than we originally thought and hand in hand with it of course.

Speaker Change: And that will come from the line of Ben Reitzes with Melius Research. Your line is open.

Ross Seymore: More and more networking so it's a combination of both.

Ben Reitzes: Hey, How're you doing thanks, guys, Hey, Hock.

Hock Tan: Networking AI networking was really strong in the quarter.

Ross Seymore: And the print side of things.

Ben Reitzes: And then it seemed like it must have beat expectations. I was wondering if you could just talk about the networking in particular, what what caused that and how much of that is your acceleration into next year.

Ross Seymore: Yeah, we're seeing much more willing for himself.

Ross Seymore: Thank you.

Speaker Change: Thank you one moment for our next question.

Operator: One moment for our next question.

Harlan Sur: And that will come from the line of Harlan Sur with J.P. Morgan.

Speaker Change: And that will come from the line of Harlan sur with Jpmorgan. Your line is open.

Unknown Executive: One moment for our next question.

Ben Reitzes: And when do you think you see tomahawk kicking in as part of that acceleration. Thanks.

Harlan Sur: Your line is open. All right, good afternoon. Thanks for taking my question and great job on the quarterly execution. Hock, you know, good to see the positive growth in suction quarter quarter, year over year growth rates in your AI business. You have the team, as mentioned, right, the quarters can be a bit lumpy. So if I smooth out kind of first three quarters of this fiscal year, your AI business is up 60% year over year. It's kind of right in line with your three year kind of Sam growth Kager, right? Given your prepared remarks, and, you know, knowing that your lead times remain at 35 weeks or better, you see the Broadcom team sustaining the 60% year over year growth rate exiting this year.

Blayne Curtis: And that will come from the line of Blayne Curtis with Jeffries. Your line is open. Hey, thanks. Great results. I just want to ask, maybe following up on the scale out opportunity. So today, I guess your main customer is not really using an MVLINK switch style scale up.

Harlan Sur: Hey, good afternoon. Thanks for taking my question and great job on the quarterly execution Hock good to see the positive growth inflection quarter over quarter.

Ben Reitzes: Well.

Ben Reitzes: I think the.

Ben Reitzes: Network AI networking.

Ben Reitzes: You probably would know goes hand in hand with deployment of AI accelerate the clusters. It is it doesn't deploy a on the timetable that is very different from the way the.

Speaker Change: Growth rates in your AI business you have the team.

Speaker Change: Mentioned right the quarters can be a bit lumpy. So if I may.

Hock Tan: I'm just kind of curious, your visibility or the timing in terms of when you might be shipping, you know, a switch Ethernet scale up network to your customer. The Punking Scale-Up Scale, scale up. Yeah, well, scale up. very rapidly converting to Ethernet. Very much so. I am for our fairly narrow band of hyperscale customers. is very much internet. Thank you. One moment for our next question.

Speaker Change: <unk> kind of first three quarters of this fiscal year.

Speaker Change: Our business is up 60% year here, it's kind of right in line with your three year kind of Sam growth CAGR right given your prepared remarks.

Ben Reitzes: Salaries does get deployed.

Ben Reitzes: Uh huh.

Ben Reitzes: Excuse or Gpus it does happen.

Speaker Change: Knowing that your lead times remain at 35 weeks of better do you see the broadcom team sustaining the 60% year over year growth rate exiting this year and I assume that potentially.

Ben Reitzes: Deploy a lot in scale out.

Ben Reitzes: Ethernet of course is the choice of protocol, but it's also increasingly moving into the space of what we all call scale up within those data centers, where you have more.

Harlan Sur: And I assume that that potentially implies that you see your AI business sustaining the 60% year of your growth rate into fiscal 26 again, based on your prepared commentary, which again is in line with your Sam growth maker. Is that kind of a fair way to think about the trajectory this year and next year?

Speaker Change: Potentially implies that you see your AI business sustaining 60% year over year growth rate into fiscal 'twenty six again based on your prepared commentary, which again is in line with your Sam growth pick or is that kind of a fair way to think about the trajectory this year and next year.

Ben Reitzes: Much higher more than we originally thought.

Ben Reitzes: Sumption or density of switches than you have in the scale out scenario.

Hock Tan: Harlan, that's a very insightful set of analysis here. And that's exactly what we're trying to do here. Because six over six months ago, we gave you guys a point a year 2027. As we come into the second now into the second half of 2025. And we've improved visibility and updates we're seeing in the way our hyperscale partners are deploying data centers AI class We are providing you some level of guidance, visibility, what we are seeing, how the trajectory of 26 might look. I'm not giving you any update on 27. We're just still establishing the update we have in 27, six months ago.

Speaker Change: Collyn, that's a very insightful set of analysis here and that's exactly what we're trying to do here because six over six months ago. We gave you guys.

Stacy Rasgon: And that will come from the line of Stacy Rasgon with Bernstein. Your line is open. Hi, guys. Thanks for taking my questions. Hock, I still wanted to follow up on that AI 2026 question. I want to just put some numbers on it, just just to make sure I've got it right. So if you you did 60% in the first three quarters of this year, if you grow 60% year over year in Q4 and put you at like, I don't know, 5.8 billion, something like 19 or 20 billion for the year. And then are you saying you're going to grow 60% in 2026, which would put you $30 billion plus in AI revenues for 2026?

Ben Reitzes: It's in fact increased density and scale up is.

Ben Reitzes: Five to 10 times more than in scale out and that's the part that kind of pleasantly surprised us and which is why in this past quarter Q2.

Speaker Change: A point of year 2027.

Speaker Change: As we've come into the second and now into the second half of 2025, and we've improved visibility and updates we are seeing in the way.

Ben Reitzes: The AI networking portion continues at about 40% from when we reported a quarter ago for Q1.

Speaker Change: Hyperscale partners are deploying data centers AI clusters.

Ben Reitzes: And at that time, I said I expect it to drop.

Ben Reitzes: It hasn't.

Speaker Change: We are providing you more some level of.

Hock Tan: I just want to know, is that the math that you're trying to communicate to us directly? I think you're doing the math. I'm giving you the trend. But I did answer that question, I think Harlan asked earlier. The rate we are seeing now so far in Fiscal 25 And we'll presumably continue. We don't see any reason why it doesn't give a lead time visibility in 25. What we're seeing today, based on what we have visibility on 26th, is to be able to Ram up this AI revenue in the same trajectory. Yes, so is the SAM going up?

Ben Reitzes: And your thoughts on Tomahawk driving acceleration for next year and when it kicks in.

Speaker Change: Our guidance visibility, what we are seeing how the trajectory of 26 might look like.

Ben Reitzes: Oh, Tom out six Oh, yeah, that's extremely strong interest.

Ben Reitzes: No we're not shipping.

Speaker Change: I'm not giving you any update on 2007, we just still establishing the update we have been 27 six months ago, but what we're doing now is giving you more visibility.

Ben Reitzes: Big orders or any orders on other than the basic.

Hock Tan: But what we're doing now is giving you more visibility into where we're seeing 26 headed. But is the framework that you laid out for us like second half of last year, which implies 60% kind of growth, Kager and your SAM opportunity, is that kind of the right way to think about it as it relates to the profile of growth in your business this year and next year? Yes.

Speaker Change: Both concepts out to customers, but there is tremendous demand for this new 102 terabits.

Speaker Change: Into where we are seeing 26 head.

Ben Reitzes: Terabits per second.

Speaker Change: But as the framework that you laid out for us like second half of last year, which implies 60% kind of growth CAGR on the Sam opportunity is that kind of the right way to think about it as it relates to the profile of growth in your business. This year and next year, yes.

Ben Reitzes: Tomahawk switches.

Ben Reitzes: Thanks Hock.

Ben Reitzes: Thank you one moment our next question.

Speaker Change: And that will come from the line of Blayne Curtis with Jefferies. Your line is open.

Harlan Sur: Okay, thank you, Hock. Thank you.

Speaker Change: Hey, Thanks, and the great results I just wanted to ask maybe following up on the scale out opportunities to today I guess your main customers not really using them in NV links with style scale up I'm just kind of curious.

Speaker Change: Okay. Thank you hock.

Hock Tan: So is the SAM going up as well? Because now you have inference on top of training. So is the SAM still 60 to 90? Or is the SAM higher now as you see it?

Speaker Change: Thank you one moment for our next question.

Ben Reitzes: One moment for our next question. And that will come from the line of Ben Reitzes with Mellius Research. Your line is open. Hey, how you doing? Thanks, guys. Hey, Hock, networking, AI networking was really strong in the quarter. And it's, it seemed like it must have beat expectations. I was wondering if you could just talk about the networking in particular, what what caused that? And how much of that is your acceleration into next year? And when do you think you see Tomahawk kicking in as part of that acceleration? Thanks. Well, I think the network AI networking, as you probably would know, goes pretty hand in hand with deployment of AI accelerator clusters.

Speaker Change: And that will come from the line of Ben Reitzes with Melius Research. Your line is open.

Hock Tan: I'm not playing a SAM game here. I'm just giving a trajectory towards where we drew the line on 27 before. So I have no response to is the SAM going up or not.

Speaker Change: You your visibility or the timing in terms of when you might be shipping soon.

Ben Reitzes: Hey, How're you doing thanks, guys, Hey, Hock.

Speaker Change: Ethernet scale up that worked to your customers.

Speaker Change: Networking AI networking was really strong in the quarter.

Speaker Change: The pumpkin scalar.

Speaker Change: And then it seemed like it must have beat expectations. I was wondering if you could just talk about the networking in particular, what what caused that and how much of that is your acceleration into next year.

Hock Tan: Stop talking about SAM now. Thanks. Oh, okay. Thank you.

Speaker Change: Scaleup.

Speaker Change: Well scale up.

Speaker Change: Very rapidly converting to Ethernet now.

Vivek Arya: One moment for our next question and that will come from the line of Vivek Arya with Bank of America. Your line is open. Thanks for taking my question. I had a near and then a longer-term question on the XTU business. So Hock, for near-term, if your networking upsided in Q2 and overall AI was in line, it means XTU was perhaps not as strong. I realize it's lumpy, but anything more to read into that, any product transition or anything else? So just a clarification there.

Speaker Change: Very much so.

Speaker Change: I want a fairly narrow band of Hyperscale customers.

Ben Reitzes: And when do you think you see tomahawk kicking in as part of that acceleration. Thanks.

Speaker Change: Scale up.

Speaker Change: It's very much Ethernet.

Ben Reitzes: Well.

Ben Reitzes: I think the.

Ben Reitzes: Network AI networking.

Speaker Change: Thank you one moment for our next question.

Ben Reitzes: You probably would know our goals.

Speaker Change: Yeah.

Ben Reitzes: Goes hand in hand with deployment of AI accelerator clusters. It is it doesn't deploy on the timetable that is very different from the way the.

Speaker Change: And that will come from the line of Stacy <unk> with Bernstein. Your line is open.

Hock Tan: It isn't. It doesn't deploy on a timetable that's very different from the way that... Accelerators get deployed, where there they are. Spews, or G.P. . It does happen, and they deploy a lot in scale out. Wei, Eternet, of course. It's not just a choice or protocol, but it's also increasingly moving into the space of what we all call scale up within those data centers, where you have much higher, more than we originally thought, consumption or density of switches than you have in the scale out scenario. In fact, the increased density in scale up is... 5 to 10 times more than in scale-out.

Speaker Change: Yeah.

Stacy: Hi, guys. Thanks for taking my questions I still wanted to follow up on that AI 'twenty 'twenty six question I wanted to just put some numbers on it.

Hock Tan: And then longer-term, you know, you have should we look forward to and what milestones are you watching to give you the confidence that you can now start adding their addressable opportunity into your 27 or 28 or other numbers? Like how do we get the confidence that these projects are going to turn into revenue in some, you know, reasonable timeframe from now? Thank you.

Ben Reitzes: Our salaries does get deployed whether they are.

Ben Reitzes: Ex feels all Gpus it does happen.

Speaker Change: To make sure I've got it right. So if you you did 60% in the first three quarters of this year.

Ben Reitzes: Deploy a lot in scale out.

Speaker Change: If you grow 60% year over year in Q4. It puts you at like I don't know five 8 billion sounds like 19 or $20 billion for the year.

Ben Reitzes: Ethernet of course.

Ben Reitzes: The choice of protocol, but it's also increasingly moving into the space.

Speaker Change: And then are you are you, saying you're going to grow 60%.

Speaker Change: In 2026 would put your 30 billion Clos and AI revenues for 'twenty, two because I just want to make is that the math.

Ben Reitzes: What we all call scale up within those data centers, where you have much higher more than we originally thought.

Hock Tan: Okay, on the first part that you're asking it You know, it's like, you're trying to be, you're trying to count how many angels on the head of a pin here. I mean, whether it's XPU or networking. Networking is hard, but that doesn't mean XPU is any softer. It's very much along the trajectory we expect. There's no lumpiness. There's no softening. It's pretty much what we expect. the short-term trajectory.

Speaker Change: And to communicate to us directly.

Speaker Change: Doing the math I'm, giving you the trend.

Ben Reitzes: Samson or density of switches than you have in the scale out scenario.

Harlan Sur: But I did I answer that question I think Harlan asked earlier.

Speaker Change: The rate we are seeing that.

Ben Reitzes: It's increased density and scale up is.

Speaker Change: Now so far in fiscal 'twenty five.

Ben Reitzes: Five to 10 times more than in scale out and that's the part that kind of pleasantly surprised us and which is why in this past quarter Q2.

Speaker Change: And presumably continue we don't see any reason why it doesn't given the lead time visibility in 'twenty five.

Hock Tan: And that's the part that kind of pleasantly surprises, and which is why in this past quarter Q2... The AI networking portion continues at about 40% from when we reported a quarter ago for Q1. And at that time I said I expect it to drop. It hasn't.

Speaker Change: What we are seeing today based on what we have visibility on 26.

Ben Reitzes: The AI networking portion continues at about 40% from when we reported a quarter ago for Q1.

Speaker Change: To be able to.

Speaker Change: Ram.

Speaker Change: This AI revenue in the same.

Ben Reitzes: And at that time, I said I expect it to drop.

Speaker Change: Trajectory.

Speaker Change: Yes.

Ben Reitzes: It hasn't.

Speaker Change: But Sam going up so as you see them going up as well because now you have influence on top of training. So it was a scam Sam still sits in the 90 or is the same hiring how high is high.

Hock Tan: and and your thoughts on Tomahawk driving acceleration for next year and when it kicks in. Oh, Tomahawk 6. Oh, yeah, that's extremely strong interest. Now, we're not shipping big orders or any orders other than basic proof of concepts out to customers. But there is tremendous demand for this new 102 terabit per second Tomahawk Thanks, Hock.

Ben Reitzes: And and your thoughts on Tomahawk driving acceleration for next year and when it kicks in.

Hock Tan: In terms of going on to 27, no, we are not updating any numbers here. We six months ago, we drew a sense for the size of the SAM based on... You know, a million. The Union of Avid pedals. We'll be happy to give an update to the audience. But right now though, in today's prepared remarks and answering a couple of questions, we have, we are, as we are doing, as we have done here, we are intending to give you guys more visibility what we think. Growth Trajectory in 26. Thank you. One moment for our next question.

Ben Reitzes: Oh, Tom out six Oh, yeah, that's extremely strong interest.

Speaker Change: As you see it I don't I'm not playing the same game here I'm, just giving you trajectory towards where we draw the line on 27 before so I have no. No response, though is the same going up but not stopped talking about the same now.

Ben Reitzes: No we're not shipping.

Ben Reitzes: Big orders or any orders or other than basically.

Ben Reitzes: Concepts out to customers, but there is tremendous demand for this new 102 terabits.

Speaker Change: [laughter] Oh, Oh, okay. Thank you.

Ben Reitzes: Terabits per second.

Speaker Change: One moment for our next question.

Ben Reitzes: Tomahawk switches.

Speaker Change: Yeah.

Conference Operator: And that will come from the line of Vivek Arya with Bank of America. Your line is open.

Ben Reitzes: Thanks Hock.

Operator: Thank you. One moment for our next question.

Ben Reitzes: Thank you one moment for our next question.

Vivek Arya: Thanks for taking my question I had a near and then a longer term question on the ex U S business. The Hawk for near term if youre networking upside in Q2 and overall the I wasn't mine. It means extra you was perhaps not as strong I realize it's lumpy, but anything more to read into that any product.

Blayne Curtis: And now we'll come from the line of Blayne Curtis with Jeffries. Your line is open. Hey, thanks, and great results. I just want to ask, maybe following up on the scale out opportunity. So today, I guess your main customer is not really using an MVLink switch style scale up. I'm just kind of curious, your visibility or the timing in terms of when you might be shipping, you know, a switch Ethernet scale up network to your customer.

Speaker Change: And that will come from the line of Blayne Curtis with Jefferies. Your line is open.

Blayne Curtis: Hey, Thanks, and great results I just wanted to ask maybe following up on the scale out opportunity. So today I guess your main customers not really using kind of an empty links with style scale up I'm just kind of curious.

Blayne Curtis: Your visibility or the timing in terms of when you might be shipping.

Vivek Arya: Or anything else, so just a clarification there.

Vivek Arya: And then longer term you have outlined.

Blayne Curtis: Ethernet scale up network to your customers.

Speaker Change: Number of additional customers that you're working with what milestones should we look forward to and what milestones are you watching.

Hock Tan: How does poking scale up? Scale, scale up. Well, scale up. very rapidly converting to Ethernet. Very much so. For our fairly narrow band of hyperscale customers...

Blayne Curtis: The pumpkin scalar.

Blayne Curtis: Scale up well.

Blayne Curtis: Well scale up very rapidly converting to Ethernet now.

Speaker Change: To give you the confidence that you can now start adding that addressable opportunity into your 27 or 28 or other.

CJ Muse: and that will come from the line of CJ Muse with Cantor Fitzgerald. Your line is open. Yeah, good afternoon. Thank you for taking the question. I was hoping to follow up on Ross's question regarding inference opportunity. Can you discuss workloads that are optimal that you're seeing for custom silicon? And that over time, what percentage of your XPU business could be inference versus training? Thank you.

Blayne Curtis: Three months so this fall.

Blayne Curtis: Our fairly narrow band of Hyperscale customers.

Speaker Change: Like how do we get the confidence that these projects are going to turn into revenue in some reasonable timeframe from now thank you.

Hock Tan: Gayle R. is very much Ethernet. Thank you.

Blayne Curtis: Scale up.

Blayne Curtis: It's very much Ethernet.

Speaker Change: Okay on the first spun that youre asking.

Speaker Change: It's like you're trying to be Youre trying to count how many angels on the head of opinion, I mean windows XP, Oh networking networking is hard but that doesn't mean excuse any source suffer.

Blayne Curtis: Thank you one moment for our next question.

Operator: One moment for our next question.

Hock Tan: I think there's no differentiation between training and inference in using merchants. Accelerators versus custom accelerators. I think that all under the whole premise Behind. Going Towards Custom Accelerators Continues, which is... Not a matter of cost alone, it is that as custom accelerators get used. and get developed on a roadmap with any particular... hyposcaler That's a learning curve, a learning curve on how they could optimize the way the algorithms on their large language models get... Written and Typed to Silicon Valley. And that ability to do so is a huge value added in creating algorithms that can drive their LLMs to higher and higher performance.

Blayne Curtis: Yeah.

Stacy Rasgon: And that will come from the line of Stacy Rasgon with Bernstein. Your line is open. Hi, guys. Thanks for taking my questions. Um, Hock, I still wanted to follow up on that AI 2026 question. I want to just put some numbers on it, just just to make sure I've got it right. So if you you did 60% in the first three quarters of this year, if you grow 60% year over year in Q4 and put you at like, I don't know, 5.8 billion, something like 19 or 20 billion for the year. And then are you saying you're going to grow 60% in 2026, which would put you 30 billion plus in AI revenues for 2026?

Speaker Change: And that will come from the line of Stacy <unk> with Bernstein. Your line is open.

Speaker Change: Okay.

Stacy: Hi, guys. Thanks for taking my questions.

Speaker Change: It's very much along the trajectory we expect it to be.

Blayne Curtis: I still wanted to follow up on that AI 2026 question I wanted to just put some numbers on it.

Speaker Change: That's known Lumpiness, there still softening, it's pretty much what we expect.

Blayne Curtis: Just to make sure I've got it right. So if you you did 60% in the first three quarters of this year.

Speaker Change: The trajectory to go so far.

Blayne Curtis: If you grow 60% year over year in Q4. It puts you at like I don't know five 8 billion sounds like 19 or $20 billion for the year.

Speaker Change: And into next quarter, Israel, and probably beyond so we have a fairly it's a family.

Blayne Curtis: And then are you are you, saying you're going to grow 60%.

Speaker Change: I guess in our view a fairly clear visibility on the short term trajectory.

Blayne Curtis: In 2026 would put you 30 billion Clos and AI revenues for 2020. So I'm just wondering is that the math that you're trying to communicate to us directly.

Hock Tan: I just want to know, is that the math that you're trying to communicate to us directly? I think you're doing the math. I'm giving you the trend. But I did answer that question, I think Harlan asked earlier. The rate we are seeing now so far in Fiscal 25 and we'll presumably continue. We don't see any reason why it doesn't give a lead time visibility in 25. What we're seeing today, based on what we have visibility on 26th, is to be able to Ram up this AI revenue in the same trajectory. So is the SAM going up?

Speaker Change: In terms of going onto 27, no. We are not updating any numbers here, we six months ago withdrawal essential the size of the same based on you.

Blayne Curtis: Doing the math I'm, giving you the trend.

Blayne Curtis: And but I did I answer that question I think Harlan asked earlier.

Speaker Change: No.

Speaker Change: Million.

Blayne Curtis: The rate we are seeing that.

Speaker Change: <unk> be a million GPU XP clusters for three customers and that's still very valid at that point that you'll be done by and we have not provided any further updates here nor do we intend them to at this point when we can have better visibility a clearer sense of wherever you are.

Blayne Curtis: Now so far in fiscal 'twenty five.

Blayne Curtis: And presumably continue we don't see any reason why it doesn't given lead time visibility in 'twenty five.

Blayne Curtis: What we are seeing today based on what we have visibility on 26.

Hock Tan: much more than basically a segregation approach between hardware. and the. You've literally combined end-to-end hardware and software as they take that journey. And it's a journey. They don't learn that in one year. Do it a few cycles, get better and better at it, and then learn the value, the fundamental value. Creating Your Own Hardware. versus using a third-party merchant. that you are able to optimize. way high performance than you otherwise. and we see that happening. Thank you. One moment for our next question.

Blayne Curtis: Is to be able to.

Speaker Change: Probably won't happen until 'twenty six.

Blayne Curtis: Ram this AI revenue in the same trajectory.

Speaker Change: We'll be happy to give you an update to the audience, but right now though.

Blayne Curtis: Yes.

Speaker Change: In today's prepared remarks, and answering a couple of questions. We have we are as we are doing it as we have done yet.

Blayne Curtis: But Sam going up so as you see them going up as well because now you have influence on top of training. So as the Sam Sam So 60 to 90 or is the same higher now.

Hock Tan: So is the SAM going up as well? Because now you have inference on top of training. So is the SAM still 60 to 90? Or is the SAM higher now as you see it?

Speaker Change: We are intending to give you guys more visibility what we have seen.

Hock Tan: I'm not playing a SAM game here. I'm just giving you a trajectory towards where we drew the line on 27 before. So I have no response to is the SAM going up or not.

Blayne Curtis: I don't I'm not playing the same game here I'm, just giving you a trajectory towards where we draw the line on 27 before so I have no. No response, though it's the same going up but not stopped talking about the same now.

Speaker Change: The growth.

Speaker Change: Sectary in 'twenty six.

Speaker Change: Uh huh.

Speaker Change: Thank you one moment for our next question.

Hock Tan: Stop talking about SAM now. Thanks. Oh, okay. Thank you.

Blayne Curtis: Thanks.

Conference Operator: And that will come from the line of C. J Muse with Cantor Fitzgerald. Your line is open.

Blayne Curtis: Okay. Thank you.

Operator: One moment for our next question and that will come from the line of Vivek Arya with Bank of America. Your line is open. Thanks for taking my question. I had a near and then a longer-term question on the XTU business. So, Hock, for near-term, if you're networking upsided in Q2 and overall AI was in line, it means XTU was perhaps not as strong. I realize it's lumpy, but anything more to read into that, any product transition or anything else, so just a clarification there. And then longer-term, you know, you have outlined a number of additional customers that you're working with.

Blayne Curtis: One moment for our next question.

Speaker Change: Yes, good afternoon, and thank you for taking the question I was hoping to follow up on Ross's question regarding inference opportunity can you discuss workloads to the optimal that youre seeing for custom silicon and that over time, what percentage of your <unk> business could be influenced versus training. Thank you.

Blayne Curtis: Yeah.

Speaker Change: And that will come from the line of Vivek Arya with Bank of America. Your line is open.

Vivek Arya: Thanks for taking my question I had a near and then a longer term question on the <unk> business.

Karl Ackerman: and that will come from the line of Karl Ackerman with BNP Paribas. Your line is open. Thank you. Hock, you spoke about the much higher content opportunity in scale-up networking. I was hoping you could discuss how important is demand adoption for co-package optics in achieving this 5 to 10x higher content for scale-up networks? Or should we anticipate much of the scale-up opportunity will be driven by Tomahawk and ThorNix? Thank you.

Vivek Arya: Look for near term, if youre networking upside in Q2 and overall.

Speaker Change: I think there's no differentiation between training and inference and using.

Vivek Arya: Wasn't mine that means ex EU was perhaps not as strong I realize it's lumpy, but anything more to read into that any product transition or anything else. So just a clarification there.

Speaker Change: Merchant.

Speaker Change: Accelerators versus customer accelerators, I think they're all under.

Speaker Change: The whole premise.

Vivek Arya: And then long ago.

Speaker Change: Behind.

Vivek Arya: Have outlined.

Speaker Change: Going towards custom accelerators.

Speaker Change: Number of additional customers that youre working with what milestones should we look forward to and what milestones are you watching to give you. The confidence that you can now start adding that addressable opportunity into your 27 or 28 or other numbers like how do we get the confidence that these projects aren't going to turn into revenue.

Vivek Arya: What milestones should we look forward to and what milestones are you watching to give you the confidence that you can now start adding their addressable opportunity into your 27 or 28 or other numbers? Like, how do we get the confidence that these projects are going to turn into revenue in some, you know, reasonable timeframe from now? Thank you.

Speaker Change: <unk> continues which is.

Speaker Change: It's not a mental half alone.

Speaker Change: Is that is custom accelerators get use and get develop on a on a roadmap if any.

Hock Tan: I'm trying to decipher this question of yours, so let me try to answer it perhaps in a way I think you want me to clarify. First and foremost, I think most of what's scaling up, a lot of the scaling up that's going on, as I call it, which means a lot of... is done on copper, copper interconnection. And because, you know, that's the size of the size of this of this Scale-up Cluster, still not that huge yet, that you can get away with. And they're still doing it, mostly they're doing it. at some point soon.

Speaker Change: Our hyperscale.

Speaker Change: In some reasonable timeframe from now thank you.

Speaker Change: Hyperscale huh.

Speaker Change: Oh, that's a learning curve and learning curve on how they could optimize.

Hock Tan: Okay, on the first part that you are asking. You know, it's like, you're trying to be, you're trying to count how many angels on the head of a pin here. I mean, whether it's XPU or networking. Networking is hard, but that doesn't mean XPU is any softer. It's very much along the trajectory we expect. There's no lumpiness. There's no softening. It's pretty much what we expect.

Speaker Change: Okay on the first spun that you're asking.

Speaker Change: The way deal and the algorithms on their launch language models guests.

Speaker Change: It's like you're trying to be Youre trying to count how many angels on the head of opinion.

Speaker Change: Griffin.

Speaker Change: Whether it's XP.

Speaker Change: Oh networking networking is hot but that doesn't mean excuse any source suffered.

Speaker Change: Hi.

Speaker Change: <unk>.

Speaker Change: Silicon.

Speaker Change: And then the ability to do so it's a huge value add the increasing algorithms that can drive that allo ams to higher and higher performance much more than.

Speaker Change: It's very much along the trajectory we expect it to be.

Speaker Change: That's that's known Lumpiness, there still softening, it's pretty much what we expect.

Hock Tan: project is In terms of going on to 27, no, we are not updating any numbers here. We six months ago, we drew a sense for the size of the SAM based on... You know, a million.

Speaker Change: The trajectory to go so far.

Speaker Change: Basically.

Speaker Change: And into next quarter, Israel, and probably beyond so we have a fairly it's a family.

Speaker Change: Segregation that approach between hardware.

Speaker Change: And the software is that you were literally combined end to end hardware and software as they.

Speaker Change: I guess in our view.

Hock Tan: I believe when you start trying to go beyond maybe 72. Interconnect. You may have to push towards a different protocol by protocol mode at a different medium from copper. And when we do that, yeah, perhaps then things like exotic stuff like cold packaging might become, of silicon with optical, might become relevant. But truly what we really are talking about is that at some stage, as the clusters get larger, which means... Scale up becomes much bigger and you need to interconnect. And that's when optical will start replacing copper. And when that happens, the question is, what's the best way to deliver on optical?

Speaker Change: Fairly clear visibility on the short term trajectory.

Speaker Change: That journey and it's a journey they don't learn that in one year doing a few cycles get better and better at it and analyze them.

Speaker Change: In terms of going on to 'twenty seven no. We are not updating any numbers here. We are six months ago. We draw the same so the size of the same base on.

Speaker Change: The fundamental value in creating your own hardware.

Speaker Change: You know me.

Speaker Change: X P a million GPU XP clusters for three customers and Thats still very valid at that point that you'll be done by and we have not provided any further updates here nor do we intend them to at this point when we can have better visibility a clearer sense of.

Speaker Change: Buses using third party merchant silicon.

Speaker Change: You are able to optimize.

Hock Tan: Website Copyright 2018 NET Free Tudo Copyright 2018 NET Free Tudo Copyright 2018 NET Free Tudo I will be happy to give an update to the audience. but right now don't Today's prepared remarks and answering a couple of questions We have, we are, as we are doing, as we have done yet, we are intending to give you guys more visibility what we've seen. Growth Trajectory in 26. Thank you.

Speaker Change: Youll software to the hardware and event and eventually achieve.

Speaker Change: Weihai performance than you otherwise could.

Speaker Change: See that happening.

Speaker Change: Wherever you are and then probably won't happen until 2006.

Speaker Change: Yeah.

Speaker Change: Thank you one moment our next question.

Speaker Change: We'll be happy to give you an update to the audience, but right now though.

Conference Operator: And that will come from the line of Karl Ackerman with BNP Paribas. Your line is open.

Speaker Change: In today's prepared remarks, and answering a couple of questions. We have we are as we are doing as we have done yet.

Speaker Change: Yeah. Thank you.

Speaker Change: Hockey spoke about the the.

Speaker Change: We are intending to give you guys more visibility what we have seen.

Speaker Change: The much higher content opportunity and scale up networking I was hoping you could discuss how important is demand adoption for cuss co package optics in it.

Speaker Change: The growth trajectory in 'twenty six.

Speaker Change: Okay.

Speaker Change: Achieving this followed connect higher content for scarp networks.

Speaker Change: Thank you one moment our next question.

CJ Muse: One moment for our next question. And that will come from the line of CJ Muse with Cantor Fitzgerald.

Speaker Change: Or should we anticipate much of a scale up opportunity will be driven by tomahawk and toward next thank you.

Hock Tan: And one way is cold package. but it's on the only way. You can just simply use, continue to use, perhaps plug a... at low cost. in which case then you can interconnect. The bandwidth, the radix of a switch, and our switch is now 512. Connections. So you can now connect all these XPUs. This is from Copper. connecting GPU-to-GPU to... Connecting it and the stamp in that move will be and it's not necessarily called Package Optics, though that's definitely one path we apply.

Speaker Change: And that will come from the line of C. J Muse with Cantor Fitzgerald. Your line is open.

Hock Tan: Your line is open. Yeah, good afternoon. Thank you for taking the question. I was hoping to follow up on Ross's question regarding inference opportunity. Can you discuss workloads that are optimal that you're seeing for custom silicon? And that over time, what percentage of your XPU business could be inference versus training? Thank you. I think there's no differentiation between training and inference in using merchants. Accelerators versus custom accelerators.

Speaker Change:

Speaker Change: Yes. Good afternoon. Thank you for taking the question I was hoping to follow up on Ross's question regarding in for its opportunity can you discuss what closer to optimal that youre seeing for custom silicon and then over time, what percentage of your <unk> business could be influenced versus training. Thank you.

Speaker Change: I mean I'm trying to decipher. This question of yours. So let me try to answer it perhaps in a way I think you wanted me to clarify.

Speaker Change: First and foremost I think most of what's scaling up.

Speaker Change: Yeah.

Speaker Change: So a lot of the scaling up that go in as I call, It which means a lot of.

Speaker Change: I think there's no differentiation between training and inference and using.

Speaker Change: Merchant.

Speaker Change: HBU or GPU to GPU interconnect.

Speaker Change: Accelerators versus customer accelerators, I think they're all under the the whole premise behind.

Hock Tan: I think that all under the whole premise behind Going Towards Custom Accelerators Continues, which is... Not a matter of cost alone, it is that as custom accelerators get used. and get developed on a roadmap with any particular... hyposcaler That's a learning curve, a learning curve on how they could optimize the way the algorithms on their large language models get... Written by Christopher Rakesh, Aaron Rakesh, Joshua Buchalter, Kirsten Spears, Benjamin Reitzes, Charlie Kawwas, William Stein and Typed to select. And that ability to do so is a huge value added. Today, we are creating algorithms that can drive the LLMs to higher and higher performance.

Speaker Change: It's done on copper copper interconnects.

Speaker Change: And because you know.

Speaker Change: Going towards.

Speaker Change: The size of the size of this oh this.

Speaker Change: Custom accelerators.

Speaker Change: <unk> continues which is.

Speaker Change: Scale cluster still not that huge yet they just can't get away with copper using copper interconnect.

Speaker Change: It's not a mental half alone.

Speaker Change: It is that as custom accelerates does get us and develop on them on a road map with any particular hi.

Speaker Change: And they're still doing it mostly they are doing it today.

Speaker Change: At some point soon.

Speaker Change: I believe when you start trying to go beyond maybe 72 GPU to GPU.

Speaker Change: Hyper scalar.

Speaker Change: Oh, that's a learning curve and learning curve on how they could optimize.

Speaker Change: The way deal.

Speaker Change: <unk> you may have to push towards a different protocol by protocol motor the different media.

Speaker Change: The rhythms on their large language models guests.

Speaker Change: Griffin.

Speaker Change: Copper.

Speaker Change: Hi.

Hock Tan: Very clear. Thank you.

Speaker Change: True optical.

Speaker Change: <unk>.

Speaker Change: Silicon.

Speaker Change: And when we do that yeah, perhaps then.

Unknown Executive: And one moment for our next question. And that will come from the line of Joshua Buchalter with TD Cowen. Your line is open. Hey guys, thank you for taking my question. I realize it's a bit nitpicky, but I wanted to ask about gross margins in the guide. So your revenue implies sort of $800 million incremental increase, but gross profit up, I think, $400 to $450 million, which is kind of pretty well below corporate average fall through. I appreciate that CEMI is dilutive and custom is probably diluted within CEMI, but anything else going on with margins that we should be aware of?

Speaker Change: And then the ability to do so it's a huge value add in creating our algorithms that can drive the allo ams to higher and higher performance much more than.

Speaker Change: Uh huh.

Speaker Change: Exalting stopped like co packaging might be can fall off.

Speaker Change: Silicon with optical might become relevant but truly what we really are talking about is that at some stage as the clusters get larger.

Hock Tan: much more than basically a segregation approach between hardware. and the folks. is that you've literally combined end-to-end hardware and software as they take that journey. And it's a journey. They don't learn that in one year. Do it a few cycles, get better and better at it, and then realize their value, the fundamental value. Creating Your Own Hardware versus using a third-party... that you are able to optimize. your software to the hardware and eventually achieve... way high performance than you otherwise. and we see that happening. Thank you.

Speaker Change: Basically.

Speaker Change: Segregation approach between hardware.

Speaker Change: And which means scale becomes much bigger and you need to interconnect Jeep.

Speaker Change: And the software is there.

Speaker Change: As you will literally combined end to end hardware and software.

Speaker Change: <unk> or experian to each other and scale up.

Speaker Change: Take that journey and it's a journey they don't run that in one year doing a few cycles get better and better at it.

Speaker Change: Many more than just 70 to a 100, maybe even hiring 28, you start growing more and more you want to use our optical interconnect simply because of distance and that's when optical will start replacing copper and when that happens. Our question is what's the best way to them.

Hock Tan: And how should we think about the margin profile of custom longer term as that business continues to scale and diversify? Thank you. Yeah, we've historically said that the XPU margins are slightly lower than the rest of the business other than wireless. And so there's really nothing else going on other than that. It's just just exactly what I said that the majority of it quarter over quarter, it's 130 basis point decline is being driven by more XPU. You know, there are more moving... Parts here, then your simple analysis proves here, and I think your simple analysis is totally wrong in that regard.

Speaker Change: And Dan lies.

Speaker Change: The fundamental value in creating your own hardware.

Speaker Change: Versus using third party merchant silicon.

Speaker Change: You are able to optimize.

Speaker Change: Deliver on optical.

Speaker Change: Youll software to the hardware and event and eventually achieve.

Speaker Change: One way is co package optics.

Speaker Change: But they still do you only wish.

Speaker Change: High performance than you otherwise could.

Speaker Change: You can't just simply use continue use perhaps plausible.

Speaker Change: And we see that happening.

Speaker Change: Yeah.

Speaker Change: Low cost optics, and rescale venues can interconnect.

Speaker Change: Thank you one moment our next question.

Karl Ackerman: One moment for our next question. and that will come from the line of Karl Ackerman with BNP Paribas.

Speaker Change: Yeah.

Speaker Change: The band with the ratings of our switch and our switches down 512.

Speaker Change: And that will come from the line of Karl Ackerman with BNP Paribas. Your line is open.

Hock Tan: Your line is open. Thank you. Hock, you spoke about the much higher content opportunity in scale-up networking.

Unknown Executive: All right, and thank you.

Speaker Change: Connections. So you can now connect.

Karl Ackerman: Yeah. Thank you.

Unknown Executive: And one moment for our next question.

Speaker Change: Hockey spoke about.

Timothy Arcuri: And that will come from the line of Timothy Arcuri with UBS. Your line is open. Thanks a lot. I also wanted to ask about scale-up, Hock. So there's a lot of competing ecosystems. There's UA Link, which, of course, you left. And now there's the big GPU company opening up, NVLink. And they're both trying to build ecosystems. And there's an argument that you're an ecosystem of one. What would you say to that debate? Does opening up NVLink change the landscape? And sort of how do you view of your AI network and growth next year? Do you think it's going to be primarily driven by scale-up or would still be pretty scale-out heavy?

Speaker Change: This XP use Gpus 512 for scale up phenomenon and that was huge.

Speaker Change: The much higher content opportunity and scale up networking I was hoping you could discuss how important is the man adoption for.

Hock Tan: I was hoping you could discuss how important is demand adoption for co-package optics in achieving this 510X higher content for scale-up networks? Or should we anticipate much of the scale-up opportunity will be driven by Tomahawk and ThorNix? Thank you.

Speaker Change: It's when you go to optical that's going to happen.

Speaker Change: Co package optics in.

Speaker Change:

Speaker Change: My view, a year or two and we'll be right in the forefront of it and it may be co package optics, which we are very much in development, but it's a locking co package or could just be that's the first step plug a hole or whatever it is I think the bigger question is when does it go from optical.

Speaker Change: Achieving this fired connects higher content per scope networks.

Speaker Change: Or should we anticipate much of a scale up opportunity will be driven by our tomahawk and toward next thank you.

Speaker Change:

Hock Tan: I'm trying to decipher this question of yours, so let me try to answer it perhaps in a way I think you want me to clarify. First and foremost, I think most of what's scaling up, a lot of the scaling up that's going on, as I call it, which means a lot of... is done on copper, copper indicator. And because, you know, that's the size of the size of this of this Scale-up Cluster, still not that huge yet, they can get away with it. And they're still doing it, mostly they're doing it. at some point soon.

Speaker Change: Uh huh.

Speaker Change: Trying to decipher at this question of Yours. So let me try to answer it perhaps in a way I think you wanted me to clarify.

Speaker Change: From copper.

Speaker Change: The nice thing G. P S attribute to optical commenting and.

Hock Tan: Thanks.

Hock Tan: It's, you know, people do like to create platforms and new protocols and systems. The fact of the matter is... JLM can just be done easily and it's currently available. Open Standards, Open Source, Ethernet, just as well, you don't need to create new systems for the sake of... doing something that you could easily be doing in networking in Ethernet. And so, yeah, I hear a lot of this interesting new protocol standards that are trying to be created. And most of them, by the way, are proprietary, much as they like to call it otherwise. What is really open source and open standard is the Internet.

Speaker Change: First and foremost I think.

Speaker Change: Most of what's scaling.

Speaker Change: And that move will be huge.

Speaker Change: There are a lot of the scaling up that go in as I call, It which means a lot of.

Speaker Change: And it's not necessary co package healthy so that's definitely one pad we are pursuing.

Speaker Change: <unk> or GPU to GPU interconnect.

Speaker Change: Very clear thank you.

Speaker Change: It's done on copper copper interconnects.

Conference Operator: And one moment our next question.

Speaker Change: And because you know.

Speaker Change: And that will come from the line of Joshua Buchalter with TD Cowen Your line is open.

Speaker Change: The size of the size of this of.

Speaker Change: This.

Joshua Buchalter: Hey, guys. Thank you for taking my question.

Speaker Change: Scale cluster still not that huge it they just can't get away with copper using copper.

Speaker Change: Realize it's a bit nitpicky, but I wanted to ask about gross margins in the guide. So your revenue implies sort of $800 million incremental increase with gross profit up I think $400 million to $450 million, which is kind of pretty well below corporate average fall through I.

Speaker Change: Interconnect.

Speaker Change: And they're still doing it mostly they are doing it today.

Speaker Change: At some point soon.

Hock Tan: I believe, when you start trying to go beyond maybe seven... Interconnect. You may have to push towards a different protocol by protocol mode at a different medium from copper. And when we do that, yeah, perhaps then things like exotic stuff like cold packaging might become, of silicon with optical, might become relevant. But truly what we really are talking about is that at some stage, as the clusters get larger, which means... Scale up becomes much bigger and you need to interconnect. GPUs or XPUs to each other in scale-up. And that's when optical will start replacing copper.

Speaker Change: I believe.

Speaker Change: When you start trying to go beyond maybe 72 GPU to GPU.

Speaker Change: Hate that term eases dilutive and customers, probably dilutive within semi but anything else going on with margins that we should be aware of it.

Hock Tan: And we believe the Internet will prevail as it does for the last 20 years in traditional networking. There's no reason to create a new standard for something that could be easily done in transferring bits and bytes of data. Got it, Hock. Thank you. And one moment for our next question.

Speaker Change: Interconnect you may have to push towards a different protocol by protocol model different media from copper.

Speaker Change: And how should we think about the margin profile of custom longer term that that business continues to scale and diversify.

Speaker Change: Yeah, we've historically said that the XP your margins are slightly lower than the rest of the business other than wireless and so there's really nothing else going on other than that it's just exactly what I said that the majority of it quarter over quarter at 130 basis point decline is being driven by more axes.

Speaker Change: To optical.

Speaker Change: And when we do that yeah, perhaps then.

Speaker Change: Uh huh.

Speaker Change: Exotic stoplight co packaging might be can fall off.

Speaker Change: <unk> silicon with optical might become relevant but truly what we really are talking about is that at some stage as the clusters get larger.

Christopher Rolland: And that will come from the line of Christopher Rolland with Susquehanna. Your line is open. Thanks for the question. Yeah, my question is for you, Hock. It's a kind of a bigger picture one here. And this kind of acceleration that we're seeing in AI demand, do you think that this acceleration is because of a marked improvement in ASICs or XPUs closing the gap on the software side at your customers? Do you think it's these require token tokenomics around inference, test time compute driving that, for example? What do you think is actually driving the upside here?

Speaker Change: You know there are more moving.

Speaker Change: But here Danielle simple analysis process and I think the simple analysis totally wrong in that regard.

Speaker Change: And which means scale becomes much bigger and you need to interconnect.

Speaker Change: Yeah.

Speaker Change: <unk> all experienced to each other and scale up.

Speaker Change: Alright, thank you.

Speaker Change: And one moment for our next question.

Speaker Change: Many more than just 70 to a 100, maybe even hiring 28, you'll start going more and more you want to use our optical interconnect simply because of distance and that's when opt.

Conference Operator: And that will come from the line of Timothy Arcuri with UBS. Your line is now open.

Speaker Change: Thanks, a lot.

Speaker Change: I also wanted to ask about scale up hockey so theres a lot of competing ecosystems knows you a language course, you left and that is the big GPU companies opening up and be like and they're both trying to build eco systems and there is an argument that youre an ecosystem of one.

Speaker Change: Optical will start replacing copper and when that happens. Our question is what's the best way to deliver.

Hock Tan: And when that happens, the question is, what's the best way to do it? and Deliver on Optical. One way is called package op. but it's not the only way. You can just simply use, continue to use, perhaps plug a... at low cost. In which case then you can interconnect. Ben with the radix of a switch and our switch is down 512 Connections. So you can now connect all these XPUs.

Speaker Change: Deliver on optical one way is co package optics.

Speaker Change: What would you say that that debate does does does opening up MZ link change the landscape and sort of how do you view of your AI networking growth next year do you think is going to be primarily driven by scale up or what's still be pretty scale out heavy thanks.

Speaker Change: But they still do you only wish.

Speaker Change: You can't just simply use continue use perhaps plausible.

Hock Tan: And do you think it leads to a market share shift faster than we were expecting towards XPU from GPU? Thanks. Yeah, interesting question. But no, none of the foregoing that you outlined. It's very simple. The way inference has come out very, very hot lately is, remember, we're only selling to a few. customers, hyperscalers with platforms and LLMs. That's it. There are not that many. And you, we told you how many we have and we haven't increased any. But what is happening is These hyperscalers and those with LNM... me to me to Justify all the spending they're doing.

Speaker Change: Low cost optics, and risque venues can interconnect.

Speaker Change: The bandwidth.

Speaker Change: It's people do like to create platforms and new protocols and system. The fact of the Manta is.

Speaker Change: Ratings of switch and our switches down 512.

Speaker Change: Connections. So you can now connect.

Speaker Change: Scale.

Speaker Change: This <unk> Gpus 512 for scale up phenomenon and that was huge.

Speaker Change: It can just be done easily and it's currently available is.

Speaker Change: Open standards open source Ethernet, just as well just as well you don't need to create new systems for the sake of.

Speaker Change: It's when you go to optical that's going to happen.

Unknown Attendee: The opinions voiced in this meeting are those of the guests, and not necessarily those of from Copper. the next thing, GPU to GPU to all. Connecting it and the stamp in that move will be And it's not necessarily called Package Optics, though that's definitely one path we are taking. Very clear. Thank you. And one moment for our next question.

Speaker Change:

Speaker Change: My view, a year or two and we'll be right in the forefront of it and then maybe co package optics, which we are very much in development, but is locking co package or could just be that's the first step plug a hole or whatever it is I think the bigger question is when does it go for optical and <unk>.

Speaker Change: Doing something that you could easily be doing in networking in Ethernet.

Speaker Change: And so yeah I hear a lot of this interesting new protocol standards that I'm trying to be creative and most of them by the way a proprietary much as they like to call. It otherwise one is really open source and open standards is Ethernet and we believe Ethernet will prevail as it does.

Hock Tan: Doing training makes your frontier models smarter. That's my question. It's almost like science, research and science. Make your frontier models by creating very clever algorithms that consume a lot of compute for training. Training makes us smarter. You want to monetize? And that's what's driving it. Money, as I indicated in my prepared remarks, the drive to justify a return on investment. and a lot of their investment is training. And then return on investment is by creating use cases, a lot of AI use cases, AI Consumption. out there through availability of a lot of And that's what we are now starting to see among a small group of customers.

Speaker Change: From copper.

Speaker Change: Connecting GPS attribute to optical commencing and.

Speaker Change: And that move will be huge.

Speaker Change: And it's not necessary co package, so that definitely one pad we are pursuing.

Speaker Change: And before the last 20 years in traditional networking, there's no reason to create a new standard for something that can be easily done in transferring bits and bytes of data.

Speaker Change: Very clear thank you.

Speaker Change: And one moment our next question.

Joshua Buchalter: and that will come from the line of Joshua Buchalter with TD Cowen. Your line is open. Hey guys, thank you for taking my question.

Speaker Change: And that will come from the line of Joshua Buchalter with TD Cowen Your line is open.

Speaker Change: Yeah.

Speaker Change: Got it thank you.

Speaker Change: And one moment for our next question.

Joshua Buchalter: Hey, guys. Thank you for taking my question.

Hock Tan: I realize it's a bit nitpicky, but I wanted to ask about gross margins in the guide. So your revenue implies sort of $800 million incremental increase, but gross profit up, I think, $400 to $450 million, which is kind of pretty well below corporate average fall through. I appreciate that CEMI is dilutive and custom is probably diluted within CEMI, but anything else going on with margins that we should be aware of? And how should we think about the margin profile of custom longer term as that business continues to scale and diversify?

Speaker Change: Realize it's a bit nitpicky, but I wanted to ask about gross margins in the guide. So your revenue implies sort of $800 million incremental increase with gross profit up I think $400 million to $450 million, which is kind of pretty well below corporate average fall through.

Speaker Change: And that will come from the line of Christopher Rolland with Susquehanna. Your line is open.

Christopher Rolland: Thanks for the question.

Christopher Rolland: Yeah. My question is for you Hock, it's kind of a bigger picture one here and.

Speaker Change: This kind of acceleration that we're seeing in AI demand.

Speaker Change: I appreciate the time, either dilutive and customers probably dilutive within semi but anything else going on with margins that we should be aware of it.

Speaker Change: Do you think that this acceleration is because of a marked improvement and 86.

Speaker Change: And how should we think about the margin profile of custom longer term that that business continues to scale and diversify.

Unknown Executive: Excellent. Thank you. And one moment for our next question.

Hock Tan: Thank you. Yeah, we've historically said that the XPU margins are slightly lower than the rest of the business other than wireless. And so there's really nothing else going on other than that. Exactly what I said, that the majority of it, quarter over quarter, it's 130 basis point decline is being driven by more... You know, there are more moving...

Speaker Change: <unk> X to us closing the gap on the software side.

Speaker Change: We've historically said that the <unk> margins are slightly lower.

Speaker Change: At your customers do you think it's these require token.

Vijay Rakesh: And that will come from the line of Vijay Rakesh with Mizuho. Your line is open. Yeah, thanks. Hey, Hock, just going back on the AI server revenue side. I know you said fiscal 25 kind of tracking to like, up 60% ish growth. If you look at fiscal 26, you know, you have many new customers lapping a meta and probably, you know, you have the four of the six hyperscalers that you're talking about. Would you expect that growth to accelerate into fiscal 26 above that, you know, 60% you talked about? You know, my prepared remarks, which I clarified, that the rate of growth we are seeing in 2025 will sustain into 2026 based on improved visibility and the fact that we're seeing inference coming in on top of the demand for training as the clusters get built up bigger and bigger still stands.

Speaker Change: Then the rest of the business other than wireless and so there's really nothing else going on other than that it's just exactly what I said that the majority of it quarter over quarter at 130 basis point decline is being driven by more axes.

Speaker Change: Token omics around infringe our test time compute driving that for example, what do you think is actually driving the upside here and do you think it leads to a market share shift faster than we were expecting towards XP from GPU.

Speaker Change: You know there are more moving.

Operator: Parts here, then your simple analysis proves here, and I think your simple analysis is totally wrong in that regard. All right, and thank you. Then one moment for our next question.

Speaker Change: But.

Speaker Change: Then your simple analysis process and I think the simple analysis totally wrong in that regard.

Speaker Change: Yeah interesting question about no none of the forgoing that you outlined sort of simple the way inference that's come out.

Speaker Change: Okay.

Speaker Change: Alright, thank you.

Speaker Change: And one moment for our next question.

Speaker Change: Very very hot lately.

Timothy Arcuri: And that will come from the line of Timothy Arcuri with UBS. Your line is open. Thanks a lot. I also wanted to ask about scale up, Hock. So there's a lot of competing ecosystems. There's UA Link, which, of course, you left. And now there's the big GPU company opening up, NVLink. And they're both trying to build ecosystems. And there's an argument that you're an ecosystem of one. What would you say to that debate? Does opening up NVLink change the landscape? And how do you view of your AI network and growth next year? Do you think it's going to be primarily driven by scale up or would still be pretty scale out heavy things?

Speaker Change: And that will come from the line of Timothy Arcuri with UBS. Your line is now open.

Speaker Change: No remember, we're only selling to a few.

Timothy Arcuri: Thanks, a lot.

Speaker Change: Customers Hyperscale is voice platforms and L. L M. That's it no not that much.

Speaker Change: I also wanted to ask what scale up hock, so theres a lot of competing ecosystems. As you a language course, you left and that is the big GPU companies opening up Andy Lake and they're both trying to build eco systems and there is an argument that youre an ecosystem of one.

Speaker Change: And we had told you how many we have and we have an increase any but what is happening is.

Hock Tan: I don't think we are getting very far by trying to pass through my words or data here. It's just and we see that going from 2025 into 2026 as the best forecast we have at this.

Speaker Change: This all on this hyperscale us and those with <unk>.

Speaker Change: What would you say that that debate does opening up MZ link change the landscape and sort of how do you view of your AI networking growth next year do you think it's going to be primarily driven by scale up or what's still be pretty scale out heavy thanks.

Speaker Change: Me too.

Speaker Change: Justify all this spending theyre doing doing training makes your frontier models. Some models. That's one question.

Hock Tan: It's, you know, people do like to create platforms and new protocols and systems. The fact of the matter is, scale up. can just be done easily and it's currently available. Open Standards, Open Source, Ethernet, just as well, just as well, you don't need to create new systems for the sake of... doing something that you could easily be doing in networking in Ethernet. And so, yeah, I hear a lot of this interesting new protocols, standards that are trying to be created. And most of them, by the way, are proprietary, much as they like to call it otherwise.

Speaker Change: I know people do like to create platforms and new protocols and system. The fact of the matter is.

Hock Tan: Got it. And on the NVLink, the NVLink Fusion versus the scale up, do you expect that market to go the route of top of the rack where you've seen some move to the Ethernet side in kind of scale out? Do you expect scale up to kind of go the same route? Broadcom do not participate in NVLink. So I'm really not qualified to answer that question.

Speaker Change: Almost like Science research and science manual frontier models by creating very clever algorithms that deep and consumes a lot of compute for training Smart training makes this model.

Speaker Change: Scale.

Speaker Change: Can just be done easily and it's currently available.

Speaker Change: Open standards open source Ethernet, just as well just as well.

Speaker Change: You wanted to monetize.

Speaker Change: Inference.

Speaker Change: And that's what's driving it monetize I indicated in my prepared remarks, the drive to justify.

Speaker Change: Don't need to create new systems for the sake of.

Speaker Change: Doing something that you could easily be doing in networking in Ethernet.

Speaker Change: Our return on investment and a lot of that investment is training and then return on investment is by creating use cases, a lot AI use cases.

Unknown Executive: All right.

Aaron Rakers: Thank you. One moment for our next question. And that will come from the line of Aaron Rakers with Wells Fargo. Your line is open. Yeah, thanks for taking the question.

Speaker Change: And so yeah I hear a lot of this interesting new protocol standards that are trying to be creative and most of them by the way a proprietary much as they like to call. It otherwise one is really open source.

Speaker Change: I can.

Speaker Change: Consumption.

Operator: What is really open source and open standard is the Internet. And we believe the Internet will prevail as it does for the last 20 years in traditional networking. There's no reason to create a new standard for something that could be easily done in transferring. and Bites of Data. Got it, Hock. Thank you. And one moment for our next question.

Speaker Change: All day through availability of a lot of influence.

Hock Tan: I think all my questions on scale up have been asked. But I guess, given the execution that you guys have been able to do with the VMware integration, looking at the balance sheet, looking at the debt structure, I'm curious if, you know, if you could give us your thoughts on how the company thinks about, you know, capital return versus, you know, the thoughts on, you know, M&A and the strategy going forward.

Speaker Change: And stay open standards.

Speaker Change: And we believe Ethernet will prevail as it does.

Speaker Change: And that's what we are now starting to see among a small group of them.

Speaker Change: Before the last 20 years in traditional networking, there's no reason to create a new standard for something that can be easily done in transferring bits and bytes of data.

Speaker Change: Awesome.

Speaker Change: Excellent. Thank you.

Speaker Change: One moment for our next question.

Speaker Change: And that will come from the line of Jive of cash with Mizuho. Your line is open.

Hock Tan: Thank you. Okay, that's a that's an interesting question. And, and I agree, not to not to untimely, I would say, because yeah, we've done a lot of the integration of VMware now. And you can see that in a in a level of free cash flow we're generating from operations. And as we said, the user capital has always been, we're very I guess measured and upfront with a return through dividends. which is half our free cash flow of the preceding year. And, frankly, as Kirsten has mentioned... Three months ago and six months ago during that last two earnings call, the first choice Typically of the other part of the free cash flow is to bring down our debt to a more, to a level that we feel closer to no more than two.

Speaker Change: Yeah, Thanks, Hey, Hock just going back on the.

Speaker Change: Got it thank you.

Speaker Change: And then just nice.

Speaker Change: And one moment our next question.

Speaker Change: I know you said fiscal 'twenty, probably kind of tracking.

Christopher Rolland: And that will come from the line of Christopher Rolland with Susquehanna. Your line is open. Thanks for the question. Yeah, my question is for you, Hock. It's a kind of a bigger picture one here. And this kind of acceleration that we're seeing in AI demand, do you think that this acceleration is because of a marked improvement in ASICs or XPUs closing the gap on the software side at your customers? Do you think it's these require tokenomics around inference, test time compute driving that, for example? What do you think is actually driving the upside here? And do you think it leads to a market share shift faster than we were expecting towards XPU from XPU?

Speaker Change: And that will come from the line of Christopher Roland with Susquehanna. Your line is open.

Speaker Change: 60% ish growth if you look at fiscal 'twenty six you know you have.

Speaker Change: Thanks for the question.

Speaker Change: Many new customers that matter and probably even before the.

Speaker Change: Yes. My question is for you Hock, it's kind of a bigger picture one here and.

Speaker Change: <unk> talked in the past when do you expect that growth to accelerate into fiscal 'twenty six.

Speaker Change: This kind of acceleration that we're seeing in AI demand.

Speaker Change: 60 person you're talking about.

Speaker Change: You know minor minor it in my prepared remarks, which I clarifying that the rate of growth. We are seeing in 'twenty five will sustain into 2006 based on improved visibility and the fact that we're seeing inference coming in on top of the demand for training.

Speaker Change: Do you think that this acceleration is because of a marked improvement in Essex Rx P. R. <unk> closing the gap on the software side.

Speaker Change: At your customers do you think it's.

Speaker Change: These require token.

Speaker Change: It cost us getting people up again, because still stance I don't think we are getting very far by trying to pass through my words, all data him. It just a trick.

Speaker Change: <unk> around infringe.

Speaker Change: Test time compute driving that for example, what do you think is actually driving the upside here and do you think it leads to a market share shift faster than we were expecting towards ex view from GPU. Thanks.

Speaker Change: And we see that going from 25 into 26 is the best.

Hock Tan: a ratio of debt to EBITDA. And that doesn't mean that opportunistically we may go out there and buy back our share. as we did last quarter and indicated by Kirsten when we did 4.2 billion dollars of stock buyback. Now part of it is used to... Basically, when RSUs, employee RSUs, VESTs, Basically use we basically buy back part of the sheds uce to be paying taxes on a vested ERsu but the other part of it I do agree...we used it opportunistically last quarter when we see an opportunistic situation when Basically, we think that it's a good time to buy some shares back.

Speaker Change: The forecast we have at this point.

Hock Tan: Thanks. Yeah, interesting question. But no, none of the foregoing that you outlined. It's very simple. The way inference has come out very, very hot lately is, remember, we're only selling to a few. customers, hyperscalers with platforms and LLMs. That's it. There are not that many. And we told you how many we have, and we haven't increased any. But what is happening is... These hyperscalers and those with LNN... justifying all the spending they're doing. Doing training makes your frontier model smarter. That's all questions. It's almost like science, research and science. Make your frontier models by creating very clever algorithms that consume a lot of compute for training, smarter.

Speaker Change: Got it and on the and really.

Speaker Change: Yeah interesting question about no none of the forgoing that you outline sort of a simple the way inference that's come out.

Speaker Change: And then really chosen versus Kayla.

Speaker Change: Kayla or do you expect that market to go the route of top of the rack. There has been some more data on that site.

Speaker Change: Very very hot lately.

Speaker Change: No remember, we're only selling to a few.

Speaker Change: And kind of scale out do you expect to scale up to kind of go to the same boat.

Speaker Change: Customers Hyperscale as Voip platforms and.

Speaker Change: Thanks, well broadcom do not participate in NV Ling, So I'm really not qualified to answer that question I think.

Speaker Change: That's it there are not that many.

Speaker Change: And you were told you how many we have and we have an increase any but what is happening is.

Speaker Change: Alright, thank you.

Speaker Change: Yeah.

Speaker Change: This all on this hyperscale is and also with <unk>.

Speaker Change: Okay.

Speaker Change: Thank you one moment our next question.

Speaker Change: We need to.

Speaker Change: And that will come from the line of Aaron Rakers with Wells Fargo. Your line is open.

Speaker Change: Justify all this spending theyre doing doing training makes your frontier models smarter. That's one question.

Aaron Rakers: Yes. Thanks for taking the question I think all my questions on scale up has been asked but I guess given the execution that you guys have been able to do with the Vmware integration.

Hock Tan: We do. But having said all that... Our use of cash... outside the dividends would be at this stage used towards reducing our debt.

Speaker Change: Most life Science research and science manual frontier models by creating very clever algorithms that consumes.

Aaron Rakers: Looking at the balance sheet looking at the debt structure I am curious if you could give us your thoughts on how the company thinks about capital return versus the thoughts on M&A and the strategy going forward.

Hock Tan: And I know you're going to ask, what about M&A? Well. The kind of M&A we will do. What in our view would be significant, would be substantial enough that... We need that. In any case, and it's a good use of a free cash flow to bring down debt to, in a way, expand, if not preserve, our borrowing capacity.

Speaker Change: A lot of compute for training Smart training makes us smarter.

Hock Tan: Training makes us smarter. You want to monetize? Inferno. And that's what's driving it. Many times I indicated in my prepared remarks, the drive to justify... A Return on Investment. and a lot of their investment is training. And then return on investment is by creating use cases, a lot of AI use cases, AI. Consumption out there through availability of a lot of And that's what we are now starting. among a small group of customers. Excellent. Thank you.

Speaker Change: You want to monetize.

Speaker Change: Inference.

Aaron Rakers: Yes.

Speaker Change: And that's what's driving it monetize I indicated in my prepared remarks, the drive to justify.

Speaker Change: Okay.

Speaker Change: It's an interesting question and I.

Speaker Change: And I agree not to not dwell on timely I would say because yeah. We've done a lot of the integration of Vmware now and you didn't see that in a in a level of free cash flow, we're generating from operations and as we said the use of capital has always been a win win.

Speaker Change: Our return on investment and a lot of the investment is training.

Speaker Change: And then return on investment is by creating use cases.

Hock Tan: If we have to do another M&A Thank you. One moment for our next question. And that will come from the line of Srini Pajjuri with Raymond James. Your line is open. Thank you. Hock, a couple of clarifications. First, on your 2026 expectation, are you assuming any meaningful contribution from the four prospects that you talked about? No comment. We don't talk about prospects, we only talk about the customers.

Speaker Change: AI use cases.

Speaker Change: <unk>.

Speaker Change: Consumption.

Speaker Change: Very.

Speaker Change: Through availability of a lot of interests.

Speaker Change: I guess measures.

Speaker Change: Upfront, we've a return through dividends.

Speaker Change: And that's what we are now starting to see among a small group of cost.

Speaker Change: Half of our free cash flow of the preceding year.

Speaker Change: Awesome.

Speaker Change: Excellent. Thank you.

Christian: And frankly as Christian has mentioned.

Vijay Rakesh: And one moment for our next question. And that will come from the line of Vijay Rakesh with Mizuho. Your line is open. Yeah, thanks. Hey, Hock, just going back on the AI server revenue side. I know you said fiscal 25 kind of tracking to like, up 60% ish growth. If you look at fiscal 26, you know, you have many new customers ranking as a meta and probably, you know, you have the four of the six hyperscalers that you're talking about. Would you expect that growth to accelerate into fiscal 26 above that, you know, 60% you talked about?

Speaker Change: One moment for our next question.

Speaker Change: And that will come from the line of Jai Rakesh with Mizuho. Your line is open.

Speaker Change: Three months ago, and six months ago to in our last earnings call.

Speaker Change: First choice typically off your other part of the free cash flow is to bring down our debt.

Speaker Change: Yeah, Thanks, Hey, Hock just going back on the.

Speaker Change: Just nice.

Speaker Change: I know you said fiscal 'twenty five.

Speaker Change: To a mall tour level.

Speaker Change: But 60% ish growth if you look at fiscal 'twenty six you have alright.

Srinivas Pajjuri: Okay, fair enough. And then my other clarification is that I think you talked about networking being about 40% of the mix within AI. Is that the right kind of mix that you expect going forward? Or is that, you know, going to materially change as we, I guess, you know, see XPS ramping, you know, going forward? No, I've always said, and I expect that to be the case in going forward in 26, as we grow, that Networking ratio to XPU should be closer in the range of less than 30%.

Speaker Change: We feel closer to no more than two.

Speaker Change: Many new customers that matter and probably before the Hyperscale guys that you talked in the past when do you expect that growth to accelerate into fiscal 'twenty six.

Speaker Change: Our ratio a ratio ratio of debt to EBITDA.

Speaker Change: And that doesn't mean that Opportunistically, we may go out and buyback our shares.

Speaker Change: 60 person you're talking about.

Speaker Change: We did last quarter.

Hock Tan: You know, my prepared remarks, in which I clarify that the rate of growth we are seeing in 2025 will sustain into 2026, based on improved visibility, and the fact that we're seeing inference coming in on top of the demand for training, as the clusters get built up bigger and bigger, still stands. I don't think we are getting very far by trying to pass through my words or data here. It's just a tradition. And we see that going from 2025 into 2026, as the best forecast we have at this Got it. And on the NVLink, the NVLink Fusion versus the scale up, do you expect that market to go the route of top of the rack where you've seen some move to the Ethernet side in kind of the scale out?

Speaker Change: You know minor minor it in my prepared remarks, which I clarified that the rate of growth. We are seeing in 'twenty five will sustain into 2006 based on improved visibility and the fact that we're seeing inference coming in on top of the demand for training.

Speaker Change: And indicated that kicks in when we did it for $2 billion of stock buyback now part of it is used to.

Speaker Change: Basically when Rs employ our issues.

Speaker Change: Yes.

Speaker Change: Basically use a we basically buy bank part of the shares in lieu of pulp used to be paying Texas on the investor a rescue but the other part of it I do I do I mean, we use it opportunistically last quarter, when we see an opportunity a situation when.

Speaker Change: Clusters can be lumpy again biggest steel stance I don't think we are getting very far by trying to pass through my words on data him. It just a trick.

Unknown Executive: Thank you. One moment for our next question.

Joe Moore: and that will come from the line of Joe Moore with Morgan Stanley. Your line is open. Great, thank you. You said you're not going to be impacted by export controls on AI. I know there's been a number of changes since in the industry since the last time you made the call. Is that still the case? And just, you know, can you give people comfort that there's no impact from that down the road?

Speaker Change: And we see that going from 25 into 26 is the best.

Speaker Change: Basically we think that it's a good time to buy some shares back we do but having said all that.

Speaker Change: The forecast we have at this point.

Speaker Change: Got it and on the <unk>.

Speaker Change: Our use of cash.

Speaker Change: <unk> fusion versus Kayla.

Speaker Change: Outside of the dividends would be at this stage used towards reducing our debt.

Speaker Change: Kayla or do you expect that market to grow that at the top of the rack. There has been some more data on that site.

Hock Tan: Nobody can give anybody comfort in this environment, Joe. You know that rules are changing quite dramatically as trade bilateral. Transcripts provided by Transcription Outsourcing, LLC.

Speaker Change: And I know you're going to ask one about M&A well.

Speaker Change: And kind of the scale out do you expect to scale up to kind of go the same route.

Hock Tan: Do you expect scale up to kind of go the same route? Broadcom do not participate in NVLink. So I'm really not qualified to answer that question.

Speaker Change: The kind of and I am in a we will do well.

Speaker Change: In our view would be significant would be substantial enough that we.

Speaker Change: Thanks, well broadcom do not participate in NV Ling.

Speaker Change: So I'm really not qualified to answer that question I think.

Speaker Change: We need that in.

Speaker Change: In any case and it's a good use of our free cash flow to bring down debt due in a way expand if don't preserve our borrowing capacity if we have to do another M&A deal.

Operator: All right, thank you. Thank you. One moment for our next question. And that will come from the line of Aaron Rakers with Wells Fargo. Your line is open. Yeah, thanks for taking the question. I think all my questions on scale up have been asked.

Speaker Change: Alright, thank you.

Speaker Change: Yes.

Speaker Change: Thank you one moment our next question.

Speaker Change: And that will come from the line of Aaron Rakers with Wells Fargo. Your line is open.

Conference Operator: Thank you one moment our next question.

Aaron Rakers: Yes. Thanks for taking the question I think all my questions on scale up has been asked but I guess given the execution that you guys have been able to do with the Vmware integration.

Hock Tan: So, I'd rather not answer that because, no, I don't know whether it will be...

Conference Operator: And that will come from the line of shiny Pensionary with Raymond James Your line is open.

Aaron Rakers: But I guess, given the execution that you guys have been able to do with the VMware integration, looking at the balance sheet, looking at the debt structure, I'm curious if, you know, if you could give us your thoughts on how the company thinks about, you know, capital return versus, you know, the thoughts on, you know, M&A and the strategy going forward. Thank you. Okay, that's a, that's an interesting question. And, and I agree, not to not to untimely, I would say, because, yeah, we have done a lot of the integration of VMware now. And you can see that in a, in a level of free cash flow we're generating from operation.

Unknown Executive: Thank you and we do have time for one final question.

Conference Operator: Okay.

Speaker Change: Thank you Hock couple of clarifications first on your 2026 expectation are you assuming any meaningful contribution from the four prospects that you're talking about.

Speaker Change: Looking at the balance sheet looking at the debt structure I am curious if you could give us your thoughts on how the company thinks about capital return versus the thoughts on M&A and the strategy going forward.

William Stein: And that will come from the line of William Stein with Truist Securities. Your line is open. Great. Thank you for squeezing me in. I wanted to ask about VMware. Can you comment as to how far along you are in the process of converting customers to the subscription model? Is that close to complete or is there still a number of quarters that you should expect that that conversion continues? That's a good question. And so let me start off by saying a good way to measure it is, you know, most of our VMOI contracts are about three, typically three And that was one VMware date before we acquired them, and that's pretty much what we continued to do three years later.

Speaker Change: Yes.

Speaker Change: Okay.

Conference Operator: [laughter] no comment.

Speaker Change: It's an interesting question and I.

Speaker Change: And I agree not to not dwell on timely I would say because yeah. We've done a lot of the integration of Vmware now and you can see that in the in the level of free cash flow, we're generating from operations and as we said the use of capital has always been a win win.

Speaker Change: We don't know okay phosphates, we only talk about customers.

Speaker Change: Okay Fair enough and then my other clarification is that I think you talked about networking being about 40% of the mix within AI is that the right kind of mix that you expect going forward or is that going to materially change as we I guess you know see XP is ramping.

Hock Tan: And as we said, the user capital has always been, we're very... I guess measured and upfront with a return through dividends. This is half our free cash flow of the preceding year. And, frankly, as Kirsten has mentioned... Three months ago and six months ago during those last two earnings calls, the first choice Typically of the other part of the free cash flow is to bring down our debt to a more, to a level that we feel closer to no more than two. uh... ratio ratio of debt to EBITDA uh... and that doesn't mean that opportunistically we may go out there and buy back our share as we did last quarter.

Speaker Change: Very.

Speaker Change: I guess are measured.

Speaker Change: You know going forward.

Speaker Change: I've always said and I expect that to be the case and going forward in 'twenty six industry grow that.

Speaker Change: Upfront, we've a return through dividends.

Speaker Change: Half, our free cash flow of the preceding year.

Speaker Change: Networking should be a ratio to XP you should be closer in the in the range of less than 30%.

Hock Tan: So, based on that, the renewals, we're like two-thirds of the way, almost two-thirds, halfway, more than halfway through the renewals, so we probably have at least another year plus, maybe a year and a half to go. Thank you.

Speaker Change: And frankly right.

Kirsten: Kirsten has mentioned.

Speaker Change: Three months ago, and six months ago to an advanced two earnings call.

Speaker Change: The 40%.

Speaker Change: First choice typically of the other three are part of the free cash flow is to bring down our debt.

Speaker Change: Okay.

Speaker Change: Thank you one moment our next question.

Speaker Change: To a mall tour level.

Conference Operator: And that will come from the line of Joe Moore with Morgan Stanley. Your line is open.

Ji Yoo: And with that, I'd like to turn the call over to Ji Yoo for closing remarks.

Speaker Change: We feel closer to no more than two.

Speaker Change: Great. Thank you.

Ji Yoo: Thank you, operator. Broadcom currently plans to report its earnings for the third quarter of fiscal year 2025. After close of market on Thursday, September 4 2025.

Speaker Change: Our ratio a ratio ratio of debt to EBITDA.

Speaker Change: You said youre not going to be impacted by export controls on AI I know.

Speaker Change: Theres been a number of changes in the industry. Since the last time you made the call is that still the case and just can you give people comfort that youre.

Speaker Change: And that doesn't mean that Opportunistically, we may go out and buyback our shares.

Ji Yoo: A public webcast of Broadcom's earnings conference call will follow at 2pm.

Speaker Change: There is no impact from that down the road.

Speaker Change: We did last quarter.

Hock Tan: and indicated by Kirsten when we did $4.2 billion of stock buyback. Now part of it is used to... Basically, when employee RSUs vest. Basically use of we basically buy back part of the shed Basically, we think that it's a good time to buy some shares back. We do. But having said all that... Our use of cash outside of dividends would be at this stage used towards reducing our debt.

Speaker Change: And indicate that my kitchen, when we did it for $2 billion of stock buyback now part of it is used to.

Speaker Change: Nobody can give anybody comfort in this environment, Joe you know the rules are changing quite dramatically.

Ji Yoo: That will conclude our earnings call today. Thank you all for joining.

Unknown Executive: Operator, you may end the call.

Speaker Change: Trade bilateral.

Speaker Change: Basically when Rs employee RF skus.

Speaker Change: Great agreements continues to be negotiated.

Speaker Change: S based.

Speaker Change: <unk> use a we basically buy bank part of the shares in.

Speaker Change: Why not very very dynamic environment. So I'll be honest I don't I don't know I know as little as probably you probably know more than I do maybe I rephrase, then I know very little about this whole thing about whether there's any export control how they export control will take place.

Speaker Change: Fault used to be paying taxes on the investor a rescue but the other part of it I do I do I mean, we use it opportunistically last quarter, when we see an opportune situation win.

Speaker Change: Basically we think that it's a good time to buy some shares back we do but having said all that.

Speaker Change: Yeah.

Speaker Change: So I'd, rather not answer that because I don't know, but they won't be.

Speaker Change: Our use of cash.

Speaker Change: Outside of the dividends would be at this stage use towards reducing our debt.

Speaker Change: Thank you and we do have time for one final question.

Speaker Change: Okay.

Hock Tan: And I know you're going to ask, what about M&A? Well. The kind of M&A we will do... What in our view would be significant, would be substantial enough that... We need that. And it's a good use of a free cash flow to bring down debt to, in a way, expand, if not preserve, our borrowing capacity. We have to do another M&A.

Speaker Change: And that will come from the line of William Stein with Jewish Securities. Your line is open.

Speaker Change: And I know you're going to ask one about M&A well.

Speaker Change: The kind of M&A, we will do well.

William Stein: Yeah, great. Thank you for squeezing me in I wanted to ask you about Vmware can.

Speaker Change: In our view would be significant would be substantial enough that we.

William Stein: Can you comment as to how far along you are in the process.

Speaker Change: We need that in.

Speaker Change: Converting customers to the subscription model does that.

Speaker Change: In any case and it's a good use of our free cash flow to bring down debt due in a way expand if don't preserve our borrowing capacity if we have to do another M&A deal.

Speaker Change: Flips to complete or is there still a number of quarters that you should expect that.

William Stein: Conversion continues.

William Stein: That's a good question and so let me start out by saying a good way to measure it as you know most of our Vms contracts.

Hock Tan: Thank you. One moment for our next question. And that will come from the line of Srinivas Pajjuri with Raymond James. Your line is open. Thank you. Hock, a couple of clarifications. First, on your 2026 expectation, are you assuming any meaningful contribution from the four prospects that you talked about? Um No comment. We don't talk about the prospect, we only talk about the customer. Okay, fair enough. And then my other clarification is that I think you talked about networking being about 40% of the mix within AI. Is that the right kind of mix that you expect going forward?

Speaker Change: Thank you one moment our next question.

Speaker Change: And that will come from the line of shiny Pensionary with Raymond James Your line is open.

William Stein: Three.

William Stein: Typically three years and that was once Vmware did before we acquired them and that's pretty much. What we continued to do three is ready for digital so.

Speaker Change: Okay.

Speaker Change: Thank you Hock couple of clarifications first on your 2026 expectation are you assuming any meaningful contribution from the four prospects that you're talking about.

William Stein: Based on that the renewals were like two thirds of the way almost at the halfway more than halfway through the renewals. So we'd probably have at least another year, plus maybe a year and a half to go.

Speaker Change: [laughter] no comment.

Speaker Change: We don't know okay phosphates, we only talk about customers.

William Stein: Thank you and with that I'd like to turn the call over to you for closing remarks.

Speaker Change: Okay Fair enough and then my other clarification is that I think you talked about networking being about 40% of the mix within AI is that the right kind of mix that you expect going forward or is that going to materially change as we I guess see XP is ramping.

Conference Operator: Thank you operator.

Speaker Change: <unk> Com currently plans to reported earnings for the third quarter of fiscal year 2025, after close of market on Thursday September four.

Srinivas Pajjuri: Or is that, you know, going to materially change as we, I guess, you know, see XPS ramping, you know, going forward? No, I've always said, and I expect that to be the case. going forward in 26 as we grow. Networking ratio to XPU should be closer in the range of less than 30%.

Speaker Change: 2025 public webcast at Broadcom earnings Conference call will follow at two P M Pacific.

Speaker Change: You know going forward.

Speaker Change: I've always said and I expect that to be the case and going forward in 'twenty six industry grow that.

Conference Operator: That will conclude our earnings call today. Thank you all for joining operator, you may end the call.

Speaker Change: Networking should be a ratio to XP you should be closer in the in the range of less than 30% north of 40%.

Joe Moore: Unknown Attendee, Thank you. One moment for our next question. and that will come from the line of Joe Moore with Morgan Stanley. Your line is open. Great, thank you. You said you're not going to be impacted by export controls on AI. I know there's been a number of changes in the industry since the last time you made the call. Is that still the case? And just, you know, can you give people comfort that there's no impact from that down the road? Nobody can give anybody comfort in this environment, Joe. You know that. Rules are changing quite dramatically, as trade bilateral.

Speaker Change: Thank you one moment our next question.

Joe Moore: And that will come from the line of Joe Moore with Morgan Stanley. Your line is open.

Joe Moore: Great. Thank you.

Speaker Change: You said youre not going to be impacted by export controls on AI I know.

Speaker Change: Theres been a number of changes in the industry. Since the last time you made the call is that still the case and just can you give people comfort yet.

Speaker Change: There is no impact from that down the road.

Speaker Change: Nobody can give anybody comfort in this environment, Joe you know the rules are changing quite dramatically.

Speaker Change: Trade bilateral.

Speaker Change: Great agreements continue to be negotiated.

Speaker Change: It's not really very very dynamic environment, so I'll be honest I, though I don't know I know as little as probably you probably know more than I do maybe I.

Hock Tan: Transcripts provided by Transcription Outsourcing, LLC. I know very little about this whole thing about whether there is any export control, how the export control will take place.

Speaker Change: In which case, then I know very little about this whole thing about.

Speaker Change: Whether there's any export control how the export control will take place, but get too.

Hock Tan: So, I'd rather not answer that because, no, I don't know whether it will be...

Speaker Change: So I'd, rather not answer that because I don't know whether it will be.

Hock Tan: Thank you. And we do have time for one final question. And that will come from the line of William Stein with Truist Securities. Your line is open. Great, thank you for squeezing in. I wanted to ask about VMware. Can you comment as to how far along you are in the process of converting customers to the subscription model? Is that close to complete? Or is there still a number of quarters that we should expect that that conversion continues? That's a good question. And so let me start off by saying a good way to measure it You know most of our VMOA contracts are typically three years.

Speaker Change: Thank you and we do have time for one final question.

Speaker Change: And that will come from the line of William Stein with choice Securities. Your line is open.

William Stein: Great. Thank you for squeezing me in I wanted to ask you about Vmware.

William Stein: Can you comment as to how far along you are in the process.

William Stein: Converting customers to the subscription model is that.

William Stein: Close to complete or is there still a number of quarters that you should expect that.

William Stein: That conversion continues.

William Stein: That's a good question and so let me start off by saying a good way to measure it as you know most of our Vms contracts.

William Stein: About three on it.

William Stein: Typically three years and that was once Vmware did before we acquired them and that's pretty much. What we continued to do three years very true digital so.

William Stein: And that was one VMware date before we acquired them, and that's pretty much what we continue to do, three as well.

William Stein: So, based on that, the renewals, we're like two-thirds of the way, almost two-thirds, halfway, more than halfway through the renewals, so we probably have at least another year plus, maybe a year and a half to go. Thank you.

William Stein: Based on that the renewals were like two thirds of the way almost at the halfway more than halfway through the renewals. So we'd probably have at least another year, plus maybe a year and a half to go.

Speaker Change: Thank you and with that I'd like to turn the call over to you for closing remarks.

Ji Yoo: And with that, I'd like to turn the call over to Ji Yoo for closing remarks. Thank you, operator. Broadcom currently plans to report its earnings for the third quarter of fiscal year 2025. After close of market on Thursday, September 4 2025. A public webcast of Broadcom's earnings conference call will follow at 2pm.

Speaker Change: Thank you operator, Broadcom currently plans to reported earnings for the third quarter of fiscal year 2025, after close of market on Thursday September four.

Speaker Change: 2025 public.

Speaker Change: Public webcast at Broadcom earnings Conference call will follow at two P M Pacific.

Ji Yoo: That will conclude our earnings call today. Thank you all for joining.

Speaker Change: That will conclude our earnings call today. Thank you all for joining operator, you may end the call.

Operator: Operator, you may end the call.

Speaker Change: Okay.

Speaker Change: [music].

Speaker Change: Okay.

Speaker Change: [music].

Speaker Change: Yes.

Speaker Change: [music].

Speaker Change: Okay.

Speaker Change: [music].

Speaker Change: Okay.

Speaker Change: [music].

Speaker Change: Okay.

Speaker Change: [music].

Speaker Change: Yes.

Speaker Change: Yes.

Speaker Change: [music].

Speaker Change: Okay.

Speaker Change: Okay.

Speaker Change: Sure.

Speaker Change: Okay.

Speaker Change: [music].

Speaker Change: Yes.

Speaker Change: Thank you.

Speaker Change: [music].

Speaker Change: Sure.

Speaker Change: [music].

Speaker Change: Okay.

Speaker Change: [music].

Speaker Change: Yes.

Speaker Change: [music].

Speaker Change: Yes.

Speaker Change: Yes.

Speaker Change: Yes.

Speaker Change: Okay.

Speaker Change: Okay.

Speaker Change: [music].

Speaker Change: Okay.

Speaker Change: Thank you.

Speaker Change: [music].

Speaker Change: Thanks.

Speaker Change: Okay.

Speaker Change: [music].

Operator: Welcome to Broadcom's Inc. Second Quarter Fiscal Year 2025 Financial Results Conference Call.

Speaker Change: Welcome to Broadcom, Inc. Second quarter fiscal year 2025 financial results Conference call.

Ji Yoo: At this time, for opening remarks and introductions, I would like to turn the call over to Ji Yoo, Head of Investor Relations of Broadcom Inc. Thank you, operator, and good afternoon, everyone. Joining me on today's call are Hock Tan, President and CEO. Kirsten Spears, Chief Financial Officer and Charlie Kawwas, President, Semiconductor Solutions. Broadcom distributed a press release and financial tables after the market closed. Describing our financial performance for the second quarter of fiscal year 2025. If you did not receive a copy, you may obtain the information from the investor section of the Broadcom's website at Broadcom.com.

Speaker Change: At this time for opening remarks, and introductions I would like to turn the call over to <unk> head of Investor Relations of Broadcom, Inc.

Speaker Change: Thank you operator, and good afternoon, everyone join.

Speaker Change: Joining me on today's call are Hock Tan President and CEO.

Speaker Change: Houston Spirit, Chief Financial Officer, and Charlie Quad, President Semiconductor solutions group.

Speaker Change: Broadcom distributed a press release and financial tables after the market closed describing our financial performance for the second quarter of fiscal year 2025.

Speaker Change: You did not receive a copy you may obtain the information from the investors section of the Broadcom website at Broadcom Dot com.

Ji Yoo: This conference call is being webcast live and an audio replay of the call can be accessed for one year through the investor section of Broadcom's website. During the prepared comments, Hock and Kirsten will be providing details of our second quarter fiscal year 2025 results. Guidance for our third quarter of fiscal year 2025, as well as commentary regarding the business environment. We will take questions after the end of our prepared Please refer to our press release today and our recent filings with the SEC for information on the specific risk factors that could cause our actual results to differ materially from the forward-looking statements made on the In addition to U.S.

Speaker Change: This conference call is being webcast live and an audio replay of the call can be accessed for one year to the investors section of Broadcom website.

Speaker Change: During the prepared comments Hawkins Kirsten will be providing details of our second quarter fiscal year 2025 result.

Speaker Change: Guidance for our third quarter of fiscal year, 2025, as well as commentary regarding the business environment.

Speaker Change: We will take questions. After the end of our prepared comments.

Speaker Change: Please refer to our press release today and our recent filings with the SEC for information on the specific risk factors that could cause our actual results to differ materially from the forward looking statements made on this call.

Speaker Change: In addition to U S GAAP reporting Broadcom reports certain financial measures on a non-GAAP basis.

Ji Yoo: GAAP reporting, Broadcom reports certain financial measures on a non-GAAP A Reconciliation Between Gap and Non-Gap Measures. Included in the table attached to today's press conference.

Speaker Change: A reconciliation between GAAP and non-GAAP measures is included in the table attached to today's press release.

Ji Yoo: Comments made during today's call will primarily refer to our non-GAAP financial I will now turn the call over to Hock. Thank you. Thank you, Ji. and thank you everyone for joining us today. in our fiscal Q2 2025. Total revenue was a record $15 billion, up 20% year-on-year. This 20% year-on-year growth was all organic. as Q2 last year was the first full quarter with VMware. Now revenue was driven by continuous strength in AI semiconductors. and the momentum we have achieved in VMware. Now reflecting excellent all operating leverage. Q2 Consolidated Adjusted EBITDA was $10 billion, up 35% year-on-year.

Speaker Change: Comments made during today's call will primarily refer to are non-GAAP financial results.

Hawk: I will now turn the call over to Hawk.

Hawk: Yes.

Speaker Change: Thank you. Thank you Qi <unk>.

Speaker Change: And thank you everyone for joining us today.

Speaker Change: In our fiscal Q2 2025.

Speaker Change: Total revenue was a record $15 billion up 20% year on year.

Speaker Change: This 20% year on year growth was all organic.

Speaker Change: Q2 last year was the first full quarter with.

Speaker Change: Vmware.

Speaker Change: No revenue was driven by continued strength in AI semiconductors.

Speaker Change: And the momentum we have achieve in Vmware.

Speaker Change: Now, reflecting excellent operating leverage Q.

Speaker Change: Q2, consolidated adjusted EBITDA was $10 billion.

Speaker Change: Up 35% year on year.

Hock Tan: Now let me provide more color. Q2 Semiconductor Revenue was $8.4 billion, with growth accelerating to 17% year-on-year, up from 11% in Q1. and of course, driving this growth. was AI semiconductor revenue of over $4.4 billion, which is up 46% year-on-year and continues the trajectory of nine consecutive quarters of strong growth. We didn't do. Custom AI accelerators grew double digits year-on-year, while AI networking grew over 170% year-on-year. AI Network which is based on Ethernet was robust and represented 40% of our AI revenue. As a standards based open protocol, Ethernet enables one single fabric for both scale out and scale up.

Speaker Change: Now, let me provide more color.

Speaker Change: Q2 semiconductor revenue was $8 4 billion.

Speaker Change: Growth accelerating to 17% year on year up from 11% in Q1.

Speaker Change: And of course driving this growth.

Speaker Change: Also AI semiconductor revenue of over $4 4 billion, which is up 46% year on year and.

Speaker Change: And continues the trajectory of nine consecutive quarters of strong growth.

Speaker Change: Within this customer.

Speaker Change: Custom AI accelerators grew double digits year on year, while AI networking grew over 170% year on year.

Speaker Change: AI networking, which is based on Ethernet was robust and represented 40% of our AI revenue.

Speaker Change: S. A standards based open protocol.

Speaker Change: Ethernet enables one single fed break for both scale out and scale up.

Hock Tan: and remains the preferred choice by our hyperscale customers. Our Networking Portfolio of Tomahawk Switchers Cherico Routers and Nick. is what's driving our success within AI clusters in hyperscale. And the momentum continues with our breakthrough Tomahawk 6 suite. just announced this week. This represents the next generation. 102.4 terabits per second switch capacity. Tomahawk-6 enables clusters of more than 100,000 AI accelerators to be deployed in just two tiers instead of three. This flattening of the AI cluster is huge. because it enables much better performance in training next generation frontier models through a lower latency, higher bandwidth. and lower power.

Speaker Change: And remains the preferred choice by our Hyperscale customers.

Speaker Change: Our networking portfolio.

Speaker Change: Tomahawk switches.

Speaker Change: Sherri Coale routers.

Speaker Change: Nyx.

Speaker Change: It's what's driving our success within AI clusters in Hyperscale.

Speaker Change: Okay.

Speaker Change: And the momentum continues with our breakthrough Tomahawk six suites.

Speaker Change: As announced this week.

Speaker Change: This represents.

Speaker Change: Next generation.

Speaker Change: 102.

Speaker Change: 0.4, Terabits per second switch capacity.

Speaker Change: Come on six enables clusters.

Speaker Change: More than 100000 accelerators to be deployed in just two tiers instead of three.

Speaker Change: This flattening of the AI cluster is huge.

Speaker Change: Because it enables much better performance in training next generation frontier models through a low latency high bandwidth.

Speaker Change: And lower power.

Hock Tan: turning to XPUs, or Customer Accelerator. We continue to make Excellent progress on the multi-year journey of enabling our three customers and four prospects to deploy custom AI accelerators. As we had articulated over six months ago, We eventually expect at least three customers to each deploy 1 million AI accelerated clusters in 2027. largely for straining their frontier model. And we forecast and continue to do so, a significant percentage of these deployments to be custom XP. These partners are still unwavering in their plan to invest. despite this certain economic environment. In fact. What we've seen recently is that they are doubling down.

Speaker Change: Turning to <unk>.

Speaker Change: XP use our customer accelerators.

Speaker Change: We continue to make.

Speaker Change: Excellent progress.

Speaker Change: On the multi year journey of enabling our three customers.

Speaker Change: And for prospects to deploy custom AI accelerators.

Speaker Change: As we had articulated over six months ago.

Speaker Change: We eventually expect at least.

Speaker Change: Three customers to each deploy 1 million E axle.

Speaker Change: Accelerated clusters in 2027.

Speaker Change: Largely for three inning frontier models.

Speaker Change: And we forecast and continue to do so a significant percentage of these deployments.

Speaker Change: Two big customers XP use.

Speaker Change: These partners.

Speaker Change: Still unwavering India plan to invest.

Speaker Change: Despite these certain economic environment.

Speaker Change: In fact.

Speaker Change: What we've seen recently.

Speaker Change: Is that they are doubling down.

Hock Tan: on Inference in order to monetize their platform. And reflecting this, we may actually see an acceleration of XPU demand. into the back half of 2026 to meet urgent demand for inference on top of the demand we have indicated from training. And accordingly, we do anticipate now, our fiscal 2025 growth rate of AI Semiconductor Revenue to sustain. into fiscal 2026.

Speaker Change: On inference in order to monetize their platforms.

Speaker Change: And reflecting this we may actually see.

Speaker Change: A celebration of export demand.

Speaker Change: Into the back half of 'twenty 'twenty six to meet urgent demand for inference on top of the demand we have indicated.

Speaker Change: From training.

Speaker Change: And accordingly, we do anticipate now our fiscal 2025 growth rate.

Speaker Change: <unk> AI semiconductor revenue.

Speaker Change: To sustain.

Speaker Change: Into fiscal 2026.

Hock Tan: Turning to our Q3 out. As we continue our current trajectory of growth. We focus AI semiconductor revenue. to be $5.1 billion, up 60% year-on-year, which would be the 10th consecutive quarter of growth.

Speaker Change: Yeah.

Speaker Change: Turning to Q2, our Q3 outlook.

Speaker Change: As we continue our current trajectory of growth.

Speaker Change: We forecast AI semiconductor revenue.

Speaker Change: To be $5 $1 billion up 60% year on year, which would be the 10th consecutive quarter of growth.

Hock Tan: Now turning to non AI semiconductors in Q2. Revenue of $4 billion was down 5% year-on-year. Non-AI Semiconductor Revenue is close to the bottom. has been relatively slow to recover. but there are bright spots in Q2 broadband. Enterprise Networking and Server Storage Revenues were up sequentially. However... Industrial was down, and as expected, wireless was also down due to seasonality. In Q3, we expect enterprise networking and broadband. continue to grow sequentially. But server storage, wireless, and industrial are expected to be largely flat. And overall, we forecast non-AI semiconductor revenue to stay around $4 billion.

Speaker Change: Yeah.

Speaker Change: Now turning to non AI semiconductors in Q2 revenue of $4 billion was down 5% year on year.

Speaker Change: Non semiconductor revenue is close to the bottom.

Speaker Change: And it's been relatively slow to recover.

Speaker Change: But there are bright spots in Q2 broadband.

Speaker Change: Enterprise networking.

Speaker Change: Server storage revenues were up sequentially.

Speaker Change: However.

Speaker Change: Industrial was down and expect that wireless was also down due to seasonality.

Speaker Change: In Q3.

Speaker Change: We expect enterprise networking and broadband.

Speaker Change: To continue to grow sequentially.

Speaker Change: But server storage wireless and industrial.

Speaker Change: <unk> to be largely flat.

Speaker Change: Overall, we forecast non semiconductor revenue to stay around $4 billion.

Hock Tan: Now let me talk about our infrastructure software segment. Q2 infrastructure software revenue of $6.6 billion was up 25% year-on-year, above our outlook of $6.5 billion. As we have said before, This growth reflects our success in converting our enterprise customers from perpetual vSphere to the full VCF software stack subscription. Customers are increasingly turning to VCF to create a modernized private cloud on-prem, which will enable them... to repatriate workloads from public clubs. while being able to run modern container-based applications. and AI Applications. Of our 10,000 largest customers, over 87% have now adopted VCF. momentum from strong VCF sales over the past 18 months.

Speaker Change: Sure.

Speaker Change: Now, let me talk about our infrastructure software segment.

Speaker Change: Q2 infrastructure software revenue of $6 6 billion was up 25% year on year above our outlook of $6 5 billion.

Speaker Change: As we've said before this.

Speaker Change: This growth reflects our success in converting our enterprise customers from <unk>.

Speaker Change: Petrol V sphere.

Speaker Change: Full vcs software subscription.

Speaker Change: Customers are increasingly turning to <unk> to create a mood.

Speaker Change: Modernized private cloud on Prem.

Speaker Change: Which will enable NIM.

Speaker Change: To repatriate workloads from public clubs.

Speaker Change: <unk> been able to run modern container based applications.

Speaker Change: AI applications.

Speaker Change: Of our 10000 largest customers over 87% have now adopted vcs.

Speaker Change: The momentum from strong DCF sales over the past 18 months since the acquisition of Vmware has created annual recurring revenue or.

Hock Tan: Since the acquisition of VMware, has created annual recurring revenue, or otherwise known as ARR, growth of double digits in our core infrastructure software.

Speaker Change: Otherwise known as our growth of double digits.

Speaker Change: Core infrastructure software.

Hock Tan: in Q3. We expect infrastructure software revenue to be approximately $6.7 billion, up 16% year-on-year. So in total, we're guiding Q3 consolidated revenue to be approximately $15.8 billion, up 21% year-on-year. We expect Q3 adjusted EBITDA to be at least 66%.

Speaker Change: In Q3.

Speaker Change: We expect infrastructure software revenue to be approximately $6 7 billion.

Speaker Change: Up 16% year on year.

Speaker Change: So in total we are guiding Q3 consolidated revenue to be approximately 15 $8 billion up 21% year on year.

Speaker Change: We expect Q3, adjusted EBITDA to be at least 66%.

Kirsten Spears: With that, let me turn the call over to Kirsten. Thank you, Hock. Let me now provide additional detail on our Q2 financial performance. Consolidated revenue was a record $15 billion for the quarter, up 20% from a year ago. Gross margins was 79.4% of revenue in the quarter, better than we originally guided on product Consolidated operating expenses were $2.1 billion, of which $1.5 billion was related to R&D. Q2 operating income of $9.8 billion was up 37% from a year ago with operating margin at 65% of revenue. Adjusted EBITDA was $10 billion or 67% of revenue, above our guidance of 66%.

Speaker Change: With that.

Kirsten Hawkins: Let me turn the call over to Kirsten.

Kirsten Hawkins: Thank you Hock, let me now provide additional detail on our Q2 financial performance.

Kirsten Hawkins: Consolidated revenue was a record 15 billion for the quarter up 20% from a year ago.

Kirsten Hawkins: Gross margin was 79, 4% of revenue in the quarter better than we originally guided on product mix.

Kirsten Hawkins: Consolidated operating expenses were $2 1 billion of which one 5 billion was related to R&D.

Kirsten Hawkins: Q2 operating income of $9 8 billion was up 37% from a year ago with operating margin at 65% of revenue.

Kirsten Hawkins: Adjusted EBITDA was $10 billion or 67% of revenue above our guidance of 66%. This figure excludes 142 million of depreciation.

Kirsten Spears: This figure excludes $142 million of depreciation.

Kirsten Spears: Now a review of the P&L for Archie's segment. Starting with semiconductors. Revenue for our Semiconductor Solutions segment was $8.4 billion, with growth accelerating to 17% year-on-year driven by AI. Semiconductor revenue represented 56% of total revenue in the quarter. Gross Margin for our Semiconductor Solution Segment was approximately 69%, up 140 basis points year-on-year, driven by product. Operating expenses increased 12% year-on-year to $971 million on increased investment in R&D for leading-edge AI semiconductors. Semiconductor Operating Margin of 57% was up 200 basis points year on year.

Kirsten Hawkins: Now a review of the P&L for our two segments.

Kirsten Hawkins: Starting with semiconductors.

Kirsten Hawkins: Revenue for our semiconductor solutions segment was $8 4 billion with growth accelerating to 17% year on year driven by AI.

Kirsten Hawkins: Semiconductor revenue represented 56% of total revenue in the quarter.

Kirsten Hawkins: Gross margin for our semiconductor solutions segment with approximately 69% up 140 basis points year on year driven by product mix.

Kirsten Hawkins: Operating expenses increased 12% year on year to 971 million on increased investment in R&D for leading edge AI semiconductors.

Kirsten Hawkins: Semiconductor operating margin of 57% was up 200 basis points year on year.

Kirsten Spears: Now moving on to infrastructure software. Revenue for infrastructure software of $6.6 billion was up 25% year-on-year and represented 44% of total revenue. gross margin for infrastructure software was 93% in the quarter compared to 88% a year ago. Operating expenses were $1.1 billion in the quarter, resulting in infrastructure software operating margin of approximately 76%. This compares to operating margin of 60% a year ago. This year-on-year improvement reflects our disciplined integration of VMware.

Kirsten Hawkins: Now moving on to infrastructure software.

Kirsten Hawkins: Revenue for infrastructure software of $6 6 billion was up 25% year on year and represented 44% of total revenue.

Kirsten Hawkins: Gross margin for infrastructure software with 93% in the quarter compared to 88% a year ago.

Kirsten Hawkins: Operating expenses were $1 1 billion in the quarter, resulting in infrastructure software operating margin of approximately 76%.

Kirsten Hawkins: This compares to operating margin of 60% a year ago. This year on year improvement reflects our disciplined integration of Vmware.

Kirsten Spears: Moving on to cash flow. Free cash flow in the quarter was $6.4 billion and represented 43% of revenue. Free Cash Flow as a Percentage of Revenue continues to be impacted by increased interest expense from debt related to the VMware acquisition and increased cash tax. We spent $144 million on capital. Day sales outstanding, we're 34 days in the second quarter compared to 40 days a year ago. We ended the second quarter with inventory of $2 billion, up 6% sequentially in anticipation of revenue growth in future quarters. Our days of inventory on hand were 69 days in Q2, as we continue to remain disciplined on how we manage inventory across We ended the second quarter with $9.5 billion of cash and $69.4 billion of gross principal debt.

Kirsten Hawkins: Moving onto cash flow.

Kirsten Hawkins: Free cash flow in the quarter was $6 4 billion and represented 43% of revenue.

Kirsten Hawkins: Free cash flow as a percentage of revenue continues to be impacted by increased interest expense from debt related to the Vmware acquisition and increased cash taxes.

Kirsten Hawkins: We spent 144 million on capital expenditures.

Kirsten Hawkins: Days sales outstanding were 34 days in the second quarter compared to 40 days a year ago.

Kirsten Hawkins: We ended the second quarter with inventory of $2 billion up 6% sequentially and anticipation of revenue growth in future quarters.

Kirsten Hawkins: Days of inventory on hand were 69 days in Q2 as we continue to remain disciplined on how we manage inventory across the ecosystem.

Kirsten Hawkins: We ended the second quarter with $9 5 billion of cash and $69 4 billion of gross principal debt.

Kirsten Spears: Subsequent to quarter end, we repaid $1.6 billion of debt, resulting in gross principal debt of $67.8 billion. The weighted average coupon rate and years to maturity of our $59.8 billion in fixed rate debt is 3.8% in seven years, respectively. The weighted average interest rate and years to maturity of our $8 billion in floating rate debt is 5.3% and 2.6 years respectively.

Kirsten Hawkins: Subsequent to quarter end, we repaid $1 6 billion of debt, resulting in gross principal debt of $67 8 billion.

Kirsten Hawkins: The weighted average coupon rate in years to maturity of our $59 8 billion in fixed rate debt is three 8% and seven years respectively.

Kirsten Hawkins: The weighted average interest rate in years to maturity of our 8 billion in floating rate debt is five 3% and two six years respectively.

Kirsten Hawkins: Okay.

Kirsten Spears: Turning to capital allocations. Q2 we paid stockholders $2.8 billion of cash dividends based on a quarterly common stock cash dividend of $0.59 per share. In Q2, we repurchased $4.2 billion, or approximately 25 million shares of common stock. In Q3, we expect the non-GAAP diluted share count to be approximately 4.97 billion shares, excluding the potential impact of any share repurchase.

Kirsten Hawkins: Turning to capital allocation.

Kirsten Hawkins: In Q2, we paid stockholders $2 8 billion of cash dividends based on our quarterly common stock cash dividend of 59 per share.

Kirsten Hawkins: In Q2, we repurchased $4 2 billion or approximately 25 million shares of common stock in Q3, we expect non-GAAP diluted share count to be approximately $4 97 billion shares excluding the potential impact of any share repurchases.

Kirsten Spears: Now moving on to Guide. Our guidance for Q3 is for consolidated revenue of $15.8 billion, up 21% year-on-year. We forecast semiconductor revenue of approximately $9.1 billion of 25%. Within this, we expect Q3 AI semiconductor revenue of $5.1 billion, up 60% year-on-year. We expect infrastructure software revenue of approximately $6.7 billion, up 16% year-on-year. For modeling purposes, we expect Q3 consolidated gross margin to be down approximately 130 basis points sequentially, primarily reflecting a higher mix of XPUs within AI revenue. As a reminder, consolidated gross margins through the year will be impacted by the revenue mix of infrastructure, software, and semi-contractors.

Kirsten Hawkins: Now moving onto guidance.

Kirsten Hawkins: Our guidance for Q3 is for consolidated revenue of $15 8 billion up 21% year on year, we forecast semiconductor revenue of approximately $9 1 billion up 25% year on year.

Kirsten Hawkins: Within this we expect Q3 AI semiconductor revenue of $5 1 billion up 60% year on year.

Kirsten Hawkins: We expect infrastructure software revenue of approximately $6 7 billion up 16% year on year.

Kirsten Hawkins: For modeling purposes, we expect Q3 consolidated gross margin to be down approximately 130 basis points sequentially, primarily reflecting a higher mix of XP use within AI revenue.

Kirsten Hawkins: As a reminder, our consolidated gross margin through the year will be impacted by the revenue mix of infrastructure software and semiconductors.

Kirsten Spears: We expect Q3 adjusted EBITDA to be at least 66%. We expect the non-GAAP tax rate for Q3 in fiscal year 2025 to remain at 14%.

Kirsten Hawkins: We expect Q3, adjusted EBITDA to be at least 66%.

Kirsten Hawkins: We expect the non-GAAP tax rate for Q3 and fiscal year 2025 to remain at 14%.

Kirsten Spears: And with this, that concludes my prepared remarks. Operator, please open up the call for questions. Thank you. To ask a question, you will need to press star 11 on your telephone. To withdraw your question, please press star 11 again. Due to time restraints, we ask that you please limit yourself to one question. Please stand by while we compile the Q&A roster.

Kirsten Hawkins: And with that that concludes my prepared remarks, operator, please open up the call for questions.

Speaker Change: Thank you to ask a question you will need to press star one on your telephone to withdraw your question. Please press star one again due to time restraints. We ask that you. Please limit yourself to one question. Please standby, while we compile the Q&A roster.

Ross Seymour: And our first question will come from the line of Ross Seymour with Deutsche Bank. Your line is open. Hi, guys. Thanks for letting me ask a question. Hock, I wanted to jump onto the AI side and specifically some of the commentary you had about next year. Can you just give a little bit more color on the inference commentary you gave? And is it more the XPU side, the connectivity side, or both that's giving you the confidence to talk about the growth rate that you have this year being matched next fiscal Thank you, Ross. Good question.

Speaker Change: And our first question will come from the line of Ross Seymore with Deutsche Bank. Your line is open.

Ross Seymore: Guys. Thanks for let me ask a question Hock I wanted to jump into the AI side and specifically some of the commentary you had about next year can you just give a little bit more color on the inference commentary you gave and then is it more of the SBU side, the connectivity side or both that's giving you the confidence to talk about the growth rate that you have this year being matched next fiscal year.

Speaker Change: Thank you Ron good question.

Hock Tan: I think we're indicating that what we are seeing and what we have quite a bit of visibility, increasingly, is increased Deployment of XPUs next year. and much more than we originally thought. And hand in hand with it, of course, more and more networking. So it's a combination of both. and the imprint side of things. Yeah, we're seeing much more inference now. Thank you.

Speaker Change: I think we're indicating that what we are seeing and what we have quite a bit of visibility increasingly.

Speaker Change: This increase deployment of XP use next year.

Speaker Change: And much more than we originally thought and hand in hand with it of course.

Speaker Change: More and more networking so it's a combination of both.

Speaker Change: And the print side of things.

Speaker Change: Yes, we are seeing much more for himself.

Speaker Change: Thank you.

Speaker Change: Okay.

Harlan Sur: One moment for our next question. And that will come from the line of Harlan Sur with J.P.

Speaker Change: Thank you one moment for our next question.

Speaker Change: And that will come from the line of Harlan sur with Jpmorgan. Your line is open.

Harlan Sur: Morgan. Your line is open. Good afternoon. Thanks for taking my question. And great job on the quarterly execution. Hock, you know, good to see the positive growth in suction quarter quarter, year over year growth rates in your AI business. You have the team, as mentioned, right, the quarters can be a bit lumpy. So if I smooth out kind of first three quarters of this fiscal year, your AI business is up 60% year over year. It's kind of right in line with your three year kind of SAM growth CAGR, right? So given your prepared remarks, and, you know, knowing that your lead times remain at 35 weeks or better, do you see the Broadcom team sustaining the 60% year over year growth rate exiting this year?

Harlan Sur: Hey, good afternoon. Thanks for taking my question and great job on the quarterly execution.

Speaker Change: Good to see the positive growth inflection quarter over quarter year over year growth rates in your AI business do you have the team.

Speaker Change: <unk> mentioned the quarters can be a bit lumpy.

Speaker Change: <unk> kind of first three quarters of this fiscal year.

Speaker Change: AI business is up 60% year here, it's kind of right in line with your three year kind of Sam growth CAGR right given your prepared remarks.

Speaker Change: Knowing that your lead times remain at 35 weeks are better do you see the broadcom team sustaining the 60% year over year growth rate exiting this year and I assume that yes potentially.

Harlan Sur: And I assume that that potentially implies that you see your AI business sustaining the 60% year over year growth rate into fiscal 26, again, based on your prepared commentary, which again is in line with your SAM growth CAGR. Is that kind of a fair way to think about the trajectory this year and next year? Harlan, that's a very insightful set of analysis here. And that's exactly what we're trying to do here. Because six over six months ago, we gave you guys a point a year 2027. As we come into the second now into the second half of 2025.

Speaker Change: Potentially implies that you see your AI business sustaining 60% year over year growth rate into fiscal 'twenty six again based on your prepared commentary, which again is in line with your Sam growth take or is that kind of a fair way to think about the trajectory this year and next year.

Holland: Holland, that's a very insightful set of.

Holland: Analysis here and that's exactly what we're trying to do here because six over six months ago. We gave you guys a point a year 2027.

Holland: As we've come into the second into the second half of 2025, and we've improved visibility and updates we are seeing in the way our Hyperscale partners are deploying data centers AI clusters.

Hock Tan: And we've improved visibility and updates we're seeing in the way our hyperscale partners are deploying data centers, AI class We are providing you some level of guidance, visibility, what we are seeing, how the trajectory of 26 might look. I'm not giving you any update on 27. We're just still establishing the update we have in 27, six months ago. But what we're doing now is giving you more visibility into where we're seeing 26 headed. But is the framework that you laid out for us, like second half of last year, which implies 60% kind of growth kager in your SAM opportunity, is that kind of the right way to think about it as it relates to the profile of growth in your business this year and next year?

Speaker Change: We are providing you more some level of talent.

Speaker Change: Our guidance visibility, what we are seeing how the trajectory of 26 might look like.

Speaker Change: I'm not giving you an update on 2007, we just still establishing the update we have been 27 six months ago, but what we're doing now is giving you more visibility.

Speaker Change: Into where we are seeing 2006 head.

Speaker Change: But as the framework that you laid out for us like second half of last year, which implies 60% kind of growth CAGR on your Sam opportunity or is that kind of the right way to think about it as it relates to the profile of growth in your business. This year and next year, yes.

Harlan Sur: Yes. Okay, thank you, Hock. Thank you.

Speaker Change: Okay. Thank you hock.

Ben Reitzes: One moment for our next question. And that will come from the line of Ben Reitzes with Mellius Research.

Speaker Change: Thank you one moment for our next question.

Speaker Change: And that will come from the line of Ben Reitzes with Melius Research. Your line is open.

Hock Tan: Your line is open. Hey, how you doing? Thanks, guys. Hey, Hock, networking, AI networking was really strong in the quarter. And it's, it seemed like it must have beat expectations. I was wondering if you could just talk about the networking in particular, what what caused that? And how much of that is your acceleration into next year? And when do you think you see Tomahawk kicking in as part of that acceleration? Thanks. Well, I think the network AI networking, as you probably would know, goes pretty hand in hand with deployment of AI accelerator clusters. It doesn't deploy on a timetable that's very different from the way that...

Ben Reitzes: Hey, How're you doing thanks, guys, Hey, Hock.

Ben Reitzes: Networking AI networking was really strong in the quarter.

Speaker Change: And it seemed like it must have beat expectations. I was wondering if you could just talk about the networking in particular, what what caused that and how much of that is your acceleration into next year.

Speaker Change: And when do you think you see tomahawk kicking in as part of that acceleration. Thanks.

Speaker Change: Well I think the.

Speaker Change: Network <unk>.

Speaker Change: Looking.

Speaker Change: As you probably would know.

Speaker Change: It goes hand in hand with deployment of AI accelerator clusters. It is it doesn't deploy on the timetable that is very different from the way the.

Hock Tan: Accelerators get deployed, where there they are. Spews, or G.P. It does happen, and they deploy a lot in scale out. Way, Ethernet, of course. It's not just a choice or protocol, but it's also increasingly moving into the space of what we all call scale up within those data centers, where you have much higher, more than we originally thought, consumption or density of switches than you have in the scale out scenario. In fact, the increased density in scale up is... 5 to 10 times more than in scale-out. And that's the part that kind of pleasantly surprises, and which is why in this past quarter Q2...

Speaker Change: Salaries does get deployed.

Speaker Change: Excuse or Gpus, it does happen and they deploy a lot in scale out way Ethernet of course is the.

Speaker Change: Joyce our protocol, but it's also increasingly moving into the space of what we all call scale up within those data centers, where you have much higher more than we originally thought couldnt assumption or density of switches.

Speaker Change: Then you have in the scale out scenario.

Speaker Change: Increased density and scale up.

Speaker Change: Yes.

Speaker Change: Five to 10 times more than in scale out and that's the part that kind of pleasantly surprised us and which is why in this past quarter Q2.

Hock Tan: The AI networking portion continues at about 40% from when we reported a quarter ago for Q1. And at that time I said I expect it to drop. It hasn't.

Speaker Change: The AI networking portion continues at about 40% from when we reported a quarter ago for Q1.

Speaker Change: And at that time, I said I expect it to drop.

Speaker Change: It hasn't.

Hock Tan: and and your thoughts on Tomahawk driving acceleration for next year and when it kicks in. Oh, Tomahawk 6. Oh, yeah, that's extremely strong interest. Now, we're not shipping big orders or any orders other than basic proof of concepts out to customers. But there is tremendous demand for this new 102 terabit per second Tomahawk Thanks, Hock.

Speaker Change: And and your thoughts on Tomahawk driving acceleration for next year and when it kicks in.

Speaker Change: Tom I'll say, Oh, yes, that's extremely strong interest.

Speaker Change: No.

Speaker Change: We're not shipping.

Speaker Change: Big orders or any orders or other than basic proof of concepts out to customers, but there is tremendous demand for this new 102, Tera terabits per second.

Speaker Change: Tomahawk switches.

Speaker Change: Thanks Hock.

Hock Tan: Thank you.

Blayne Curtis: One moment for our next question. And that will come from the line of Blayne Curtis with Jeffries.

Speaker Change: Thank you one moment our next question.

Blayne Curtis: And that will come from the line of Blayne Curtis with Jefferies. Your line is open.

Hock Tan: Your line is open. Hey, thanks and great results. I just want to ask maybe following up on the scale out opportunity. So today, I guess your main customer is not really using an NVLink switch style scale up. I'm just kind of curious, your visibility or the timing in terms of when you might be shipping, you know, a switch Ethernet scale up network to your customer.

Blayne Curtis: Hey, Thanks, and great results I just wanted to ask maybe following up on the scale out opportunity. So today I guess your main customers not really using kind of an empty links with style scale up I'm just kind of curious.

Blayne Curtis: Your visibility or the timing in terms of when you might be shipping a switch Ethernet scale up network to your customers.

Hock Tan: THANKS FOR WATCHING PLEASE SUBSCRIBE Scale. Scale up. Yeah, well, scale up. very rapidly converting to Ethernet. Very much so. I for our fairly narrow band of hyperscale customers. Gale Ar… is very much Ethernet. Thank you.

Blayne Curtis: The bookings scale up scale scale up.

Blayne Curtis: Well scale up.

Blayne Curtis: Very rapidly converting to Ethernet.

Blayne Curtis: Very much so.

Blayne Curtis: For our fairly narrow band of Hyperscale customers.

Blayne Curtis: Scale up.

Blayne Curtis: It's very much Ethernet.

Stacy Rasgon: One moment for our next question.

Blayne Curtis: Thank you one moment for our next question.

Blayne Curtis: Yeah.

Stacy Rasgon: And that will come from the line of Stacy Rasgon with Bernstein. Your line is open. Hi, guys. Thanks for taking my questions. Um, Hock, I still wanted to follow up on that AI 2026 question. I want to just put some numbers on it, just just to make sure I've got it right. So if you you did 60% in the first three quarters of this year, if you grow 60% year over year in Q4 and put you at like, I don't know, 5.8 billion, something like 19 or 20 billion for the year. And then are you saying you're going to grow 60% in 2026, which would put you 30 billion plus in AI revenues for 2026?

Speaker Change: And that will come from the line of Stacy <unk> with Bernstein. Your line is open.

Speaker Change: Hi, guys. Thanks for taking my questions.

Speaker Change: I wanted to follow up on that AI 2026 question I wanted to just put some numbers on it.

Speaker Change: Just to make sure I've got it right. So if you did 60% in the first three quarters of this year, if you grow 60% year over year in Q4. It puts you at like I don't know five 8 billion sounds like 19 or $20 billion for the year.

Speaker Change: And then are you are you, saying youre going to grow 60% in.

Speaker Change: In 2026 would put you 30 billion clots in AI revenues for 2012, because I just want to make is that the math that you're trying to communicate to us directly.

Hock Tan: I just want to know, is that the math that you're trying to communicate to us directly? I think you're doing the math. I'm giving you the trend. But I did answer that question, I think Harlan asked earlier. The rate we are seeing now so far in fiscal 25... And we'll presumably continue. We don't see any reason why it doesn't give a lead time visibility in 25. What we're seeing today, based on what we have visibility on 26th, is to be able to Ram up this AI revenue in the same trajectory. So is the SAM going up?

Speaker Change: Doing the math I'm, giving you the trend.

Speaker Change: And but I did I answer that question I think Harlan asked earlier.

Speaker Change: The rate we are seeing that.

Speaker Change: Now so far in fiscal 'twenty five.

Speaker Change: And presumably continue we don't see any reason why it doesn't given lead time visibility in 'twenty five.

Speaker Change: What we are seeing today based on what we have visibility.

Speaker Change: On 26 is to be able to.

Blayne Curtis: Ram.

Blayne Curtis: This AI revenue in the same trajectory.

Blayne Curtis: Yes.

Hock Tan: So is the SAM going up as well? Because now you have inference on top of training. So is the SAM still 60 to 90? Or is the SAM higher now, as you see it? I'm not playing a SAM game here. I'm just giving you a trajectory towards where we drew the line on 27 before. So I have no response to is the SAM going up or not. Stop talking about SAM now. Thanks. Oh, okay. Thank you.

Blayne Curtis: Sam going up so is this game going up as well because now you have influence on top of training. So as the Sam Sam So 60 to 90 or is the same higher now.

Blayne Curtis: I'm not playing the same game here I'm, just giving you a trajectory towards where we draw the line on 2007 before so I have no. No response, though is the same going up but not stopped talking about the same now.

Blayne Curtis: Thanks.

Speaker Change: Okay. Thank you.

Vivek Arya: One moment for our next question and that will come from the line of Vivek Arya with Bank of America. Your line is open. Thanks for taking my question. I had a near and then a longer-term question on the XTU business. So, Hock, for near-term, if you're networking upside in Q2 and overall AI was in line, it means XTU was perhaps not as strong. So, I realize it's lumpy, but anything more to read into that, any product transition or anything else? So, just a clarification there.

Blayne Curtis: One moment for our next question.

Speaker Change: And that will come from the line of Vivek Arya with Bank of America. Your line is open.

Vivek Arya: Thanks for taking my question I had a near and longer term question on the <unk> business. The Hawk for near term, if youre networking upside in Q2 and overall.

Vivek Arya: Was in line. It means <unk> was perhaps not as strong I realize it's lumpy, but anything more to read into that any product transition or anything else.

Hock Tan: And then longer term, you know, you have outlined a number of additional customers that you're working with. What milestones should we look forward to and what milestones are you watching to give you the confidence that you can now start adding their addressable opportunity into your 27 or 28 or other numbers? Like, how do we get the confidence that these projects are going to turn into revenue in some, you know, reasonable timeframe from now? Thank you. Okay, on the first part that you are asking. You know, it's like, you're trying to be, you're trying to count how many angels on the head of a pin here.

Vivek Arya: Clarification there.

Vivek Arya: And then longer term you have outlined a number of additional customers that youre working with what milestones should we look forward to and what milestones are you watching to give you. The confidence that you can now start adding that addressable opportunity into your 2007 or 28 or other numbers like how do we get the.

Vivek Arya: <unk> that these projects are going to turn into revenue in some reasonable timeframe from now thank you.

Blayne Curtis: Okay on the first spun that you're asking.

Blayne Curtis: Slide <unk>.

Blayne Curtis: Going to be Youre trying to count how many angels on the head of opinion.

Hock Tan: I mean, whether it's XPU or networking. Networking is hard, but that doesn't mean XPU is any softer. This is very much along the trajectory we expected. There's no lumpiness. There's no softening. It's pretty much what we expect. for on PlanBusiness.

Speaker Change: Whether it's XP.

Speaker Change: Our networking networking is hot but that doesn't mean excuse any source. So it's very much along the trajectory we expected to be in.

Vivek Arya: And there's no lumpiness there still softening, it's pretty much what we expect.

Vivek Arya: The trajectory to go so far.

Vivek Arya: And into next quarter, Israel, and probably beyond so we have it's a.

Vivek Arya: Fairly.

Vivek Arya: So I guess in our view.

Vivek Arya: Fairly clear visibility on the short term trajectory.

Hock Tan: In terms of going on to 27, no, we are not updating any numbers here. We six months ago, we drew a sense for the size of the SAM based on... You know, a million. We will be happy to give an update to the audience. But right now though, in today's prepared remarks and answering a couple of questions We have, we are, as we are doing, as we have done yet, we are intending to give you guys more visibility what we've seen. Growth Trajectory in 26.

Vivek Arya: In terms of going on to 'twenty seven no. We are not updating any numbers here, we six months ago, we draw our same store the size of the same base on.

Vivek Arya: You know me.

Vivek Arya: <unk> be a million.

Vivek Arya: GPU XP clusters for three customers and Thats still very well at that point that youll be done and we have not provided any further updates year, nor do we intend them to at this point when we can have better visibility a clearer sense of where we are and that probably wont happen.

Vivek Arya: 26.

Vivek Arya: We'll be happy to give you an update to the audience, but right now though.

Vivek Arya: Today's prepared remarks, and answering a couple of questions. We have we are as we are doing as we have done yet.

Vivek Arya: Intending to give you guys more visibility what we are seeing.

Vivek Arya: The growth trajectory in 'twenty six.

Vivek Arya: Okay.

Hock Tan: Thank you.

CJ Muse: One moment for our next question.

Vivek Arya: Thank you one moment for our next question.

Hock Tan: And that will come from the line of CJ Muse with Cantor Fitzgerald. Your line is open. Yeah, good afternoon. Thank you for taking the question. I was hoping to follow up on Ross's question regarding inference opportunity. Can you discuss workloads that are optimal that you're seeing for custom silicon? And that over time, what percentage of your XPU business could be inference versus training? Thank you. I think there's no differentiation between training and inference in using merchants. Accelerators vs. Custom Accelerators, I think the whole premise behind Going Towards Custom Accelerators Continues, which is... Not a matter of cost alone, it is that as custom accelerators get used...

Vivek Arya: And that will come from the line of C. J Muse with Cantor Fitzgerald. Your line is open.

Speaker Change: Yes. Good afternoon. Thank you for taking the question I was hoping to follow up on Ross's question regarding in for its opportunity can you discuss more closer to optimal that youre seeing for custom silicon and then over time, what percentage of your <unk> business could be influence versus training. Thank you.

Vivek Arya: I think there's no differentiation between training and inference and using.

Vivek Arya: Merchant.

Vivek Arya: Accelerators versus customer accelerators, I think they're all under the.

Vivek Arya: The whole premise.

Vivek Arya: Behind.

Vivek Arya: Going towards <unk>.

Vivek Arya: Awesome accelerators.

Vivek Arya: <unk> continues which is.

Vivek Arya: It's not a metal costs alone.

Vivek Arya: Is that S custom accelerators get us and develop on a per on a roadmap with any particular hyper.

Hock Tan: and get developed on a roadmap with any particular... hyposcaler There's a learning curve, a learning curve on how they could optimize the way the algorithms on their large language models get... Written by Christopher Rakesh, Aaron Rakesh, Joshua Buchalter, Kirsten Spears, Benjamin Reitzes, Charlie Kawwas, William Stein and Typed to select. And that ability to do so is a huge value added. creating algorithms that can drive their LLMs to higher and higher performance. much more than basically a segregation approach between hardware. is that you've literally combined end-to-end hardware and software as they take that journey. And it's a journey.

Vivek Arya: Hyperscale.

Vivek Arya: There is a learning curve and learning curve on how they could optimize.

Vivek Arya: The way deal as the algorithms on their large language models gets.

Vivek Arya: Britain.

Vivek Arya: Hi.

Vivek Arya: <unk>.

Vivek Arya: Silicon.

Vivek Arya: And then the ability to do so it's a huge value add in creating.

Vivek Arya: Algorithms that can drive the LMS to higher and higher performance.

Vivek Arya: More than.

Vivek Arya: Basically.

Vivek Arya: Segregation approach between hardware.

Vivek Arya: And the software is that youll be literally combined end to end hardware and software.

Vivek Arya: Take that journey and it's a journey they don't run that in one year doing a few cycles get better and better at it.

Hock Tan: They don't learn that in one year. Do it a few cycles, get better and better at it, and then realize their value, the fundamental value. Creating Your Own Hardware versus using a third-party merchant. that you are able to optimize. way high performance than you otherwise. and we see that happening.

Speaker Change: Dan lies.

Vivek Arya: Failure, the fundamental value in creating your own hardware.

Vivek Arya: Versus using third party merchant silicon that UI able to optimize.

Vivek Arya: Youll software to the hardware and event and eventually achieve.

Vivek Arya: Weihai performance than you otherwise could.

Vivek Arya: And we see that happening.

Vivek Arya: Okay.

Hock Tan: Thank you.

Vivek Arya: Okay.

Operator: One moment for our next question.

Vivek Arya: Thank you one moment our next question.

Karl Ackerman: And that will come from the line of Karl Ackerman with BNP Paribas. Your line is open. Thank you. Hock, you spoke about the much higher content opportunity in scale-up networking. I was hoping you could discuss how important is demand adoption for co-package optics in achieving this 510X higher content for scale-up networks? Or should we anticipate much of the scale-up opportunity will be driven by Tom Hock and Thor Nix? Thank you. I'm trying to decipher this question of yours, so let me try to answer it perhaps in a way I think you want me to clarify.

Speaker Change: And that will come from the line of Karl Ackerman with BNP Paribas. Your line is open.

Vivek Arya: Yes. Thank you.

Speaker Change: Hockey spoke about.

Vivek Arya: <unk>.

Vivek Arya: The much higher content opportunity and scale up networking I was hoping you could discuss how important is demand adoption for <unk>.

Vivek Arya: Co package optics.

Vivek Arya: <unk> it.

Vivek Arya: Achieving this fiber connects higher content for scarp networks.

Vivek Arya: Or should we anticipate much of a scale up opportunity will be driven by.

Vivek Arya: Tomahawk and toward next thank you.

Vivek Arya: I mean I'm trying to decipher. This question of yours. So let me try to answer it perhaps in a way I think you wanted me to clarify.

Hock Tan: First and foremost, I think most of what's scaling up, there are a lot of scaling up that's going on, as I call it, which means a lot of... is done on copper, copper interconnection. And because, you know, that's the size of the size of this of this Scale Art Cluster, still not that huge yet, they can get away with it. And they're still doing it, mostly they're doing it. at some point soon. I believe when you start trying to go beyond maybe seven... Interconnect. You may have to push towards a different protocol by protocol mode at a different medium from copper.

Vivek Arya: First and foremost I think most of what's scaling up a lot.

Vivek Arya: Sort of the scaling up that go in as I call, it which means a lot of.

Vivek Arya: HBU or GPU to GPU interconnects.

Vivek Arya: It's done on copper copper interconnect.

Vivek Arya: And because the.

Vivek Arya: The size of the size of this.

Vivek Arya: Of this.

Vivek Arya: Scale cluster still not that huge it they can get away with it.

Vivek Arya: Copper using copper interconnect.

Vivek Arya: And they're still doing it mostly theyre doing it today.

Vivek Arya: At some point soon.

Vivek Arya: I believe when you start trying to go beyond maybe 72 GPU to GPU.

Vivek Arya: <unk> you may have to push towards a different protocol by vertical model different media.

Speaker Change: Copper too.

Vivek Arya: To optical.

Hock Tan: And when we do that, yeah, perhaps then things like exotic stuff like gold packaging of silicon with optical might become relevant. But truly what we really are talking about is that at some stage as the clusters get larger, which means... Scale-up becomes much bigger and you need to interconnect. And that's when optical will start replacing copper. And when that happens, the question is, what's the best way to do it? and Deliver on Optical. One way is called package op. but it's not the only way. You can just simply use, continue to use, perhaps plug a...

Vivek Arya: And when we do that yeah, perhaps then things site.

Vivek Arya: No.

Vivek Arya: <unk> stopped like co packaging might be can fall off of silicon with optical might become relevant but truly what we really are talking about is that at some stage as the clusters get larger.

Vivek Arya: Which means scale up becomes much bigger and you need to interconnect.

Vivek Arya: GPU or XP to each other and scale up.

Vivek Arya: Many more than just 70 to a 100, maybe even 128 you start growing more and more you want to use our optical interconnect simply because of distance and thats when <unk>.

Vivek Arya: Optical will start replacing copper and when that happens. Our question is what's the best way to.

Vivek Arya: Deliver on optical one way is co package optics.

Vivek Arya: But it's not the only way.

Vivek Arya: You can't just simply use continue use perhaps plausible.

Hock Tan: at low cost. in which case then you can interconnect. The bandwidth, the radix of a switch, and our switch is down 512. Connections. So you can now connect all these XPUs. Skills, Aditya Kumar, Yonatan plebe, Super Chief, Akram Lajang, Dhyar Howai, Samudra Hello. We're having a very productive plenums program from 2012 to 2017 that we do with the U.S. Embassy in India & India, 26 countries in total. We going to start with India from 2012 to 2017. Deoko Unchure is one of the trading card makers combined with DPA, Bank of India, U.S. International trade bank and the 3.6 billion USD new token.

Vivek Arya: Low cost optics and risky venue can interconnect.

Vivek Arya: The bandwidth.

Vivek Arya: <unk> of switch and our switches down 512.

Vivek Arya: Connections. So you can now connect.

Vivek Arya: This XP use Gpus 512 for scale up phenomenon and that was huge.

Vivek Arya: It's when you go to optical that's going to happen within my view a year or two.

Vivek Arya: And we'll be right in the forefront of it and it maybe co package optics, which we are very much in development, but is a locking co package or could just be as a first step plug a bowl of whatever it is I think the bigger question is when does it go from optical.

Hock Tan: from Copper. The next thing, GPU to GPU to Optane. Connecting it and the stamp in that move will be And it's not necessary for package updates, so that's definitely one path we are taking. Very clear. Thank you.

Vivek Arya: From copper.

Vivek Arya: Connecting GPS attribute to optical commencing and the staff.

Vivek Arya: <unk> in that move will be huge.

Vivek Arya: And it's not necessary co package of base load that definitely one pad we are pursuing.

Vivek Arya: Very clear thank you.

Operator: And one moment for our next question.

Vivek Arya: And one moment our next question.

Hock Tan: And that will come from the line of Joshua Buchalter with TD Cowen. Your line is open. Hey, guys, thank you for taking my question. I realize it's a bit nitpicky, but I wanted to ask about gross margins in the guide. So your revenue implies sort of $800 million incremental increase, but gross profit up, I think, $400 to $450 million, which is kind of pretty well below corporate average fall through. Appreciate that semis is dilutive and custom is probably dilutive within semis, but anything else going on with margins that we should be aware of? And how should we think about the margin profile of custom longer term as that business continues to scale and diversify?

Speaker Change: And that will come from the line of Joshua Buchalter with TD Cowen Your line is open.

Joshua Buchalter: Hey, guys. Thank you for taking my question.

Joshua Buchalter: Realize it's a bit nitpicky, but I wanted to ask about gross margins in the guide. So your revenue implies sort of eight and a $100 million incremental increase with gross profit up I think $400 million to $450 million, which is kind of pretty well below corporate average fall through.

Joshua Buchalter: I appreciate that <unk> is dilutive and customers probably dilutive within semi but anything else going on with margins that we should be aware of.

Joshua Buchalter: And how should we think about the margin profile of custom longer term as that business continues to scale and diversify.

Hock Tan: Thank you. Yeah, we've historically said that the XPU margins are slightly lower than the rest of the business, other than wireless, and so there's really nothing else going on other than that. Exactly what I said, that the majority of it, quarter over quarter, is the 130 basis point decline is being driven by more... You know, there are more moving... Parts here, then your simple analysis proves here. And I think your simple analysis is totally wrong in that regard. All right, and thank you.

Joshua Buchalter: Yes, we've historically said that the SBU margins are slightly lower.

Joshua Buchalter: Then the rest of the business other than wireless and so there's really nothing else going on other than that it's just exactly what I said that the majority of it quarter over quarter at 130 basis point decline is being driven by more extra years.

Joshua Buchalter: They're moving.

Joshua Buchalter: But then your simple analysis process and I think the simple analysis totally wrong in that regard.

Joshua Buchalter: Yeah.

Joshua Buchalter: Alright, thank you.

Timothy Arcuri: And one moment for our next question. And that will come from the line of Timothy Arcuri with UBS. Your line is open. Thanks a lot. I also wanted to ask about scale up, Hock. So there's a lot of competing ecosystems. There's UA Link, which, of course, you left. And now there's the big GPU company opening up, NVLink. And they're both trying to build ecosystems. And there's an argument that you're an ecosystem of one. What would you say to that debate? Does opening up NVLink change the landscape? And how do you view of your AI network and growth next year?

Joshua Buchalter: And one moment for our next question.

Speaker Change: And that will come from the line of Timothy Arcuri with UBS. Your line is now open.

Timothy Arcuri: Thanks, a lot.

Speaker Change: I also wanted to ask you what scale up hockey so theres a lot of competing ecosystems. As you a language course, you left and that is the big GPU companies opening up Andy Lake and they are both trying to build eco systems and there is an argument that youre an ecosystem of one.

Joshua Buchalter: What would you say that that debate does does opening up NZ link change the landscape and sort of how do you view of your AI networking growth next year do you think is going to be primarily driven by scale up or what's still be pretty scale out heavy thanks.

Hock Tan: Do you think it's going to be primarily driven by scale up or would still be pretty scale out heavy things? It's, you know, people do like to create platforms and new protocols and systems. The fact of the matter is, scale up. can just be done easily and it's currently available. Open Standards, Open Source, Ethernet, just as well, just as well. You don't need to create new systems for the sake of... doing something that you could easily be doing in networking in Ethernet. And so, yeah, I hear a lot of this interesting new protocols, standards that are trying to be created.

Joshua Buchalter: People do like to create platforms and new protocols and system. The fact of the matter is.

Joshua Buchalter: Scale.

Joshua Buchalter: Can just be done easily and.

Joshua Buchalter: Currently available.

Joshua Buchalter: <unk>.

Joshua Buchalter: Open standards open source Ethernet, just as well just as well you don't need to create new systems for the sake of.

Joshua Buchalter: Doing something that you could easily be doing in networking in Ethernet.

Joshua Buchalter: And so yeah I hear a lot of this interesting new protocol standards that are trying to be creative and most of them by the way a proprietary much as I like to call. It otherwise one Israeli open source.

Hock Tan: And most of them, by the way, are proprietary, much as they like to call it otherwise. What is really open source and open standard is Ethernet. And we believe Ethernet will prevail as it does for the last 20 years in traditional networking. There's no reason to create a new standard for something that could be easily done in transferring. and Bites of Data. Got it, Hock. Thank you.

Joshua Buchalter: And the open standards ASC Ethernet and we believe Ethernet will prevail as it does.

Joshua Buchalter: And before the last 20 years in traditional networking, there's no reason to create a new standard for something that could be easily done in transferring bits and bytes of data.

Joshua Buchalter: Got it thank you.

Operator: And one moment for our next question.

Speaker Change: And one moment our next question.

Christopher Rolland: And that will come from the line of Christopher Rolland with Susquehanna. Your line is open. Thanks for the question. Yeah, my question is for you, Hock. It's a kind of a bigger picture one here. And this kind of acceleration that we're seeing in AI demand, do you think that this acceleration is because of a marked improvement in ASICs or XPUs closing the gap on the software side at your customers? Do you think it's these require tokenomics around inference, test time compute driving that, for example? What do you think is actually driving the upside here? And do you think it leads to a market share shift faster than we were expecting towards XPU from GPU?

Speaker Change: And that will come from the line of Christopher Rolland with Susquehanna. Your line is open.

Speaker Change: Thanks for the question.

Christopher Rolland: Yes. My question is for you Hock, it's kind of a bigger picture one here.

Joshua Buchalter: This kind of acceleration that we're seeing in AI demand.

Joshua Buchalter: Do you think that this acceleration is because of a marked improvement in 86 Rx P. R. <unk> closing the gap on the software side.

Joshua Buchalter: At your customers do you think it's.

Joshua Buchalter: These require token.

Joshua Buchalter: <unk> around infringe.

Joshua Buchalter: Test time compute driving that for example, what do you think is actually driving the upside here and do you think it leads to a market share shift faster than we were expecting towards XP from GPU. Thanks.

Hock Tan: Thanks. Yeah, interesting question. But no, none of the foregoing that you outlined. It's very simple. The way inference has come out very, very hot lately is, remember, we're only selling to a few. customers, hyperscalers with platforms and LLMs. That's it. There are not that many. And we told you how many we have, and we haven't increased any. But what is happening is... These hyperscalers and those with LNN... Justify all the spending they're doing. Doing training makes your frontier model smarter. That's my question. It's almost like science, research and science. Make your frontier models by creating very clever algorithms that consume a lot of compute for training.

Joshua Buchalter: Yes interesting question about no none of the forgoing that you outline sort of a simple the way inference that's come out.

Joshua Buchalter: Very very hot lately.

Joshua Buchalter: No remember, we're only selling to a few.

Joshua Buchalter: Customers Hyperscale is voice platforms and LMS. That's it there are not that many.

Joshua Buchalter: And we had told you how many we have and we have an increasingly but what is happening is.

Joshua Buchalter: This all on this hyperscale us and dose with <unk>.

Joshua Buchalter: Me too.

Joshua Buchalter: Justify all this spending theyre doing doing training.

Joshua Buchalter: Thanks, you all frontier models smaller that's one question is almost life Science research and science Mega Frontier models by creating very clever algorithms that deep.

Joshua Buchalter: Deep consumes a lot of compute for training Smart training makes us model.

Hock Tan: Training makes us smarter. You want to monetize? And that's what's driving it. Money, as I indicated in my prepared remarks, the drive to justify... a return on invest and a lot of their investment is training. And then return on investment is by creating use cases, a lot of AI use cases, AI. Consumption. out there through availability of a lot of And that's what we are now starting. among a small group of customers. Excellent. Thank you.

Speaker Change: You want to monetize.

Joshua Buchalter: Inference.

Joshua Buchalter: And that's what's driving it monetize I indicated in my prepared remarks, the drive to justify.

Joshua Buchalter: Our return on investment and a lot of the investment is training and that return on investment is by creating use cases, a lot AI use cases.

Joshua Buchalter: I.

Joshua Buchalter: Consumption.

Joshua Buchalter: Through availability awful lot of inference.

Joshua Buchalter: And Thats, what we are now starting to see among a small group of customers.

Joshua Buchalter: Excellent. Thank you.

Vijay Rakesh: And one moment for our next question. And that will come from the line of Vijay Rakesh with Mizuho. Your line is open. Yeah, thanks. Hey, Hock, just going back on the AI server revenue side. I know you said fiscal 25 kind of tracking to that up 60%-ish growth. If you look at fiscal 26, you know, you have many new customers lapping a meta and probably, you know, you have the four or the six hyperscalers that you're talking about. Would you expect that growth to accelerate into fiscal 26 above that, you know, 60% you talked about?

Joshua Buchalter: One moment for our next question.

Speaker Change: And that will come from the line of Vijay Rakesh with Mizuho. Your line is now open.

Speaker Change: Yeah, Thanks, Hey, Hock just going back on the.

Joshua Buchalter: Nice.

Joshua Buchalter: So when you said fiscal 'twenty five.

Joshua Buchalter: Tracking to that.

Joshua Buchalter: 60% ish growth if you look at fiscal 2006, you have alright.

Joshua Buchalter: Many new customers that matter.

Joshua Buchalter: Probably before the six.

Joshua Buchalter: <unk> talked in the past when do you expect that growth to accelerate into fiscal 'twenty six.

Joshua Buchalter: 60% you're talking about.

Hock Tan: You know, my prepared remarks, in which I clarify that the rate of growth we are seeing in 2025 will sustain into 2026, based on improved visibility, and the fact that we're seeing inference coming in on top of the demand for training, as the clusters get built up bigger and bigger, still stands. I don't think we are getting very far by trying to pass through my words or data here. It's just a... And we see that going from 2025 into 2026. That's the best forecast we have at this moment. Got it. And then on the NVLink, the NVLink Fusion versus the scale-up, do you expect that market to go the route of top of the rack where you've seen some move to the Ethernet side in kind of scale-out?

Joshua Buchalter: You know minor minor it in my prepared remarks language clarifying that the rate of growth. We are seeing in 'twenty five will sustain into 2006 based on improved visibility and the fact that we're seeing inference coming in on top of the demand for training.

Joshua Buchalter: Cost of those get to be a lot bigger and bigger still stance I don't think we are getting very far by trying to pass through my words or data him.

Joshua Buchalter: Sure.

Joshua Buchalter: And we see that going from 25% 26 as the best.

Joshua Buchalter: Forecast, we have at this point.

Joshua Buchalter: Got it.

Joshua Buchalter: And we link.

Joshua Buchalter: Really two engine versus <unk>.

Joshua Buchalter: Kayla or do you expect that market to go that route.

Joshua Buchalter: There has been some more to the Ethernet side.

Hock Tan: Do you expect scale-up to kind of go the same route?

Joshua Buchalter: And kind of scale out do you expect to scale up to kind of broadly stable.

Hock Tan: Broadcom do not participate in NVLink. So I'm really not qualified to answer that question. Right. Thank you.

Speaker Change: Thanks, Broadcom do not participate in NV Ling, so I'm really not qualified to answer that question I think.

Speaker Change: Yeah.

Speaker Change: Alright, thank you.

Operator: One moment for our next question. And that will come from the line of Aaron Rakers with Wells Fargo. Your line is open. Yeah, thanks for taking the question. I think all my questions on scale up have been asked.

Speaker Change: Thank you one moment our next question.

Aaron Rakers: And that will come from the line of Aaron Rakers with Wells Fargo. Your line is open.

Aaron Rakers: Yes. Thanks for taking the question I think all my questions on scale up has been asked but I guess.

Aaron Rakers: But I guess, given the execution that you guys have been able to do with the VMware integration, looking at the balance sheet, looking at the depth structure, I'm curious if you could give us your thoughts on how the company thinks about capital return versus the thoughts on M&A and the strategy going forward. Thank you. Okay, that's a, that's an interesting question. And, and I agree, not to not to untimely, I would say, because yeah, we've done a lot of the integration of VMware now. And you can see that in a, in a level of free cash flow we're generating from operations.

Joshua Buchalter: Given the execution.

Joshua Buchalter: <unk> been able to do with the Vmware integration looking at the balance sheet looking at the debt structure I am curious.

Joshua Buchalter: You could give us your thoughts on how the company thinks about capital return versus the thoughts on M&A and the strategy going forward. Thank you.

Joshua Buchalter: Okay.

Joshua Buchalter: That's an interesting question and I agree not to.

Joshua Buchalter: Not on timely I would say.

Joshua Buchalter: Yeah, we have done.

Joshua Buchalter: A lot of the integration of Vmware now and you can see that in the in the level of free cash flow, we are generating from operations and as we said the use of capital has always been we're very.

Hock Tan: And as we said, the user capital has always been, we're very... I guess measured and upfront with a return through dividends. This is half our free cash flow of the preceding year. And, frankly, as Kirsten has mentioned... Three months ago and six months ago during those last two earnings calls, the first choice Typically of the other part of the free cash flow is to bring down our debt to a more, to a level that we feel closer to no more than two. uh... ratio of debt to EBITDA uh... and that doesn't mean that opportunistically we may go out there and buy back our share as we did last quarter.

Joshua Buchalter: I guess a measured.

Joshua Buchalter: And upfront.

Joshua Buchalter: Our return through dividends, which is half of our free cash flow of the preceding year.

Joshua Buchalter: And frankly.

Kirsten Hawkins: Kirsten has mentioned.

Joshua Buchalter: Three months ago six months ago to in the last two earnings call. The first choice typically of the other three are part of the free cash flow is to bring down our debt.

Joshua Buchalter: Two a more tour level.

Joshua Buchalter: We feel closer to no more than two.

Joshua Buchalter: Our ratio a ratio ratio of debt to EBITDA.

Joshua Buchalter: And that doesn't mean that Opportunistically, we may go out and buyback our shares.

Joshua Buchalter: We did last quarter.

Hock Tan: and Indicator by Kirsten when we did $4.2 billion of stock buyback. Now part of it is used to... Basically, when RSUs, employee RSUs, VETs... Basically use a we basically buy back part of the shed Thank you all! Basically, we think that it's a good time to buy some shares back. We do. But having said all that... Our use of cash outside of dividends would be at this stage used towards reducing our debt.

Joshua Buchalter: And indicate that by kitchen, when we did $4 2 billion of stock buyback now part of it is used to.

Joshua Buchalter: Basically when Rs employee RF skus.

Joshua Buchalter: Yes.

Joshua Buchalter: <unk> use a we basically buy bank part of the shares.

Joshua Buchalter: <unk> used to be paying taxes on the investor a rescue but the other part of it I do I do a mint we use it opportunistically last quarter, when we see an opportune situation win.

Joshua Buchalter: Basically we think that it's a good time to buy some shares back we do but having said all of that.

Joshua Buchalter: Our use of cash.

Joshua Buchalter: Outside of the dividends would be at this stage used towards reducing our debt.

Hock Tan: And I know you're going to ask, what about M&A? Well. The kind of M&A we will do. What in our view would be significant, would be substantial enough that... We need that. And it's a good use of a free cash flow to bring down debt to, in a way, expand, if not preserve, our borrowing capacity. We have to do another M&A. Thank you.

Joshua Buchalter: And I know you're going to ask what about M&A well.

Joshua Buchalter: The kind of M&A, we will do well.

Joshua Buchalter: In our view would be significant would be substantial enough debt.

Joshua Buchalter: We need that in.

Joshua Buchalter: In any case and it's a good use of our free cash flow to bring down debt to in a way expand if don't preserve our borrowing capacity. If we have to do another M&A deal.

Hock Tan: One moment for our next question. And that will come from the line of Srini Pajjuri with Raymond James. Your line is open. Thank you. Hock, a couple of clarifications. First, on your 2026 expectation, are you assuming any meaningful contribution from the four prospects that you talked about? Um No comment. We don't talk on prospects, we only talk on customers. Okay, fair enough. And then my other clarification is that I think you talked about networking being about 40% of the mix within AI. Is that the right kind of mix that you expect going forward? Or is that, you know, going to materially change as we, I guess, you know, see XPS ramping, you know, going forward?

Speaker Change: Thank you one moment our next question.

Speaker Change: And that will come from the line of shiny <unk> with Raymond James Your line is open.

Speaker Change: Okay.

Speaker Change: Thank you Hock couple of clarifications first on your 2026 expectation are you assuming any meaningful contribution from the four prospects that you talked about.

Joshua Buchalter: [laughter] no comment.

Joshua Buchalter: We don't have prospects, we only talk about customers.

Speaker Change: Okay Fair enough and then my other clarification is that I think you've talked about networking being about 40% of the mix within AI is that the right kind of mix that you expect going forward or is that going to materially change as we.

Joshua Buchalter: C Xps ramping.

Joshua Buchalter: You know going forward.

Srinivas Pajjuri: No, I've always said, and I expect that to be the case. going forward in 26 as we grow. Networking ratio to XPU should be closer in the range of less than 30%. Thank you. One moment for our next question.

Joshua Buchalter: I always say and I expect that to be the case in going forward in 2006, as we grow that.

Joshua Buchalter: Networking should be a ratio to XP, you should be closer in the range of less than 30%.

Joshua Buchalter: Not the 40%.

Speaker Change: Thank you one moment our next question.

Joe Moore: and that will come from the line of Joe Moore with Morgan Stanley. Your line is open. Great, thank you. You said you're not going to be impacted by export controls on AI. I know there's been a number of changes since in the industry since the last time you made the call. Is that still the case? And just, you know, can you give people comfort that there's no impact from that down the road? Nobody can give anybody comfort in this environment, Joe. You know that. Rules are changing quite dramatically as trade bilateral. Transcripts provided by Transcription Outsourcing, LLC.

Operator: And that will come from the line of Joe Moore with Morgan Stanley. Your line is open.

Joe Moore: Great. Thank you you said youre not going to be impacted by export controls on AI.

Joe Moore: Theres been a number of changes in the industry since the last time I think that's still the case and just can you give people comfort yet.

Joe Moore: There is no impact from that down the road.

Joe Moore: Nobody can give anybody come further in this environment. Joe you know the rules are changing quite dramatically.

Joe Moore: Trade bilateral.

Joe Moore: Trade agreements continue to be negotiated.

Joshua Buchalter: Very very dynamic.

Joshua Buchalter: A dynamic environment, so I'll be honest I don't I don't know I know as little as probably you probably know more than I do maybe.

Hock Tan: I know very little about this whole thing about whether there is any export control, how the export control will take place. So, I'd rather not answer that because, no, I don't know whether it will...

Joshua Buchalter: In which case, then I know very little about this whole thing about.

Joshua Buchalter: Whether there's any export control how the export control will take place.

Joshua Buchalter: So I'd, rather not answer that because I don't know whether it will be.

Hock Tan: Thank you.

Operator: And we do have time for one final question.

Speaker Change: Thank you and we do have time for one final question.

William Stein: And that will come from the line of William Stein with Truist Securities. Your line is open. Great, thank you for squeezing me in. I wanted to ask about VMware. Can you comment as to how far along you are in the process of converting customers to the subscription model? Is that close to complete? Or is there still a number of quarters that we should expect that that conversion continues? That's a good question. And so let me start off by saying a good way to measure it You know, most of our VMOA contracts are typically three-year contracts.

William Stein: And that will come from the line of William Stein with <unk> Securities. Your line is open.

William Stein: Great. Thank you for squeezing me in I wanted to ask you about Vmware can.

Joshua Buchalter: Can you comment as to how far along you are in the process.

Joshua Buchalter: Converting customers to the subscription model does that.

Joshua Buchalter: Close to complete or there is still a number of quarters that you should expect that.

Joshua Buchalter: That conversion continues.

Joshua Buchalter: That's a good question and so let me start off by saying a good way to measure it as you know most of our Vms contracts.

Joshua Buchalter: About three.

Joshua Buchalter: Typically three years and that was once Vmware did before we acquired them and that's pretty much. What we continued to do three years very true digital so.

William Stein: And that was one VMware did before we acquired them, and that's pretty much what we continue to do. Three as well. So, based on that, the renewals, we're like two-thirds of the way, almost two-thirds, halfway, more than halfway through the renewals, so we probably have at least another year, plus maybe a year and a half to go. Thank you. And with that, I'd like to turn the call over to Ji Yoo for closing remarks. Thank you, operator. Broadcom currently plans to report its earnings for the third quarter of fiscal year 2025. After close of market on Thursday, September 4, 2025.

Joshua Buchalter: Based on that the renewals were like two thirds of the way almost at the halfway more than halfway through the renewals. So we probably have at least another year, plus maybe a year and a half to go.

Speaker Change: Thank you and with that I'd like to turn the call over to <unk> for closing remarks.

Joshua Buchalter: Thank you operator, Broadcom currently plans to reported earnings for the third quarter of fiscal year 2025, after close of market on Thursday September four 2025 public.

A public webcast of Broadcom's earnings conference call will follow at 2pm. That will conclude our earnings call today. Thank you all for joining. Operator, you may end the call.

Joshua Buchalter: Public webcast of Broadcom earnings Conference call will follow at two P M Pacific.

Joshua Buchalter: That will conclude our earnings call today. Thank you all for joining operator, you may end the call.