Q1 2025 Broadcom Inc Earnings Call

[music].

Unknown Executive: 23 Welcome to the Broadcom Inc.'s first quarter fiscal year 2025 financial results conference call.

One.

Speaker Change: Welcome to the Broadcom, Inc. 's first quarter fiscal year 2025 financial results conference call. 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.

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, Sheree. And good afternoon, everyone.

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

Ji Yoo: Joining me on today's call are Hock Tan, President and CEO, Kirsten Spears, Chief Financial Officer, and Charlie Kawwas, President, Semiconductor Solutions Group. Broadcom distributed a press release and financial tables after the market closed, describing our financial performance for the first quarter of fiscal year 2025. If you did not receive a copy, you may obtain the information from the Investor section of 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 first quarter fiscal year 2025 results.

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

Speaker Change: Since spirits Chief Financial Officer.

Charlie Clark: Charlie Clark President semiconductor solutions group.

Charlie Clark: Broadcom distributed a press release and financial tables after the market closed.

Charlie Clark: Describing our financial performance for the first quarter of fiscal year 2025.

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

Charlie Clark: This conference call is being webcast live.

Charlie Clark: And then audio replay of the call can be accessed for one year to the investors section of Broadcom at the website.

Speaker Change: During the prepared comments Hawkins Kirsten will be providing details of our first quarter of fiscal year 2025 results.

Ji Yoo: guidance for our second quarter of fiscal year 2025, as well as commentary regarding the business environment. We will take questions after the end of our prepared comments.

Speaker Change: Guidance for our second 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.

Ji Yoo: 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. In addition to U.S. GAAP reporting, Broadcom reports certain financial measures on a non-GAAP basis. A reconciliation between GAAP and non-GAAP measures is included in the tables attached to today's press release. Comments made during today's call will primarily refer to our non-GAAP financial results.

Speaker Change: Please refer to our press release today.

Speaker Change: 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.

Speaker Change: <unk> Com reports certain financial measures on a non-GAAP basis.

Speaker Change: A reconciliation between GAAP and non-GAAP measures is included in the tables attached to todays press release.

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

Hock Tan: I'll now turn the call over to Hock. Thank you, Ji. and thank you everyone for joining today.

Hawk: I'll now turn the call over to Hawk.

Speaker Change: Thank you Qi.

Hawk: And thank you everyone for joining today.

Hock Tan: in a fiscal Q1 2025. Total revenue was a record $14.9 billion. up 25% year-on-year. and Consolidated Adjusted EBITDA was a record again, $10.1 billion, up 41% year-on-year.

Hawk: In our fiscal Q1 2025.

Hawk: Total revenue was a record $14 $9 billion.

Hawk: Up 25% year on year.

Hawk: And consolidated adjusted EBITDA was a record gain $10 1 billion up 41% year on year.

Hock Tan: So let me first provide color on our semiconductor business. Q1 Semiconductor Revenue was $8.2 billion, up 11% year-on-year. Growth was driven by AI, as AI revenue of $4.1 billion was up 77% year-on-year. We bid our guidance for AI revenue of $3.8 billion. Due to stronger shipments of networking solutions to hyperscalers on AI. Our hyperscaled partners. continue to invest aggressively in their next-generation frontier model. which do require high performance accelerators. as well as AI data centers with larger clusters.

Hawk: So let me first provide color on our semiconductor business.

Hawk: Q1 semiconductor revenue was $8 2 billion.

Hawk: 11% year on year.

Hawk: Growth was driven by E S.

Hawk: <unk> revenue of $4 1 billion was up 77% year on year.

Hawk: Beat our guidance for AI revenue of $3 8 billion due to stronger shipments of networking solutions to Hyperscale is on it yet.

Hawk: Our Hyperscale partners.

Hawk: Continued to invest aggressively in their next generation frontier models.

Hawk: We still require higher performance accelerators.

Hawk: As well as AI data centers with larger clusters.

Hawk: And consistent with this.

Hock Tan: We are stepping up our R&D investment on two fronts. One, we're pushing the envelope of technology in creating the next generation of accelerators. We're taping out the industry's first 2 nanometer AI XPU package in 3.5D as we drive towards 10,000 Teraflops XBX. Secondly, we have a view towards scaling clusters of 500,000 accelerators for hyperscale clusters. We have doubled the radix capacity of the existing Tomahawk flight. to enable AI clusters to scale up on the Internet towards 1 million XPUs. We have taped out our next generation 100 terabit Tomahawk 6 switch running 200G studies at 1.6 terabit bandwidth.

Hawk: Are stepping up our R&D investment on two fronts.

Hawk: One we're pushing the envelope of technology in creating the next generation of accelerators.

Hawk: With taping out the industry's first two nanometer XP.

Hawk: <unk> packaging 3.35 D. As we drive towards 10000, Teraflops SP X Bu.

Hawk: Secondly, we have a view towards scaling clusters of 500000 accelerators for Hyperscale customers.

Hawk: We have doubled the.

Iridex capacity of the existing amongst fine.

Hawk: And beyond this.

Hawk: To enable AI clusters to scale up on Ethernet towards 1 million ex <unk>.

Hawk: We have to tape out our next generation 100 therapy, among six switch running to 100 G studies and one six therapy.

Hawk: Bandwidth.

Hawk: We will be delivering samples to customers within the next few months.

Hock Tan: We will be delivering samples to customers within the next few months. These R&D investments are very aligned with the roadmap of our three hyperscale customers.

Hawk: These R&D investments are very aligned with our roadmap of three hyperscale customers as they each race towards $1 million SBU clusters by the <unk>.

Hock Tan: as they each race towards 1 million XPU clusters by the end of 2021. And accordingly, we do reaffirm what we said last quarter, that we expect these three hyperscale customers will generate a serviceable addressable market or SEM in the range of 60 to $90 billion in fiscal 2020.

Hawk: End of 2027.

Hawk: And accordingly, we.

Hawk: We do reaffirm what we said last quarter.

Hawk: That we expect these three hyperscale customers would generate a serviceable addressable market or Sam in the range of $60 billion to $90 billion in fiscal 2027.

Hawk: Beyond these three customers. We had also mentioned previously that.

Hock Tan: Beyond these three customers, we had also mentioned previously... that we are deeply engaged with two other hyperscalers. enabling them to create their own customized AI accelerators. We on track. take out their XPUs this year. In the process of working with the hyperscalers It has become very clear that while they are excellent in software, Broadcom is the best in Hudson. Working together is what optimizes. We are a large language mall. It is therefore no surprise to us.

Hawk: We are deeply engaged with two other hyperscale is.

Hawk: Enabling them to create their own customized AI accelerator.

Hawk: We are on track to tape out.

Hawk: XP use this year.

Hawk: In the process of working with the Hyperscale is it.

Hawk: It has become very clear.

Hawk: While they are excellent in software.

Hawk: Broadcom is the best in hardware.

Speaker Change: Working together is what optimizes.

Speaker Change: We are launch language models.

Speaker Change: It is therefore, no surprise to us.

Hock Tan: Since our last earnings call that two additional hyperscalers have selected Broadcom to develop custom accelerators to train their next generation frontier model. So even as we have three hyperscale customers, we are shipping XPUs. volume today, there are now four more who are deeply engaged with us to create their own accelerator. And to be clear, of course, these four are not included in our estimated SAM of $60-90 billion in 2021.

Speaker Change: Since our last earnings call.

Speaker Change: Two additional hyperscale is have selected broadcom to develop custom accelerators to train the next generation frontier models.

Speaker Change: So even as we have three hyperscale customers, we are shipping sbu's in volume today.

Now for more deeply engaged with us to create their own accelerators.

Speaker Change: And to be clear it's cause this fall.

Not included in our estimate the same of $60 billion to $90 billion.

Speaker Change: In 2027.

Hock Tan: So we do see an exciting trend. New Frontier Models and Techniques. put unexpected pressures on AI systems. It's difficult to serve all classes of models with a single system design point. And therefore, it is hard to imagine that a general purpose accelerator can be configured and optimized across multiple frontier models. And as I mentioned before, the trend towards XPU. is a multi-agent.

Speaker Change: So we do see an exciting trend.

Speaker Change: New frontier models.

Speaker Change: Nick.

Speaker Change: Unexpected pressures on the pressures on AI systems.

Speaker Change: It's difficult to serve all classes of models with us.

Speaker Change: Single system design point.

Speaker Change: And therefore, it is hard to imagine that a general purpose accelerator.

Speaker Change: Can be configured.

Speaker Change: And optimize across multiple frontier models.

Speaker Change: And as I mentioned before the trend towards XP use.

Speaker Change: He is a multiyear journey.

Hock Tan: So coming back to 2025. We see a steady ramp in deployment of our XPUs and networking products. Q1 AI revenue was $4.1 billion and we expect Q2 AI revenue to grow to $4.4 billion. which is up 44% year-on-year.

Speaker Change: So coming back to 2025.

Speaker Change: We see a steady ramp in deployment of our XP use in networking products.

Speaker Change: Q1 revenue was $4 1 billion and we expect Q2 revenue to grow to four 4 billion, which.

Speaker Change: Which is up 44% year on year.

Hock Tan: Turning to non-AI semiconductors. Revenue of $4.1 billion was down 9% sequentially on a seasonal decline in wildlife. In aggregate, during Q1, the recovery in non-AI semiconductors continued to be slow.

Speaker Change: Turning to non AI semiconductors revenue.

Speaker Change: Revenue of $4 1 billion was down 9% sequentially.

Speaker Change: The seasonal decline in wireless.

In aggregate during Q1, the recovery in non <unk> semiconductors.

Daniel: Daniel to be slow.

Hock Tan: Broadband, which bought them in Q4 2024, showed a double-digit sequential recovery in Q1. and is expected to be up similarly in Q2 as service providers and telcos step up spending. Server storage was down single digits sequentially in Q1. but it's expected to be up high single digits sequentially in Q2.

Daniel: Broadband, which bottom in Q4 2024 showed a double digit sequential recovery in Q1.

Daniel: <unk> is expected to be up similarly in Q2.

Daniel: Service providers and telcos step up spending.

Daniel: Server storage was down single digit sequentially in Q1.

Daniel: But it's expected to be up high single digits sequentially in Q2.

Hock Tan: Meanwhile, enterprise networking continues to remain flattish in the first half of Fiscal 25 as customers continue to work through channel inventory. One wireless was down sequentially due to a seasonal decline. remain flat year on year. Q2, Wireless is expected to be the safest. flat again year on year.

Meanwhile, enterprise networking.

Daniel: The news to remain flattish in the first half of fiscal 'twenty five.

Daniel: As customers continue to work through channel inventory.

Daniel: One wireless was down sequentially due to a seasonal decline.

Daniel: <unk> remained flat year on year.

Daniel: In Q2 wireless is expected to be the same.

Daniel: Slide again year on year.

Hock Tan: Resales in industrial were down double digits in Q1 and are expected to be down in Q2. So reflecting the foregoing puts and takes. We expect non-AI semiconductor revenue in Q2 to be flattish sequentially. Even though we are seeing bookings continue to grow year on year.

Daniel: Resales and industrial were down double digits in Q1.

Daniel: And I expect it to be down in Q2.

Daniel: So, reflecting the foregoing puts and takes.

Daniel: We expect non AI semiconductor revenue in Q2.

Daniel: To be flattish sequentially.

Daniel: Even though we are seeing bookings continue to grow.

Daniel: Yes.

Daniel: Yeah.

Hock Tan: In summary for Q2, We expect total semiconductor revenue to grow 2% sequentially and up 17% year-on-year to $8.4 billion.

Daniel: In summary for Q2.

Daniel: We expect total semiconductor revenue to grow 2% sequentially and up 17% year on year to $8 4 billion.

Hock Tan: Turning now to Infrastructure Software Center. Q1 infrastructure software revenue of $6.7 billion was up 47% year-on-year and up 15% sequentially. Exaggerated, though, by deals which slip from Q4 into Q1.

Daniel: Turning now to infrastructure software.

Daniel: The segment.

Daniel: Q1 infrastructure software revenue of $6 $7 billion was up 47% year on year and up 15% sequentially.

Daniel: As they generated by deals, which slipped from Q to Q4 Q1.

Hock Tan: Now this is the first quarter, Q1-25, where the year-on-year comparables include VMware in both quarters. We're seeing significant growth in the software segment for two reasons. One, we're converting to a footprint of large, we're, sorry, we're converting. from a footprint of largely perpetual life. to one of full subscriptions. And as of today, we are over 60% done. These perpetual licenses were only largely for compute virtualization. otherwise called The Sphere. We are. upselling customers to a full stack. VCS. which enables the entire data centers to be virtualized. And this enables customers to create their own private cloud environment on-prem.

Daniel: Now this is the first quarter Q1, 'twenty five where the year on year Comparables include Vmware.

Daniel: In both quarters.

Daniel: We're seeing significant growth in our software segment for two reasons.

Daniel: One we're converting to a footprint of launched.

Daniel: We are considering we're converting.

Daniel: From a footprint of largely perpetual license.

Daniel: Two one off full subscription.

Daniel: And as of today.

Daniel: Over 60% done.

Daniel: Two.

Daniel: This perpetual licenses will only largely for compute virtualization.

Daniel: One is called the sphere.

Daniel: We are.

Daniel: Upselling customers to a full stack.

Daniel: Which enables the entire data centers to be Virtualized.

Daniel: And this enables customers to create their own private cloud environment on Prem.

Hock Tan: and that's the end of Q1. Approximately 70% of our largest 10,000 customers. have adopted. As these customers consume VCF We still see a further opportunity for future growth.

Daniel: And as of the end of Q1.

Daniel: Approximately 70% of our largest 10000 customers.

Daniel: <unk> adopted vcs.

Daniel: As these customers consume vcs.

Daniel: Just to see a further opportunity for future growth.

Daniel: As large enterprises adopt AI.

Hock Tan: at Large Enterprises at Dock AI. They have to run their AI workload. on their on-prem data center. which will include both GPU servers as well as traditions. And just as VCF virtualizes these traditional data centers using CPUs. VCF will also virtualize GPUs on a common platform and enable enterprises to import AI models to run their own data on press.

Daniel: They have to run the AI workloads.

Daniel: On their on Prem data centers, which will include both GPU servers as well as traditional Cpus.

Daniel: And just as Vcs Virtualized us these traditional datacenters using gpus.

Daniel: Vcs will also virtualized gpus on the on a common platform.

Daniel: Enable enterprises to import AI models to run their own data on Prem.

Hock Tan: This platform, which virtualized the GPU, is called the VMware Private AI Foundation. And as of today, in collaboration with NVIDIA, we have 39 enterprise customers for the VMware Private AI Foundation. Customer demand has been driven by our open ecosystem. Superior Load Balancing and Automation Capability. that allows them to intelligently pull and run workloads across both GPU and CPU infrastructure and leading to very reduced costs.

Daniel: This platform, which virtualized the GPU is called the Vmware private AI Foundation.

Daniel: And as of today in collaboration with Nvidia.

Daniel: We have 39 enterprise customers for the Vmware private AI Foundation.

Daniel: Customer demand has been driven by our open ecosystem superior load balancing and automation capabilities.

Daniel: That allows them to intelligently pool and run workloads across both GPU and CPU infrastructure and leading to very reduced costs.

Hock Tan: Moving on to Q2 Outlook for software. We expect revenue of $6.5 billion. Up 23% year-on-year.

Daniel: Moving onto Q2 outlook for software.

Speaker Change: We expect revenue of $6 5 billion.

Daniel: Up 23% year on year.

Speaker Change: So in <unk>.

Hock Tan: So in total, we're getting Q2 consolidated revenue to be approximately $14.9 billion. Up 19% year-on-year. This, we expect, this will drive Q2 adjusted EBITDA to approximately...

Speaker Change: Total, we're guiding Q2 consolidated revenue to be approximately $14 $9 billion.

Speaker Change: Up 19% year on year end.

And this we expect this will drive Q2, adjusted EBITDA to approximately 66% of revenue.

Kirsten Spears: With that, let me turn the call over to Kirsten. Thank you, Hock. Let me now provide additional detail on our Q1 financial performance.

Speaker Change: With that let me turn the call over to kiss them.

Speaker Change: [laughter].

Speaker Change: Thank you Hock, let me now provide additional detail on our Q1 financial performance.

Kirsten Spears: From a year on year comparable basis, keep in mind that Q1 of fiscal 2024 was a 14 week quarter and Q1 of fiscal 2025 is a 13 week quarter. Consolidated revenue was $14.9 billion for the quarter, up 25% from a year ago. Gross margin was 79.1% of revenue in the quarter, better than we originally guided on higher infrastructure software revenue and more favorable semiconductor revenue mix. Consolidated operating expenses were $2 billion, of which $1.4 billion was for R&D. Q1 operating income of $9.8 billion was up 44% from a year ago, with operating margin at 66% of revenue.

Speaker Change: On a year on year comparable basis keep in mind that Q1 of fiscal 2024 was a 14 week quarter in Q1 of fiscal 2025 is a 13 week quarter.

Speaker Change: Validated revenue was $14 9 billion for the quarter up 25% from a year ago.

Speaker Change: Gross margin was 79, 1% of revenue in the quarter better than we originally guided on higher infrastructure software revenue and more favorable semiconductor revenue mix.

Speaker Change: Consolidated operating expenses were $2 billion of which one 4 billion was for R&D.

Speaker Change: Q1 operating income of $9 8 billion was up 44% from a year ago with operating margin at 66% of revenue.

Kirsten Spears: Adjusted EBITDA was a record $10.1 billion, or 68% of revenue, above our guidance of 66%. This figure excludes $142 million of depreciation.

Speaker Change: Adjusted EBITDA was a record $10 1 billion or 68% of revenue above our guidance of 66%.

Speaker Change: This figure excludes 142 million of depreciation.

Kirsten Spears: Now a review of the PNL for our two segments.

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

Kirsten Spears: starting with Semiconductor. Revenue for our Semiconductor Solutions segment was $8.2 billion and represented 55% of total revenue in the quarter. This was up 11% year on year.

Speaker Change: <unk> with semiconductors.

Speaker Change: Revenue for our semiconductor solutions segment was $8 2 billion and represented 55% of total revenue in the quarter. This was up 11% year on year.

Kirsten Spears: Gross Margin for our Semiconductor Solutions segment was approximately 68%, up 70 basis points year-on-year, driven by revenue mix. Operating expenses increased 3% year-on-year to $890 million on increased investment in R&D for leading-edge AI semiconductors, resulting in a semiconductor operating margin of 57%.

Speaker Change: Gross margin for our semiconductor solutions segment with approximately 68% up 70 basis points year on year driven by revenue mix.

Speaker Change: Operating expenses increased 3% year on year to 890 million on increased investment in R&D for leading edge AI semiconductors, resulting in semiconductor operating margin of 57%.

Kirsten Spears: Now moving on to infrastructure software. Revenue for infrastructure software of $6.7 billion with 45% of total revenue and up 47% year-on-year based primarily on increased revenue from VMware. Gross margin for infrastructure software was 92.5% in the quarter compared to 88% a year ago. Operating expenses were approximately $1.1 billion in the quarter, resulting in infrastructure software operating margin of 76%. This compares to operating margin of 59% a year ago.

Speaker Change: Now moving on to infrastructure software revenue.

Speaker Change: Revenue for infrastructure software of $6 7 billion was 45% of total revenue and up 47% year on year based primarily on increased revenue from Vmware.

Speaker Change: Gross margin for infrastructure software was 92, 5% in the quarter compared to 88% a year ago.

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

Speaker Change: This compares to operating margin of 59% a year ago.

Kirsten Spears: This year on your improvement reflects our disciplined integration of VMware and sharp focus on deploying our VCF strategy.

Speaker Change: This year on year improvement reflects our disciplined integration of Vmware and sharp focus on deploying our vcs strategy.

Kirsten Spears: Moving on to cash flow. Free cash flow in the quarter was $6 billion and represented 40% of revenue. Free cash flow as a percentage of revenue continues to be impacted by cash interest expense from debt related to the VMware acquisition and cash taxes due to the mix of U.S.

Speaker Change: Moving on to cash flow free cash flow in the quarter was 6 billion and represented 40% of revenue free cash flow as a percentage of revenue continues to be impacted by cash interest expense from debt related to the Vmware acquisition and cash taxes due to the mix of U S. Taxable income the continued delay.

Kirsten Spears: taxable income, the continued delay in the reenactment of Section 174, and the impact of corporate AMT. We spent $100 million on capital expenditures. Day sales outstanding were 30 days in the first quarter compared to 41 days a year ago. We ended the first quarter with inventory of $1.9 billion up 8% sequentially to support revenue in future quarters. Our days of inventory on hand were 65 days in Q1, as we continue to remain disciplined on how we manage inventory across the ecosystem. We ended the first quarter with $9.3 billion of cash and $68.8 billion of gross principal debt.

Speaker Change: And the re enactment of section 174, and the impact of corporate AAM team.

Speaker Change: We spent 100 million on capital expenditures.

Speaker Change: Days sales outstanding were 30 days in the first quarter compared to 41 days a year ago. We ended the first quarter with inventory of $1 9 billion up 8% sequentially to support revenue in future quarters.

Speaker Change: Days of inventory on hand were 65 days in Q1 as we continue to remain disciplined on how we manage inventory across the ecosystem.

Speaker Change: We ended the first quarter was $9 3 billion of cash and $68 8 billion of gross principal debt.

Kirsten Spears: During the quarter, we repaid $495 million of fixed rate debt and $7.6 billion of floating rate debt with new senior notes, commercial paper, and cash on hand, reducing debt by a net $1.1 billion. Following these actions, the weighted average coupon rate and years to maturity of our $58.8 billion in fixed rate debt is 3.8% and 7.3 years respectively. The weighted average coupon rate and years to maturity of our $6 billion in floating rate debt is 5.4% and 3.8 years respectively. And our $4 billion in commercial paper is at an average rate of 4.6%.

Speaker Change: During the quarter, we repaid $495 million of fixed rate debt and $7 6 billion of floating rate debt with new senior note commercial paper and cash on hand, reducing debt by a net $1 1 billion.

Speaker Change: Following these actions the weighted average coupon rate in years to maturity of our $58 8 billion in fixed rate debt is three 8% and seven three years respectively.

Speaker Change: The weighted average coupon rate in years to maturity of our 6 billion in floating rate debt is five 4% and three eight years, respectively, and our 4 billion of commercial paper is at an average rate of four 6%.

Kirsten Spears: Turning to Capital Allocation. In Q1, we paid stockholders $2.8 billion of cash dividends based on a quarterly common stock cash dividend of $0.59 per share. We spent $2 billion to repurchase 8.7 million ABGO shares from employees as those shares vested for withholding tax.

Speaker Change: Turning to capital allocation.

Speaker Change: Q1, we paid stockholders $2 8 billion of cash dividends based on our quarterly common stock cash dividend of <unk> 59 per share. We spent 2 billion to repurchase eight 7 million <unk> shares from employees as those shares vested for withholding taxes.

Kirsten Spears: In Q2, we expect the non-GAAP diluted share count to be approximately 4.95 billion shares.

Speaker Change: In Q2, we expect our non-GAAP diluted share count to be approximately $4 95 billion shares now.

Kirsten Spears: Now moving on to guidance. Our guidance for Q2 is for consolidated revenue of $14.9 billion with semiconductor revenue of approximately $8.4 billion, up 17% year-on-year. We expect Q2 AI revenue of $4.4 billion, up 44% year-on-year. For non-AI semiconductors, we expect Q2 revenue of $4 billion. We expect Q2 infrastructure software revenue of approximately $6.5 billion, up 23% year-on-year. We expect Q2 Adjusted EBITDA to be about 66%.

Speaker Change: Now moving onto guidance our guidance for Q2 is for consolidated revenue of $14 9 billion with semiconductor revenue of approximately $8 4 billion up 17% year on year.

Speaker Change: We expect Q2 revenue of $4 4 billion up 44% year on year.

Speaker Change: Fernand AI semiconductors, we expect Q2 revenue of $4 billion.

Speaker Change: We expect Q2 infrastructure software revenue of approximately $6 5 billion up 23% year on year.

Speaker Change: We expect Q2, adjusted EBITDA to be about 66%.

Kirsten Spears: For modeling purposes, we expect Q2 Consolidated Growth Margin to be down approximately 20 basis points sequentially on the revenue mix of infrastructure, software, and product mix within semiconductors. As Hock discussed earlier, we are increasing our R&D investment in leading-edge AI in Q2, and accordingly, we expect adjusted EBITDA to be approximately 66%.

Speaker Change: For modeling purposes, we expect Q2 consolidated gross margin to be down approximately 20 basis points sequentially on the revenue mix up infrastructure software and product mix within semiconductors.

Speaker Change: As Hock discussed earlier, we are increasing our R&D investment in leading edge AI in Q2, and accordingly, we expect adjusted EBITDA to be approximately 66%.

Kirsten Spears: We expect the non-GAAP tax rate for Q2 and fiscal year 2025 to be approximately 14%.

Speaker Change: We expect the non-GAAP tax rate for Q2 and fiscal year 2025 to be approximately 14%.

Kirsten Spears: That concludes my prepared remarks.

Speaker Change: That concludes my prepared remarks, operator, please open up the call for questions.

Unknown Executive: 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, 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: Thank you to ask a question you will need to press star one one on your telephone to withdraw your question 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: Okay.

Benjamin Reitzes: And our first question will come from the line of Ben Reitzes with Mellius. Your line is open. Hey, guys, thanks a lot. And congrats on the results. Hock, you talked about four more customers coming online. Can you just talk a little bit more about, you know, the trend you're seeing? Can any of these customers be as big as the current three? And what what does this say about the custom silicon trend overall, and your optimism and upside to the business long term? Thanks.

Speaker Change: And our first question will come from the line of Ben Reitzes with Meli asked your line is open.

Ben Reitzes: Hey, guys. Thanks, a lot and congrats on the results Hock you talked about four more customers coming online.

Speaker Change: Can you just talk a little bit more about that.

Speaker Change: The trend Youre seeing can any of these customers.

Speaker Change: As big as the current three and.

Speaker Change: What does this say about the custom silicon trend overall, and your optimism and upside to the business long term. Thanks.

Ben Reitzes: Well very interesting question Ben.

Hock Tan: Well, very interesting question, Ben, and thanks for your kind wishes, but... What we think as, and by the way, these four are not yet customers as we defined them. As I've always said... You know, in developing and creating XPU. You know, we are not really the, we are not really the creator of those XPUs, to be honest. We enable... These are those hyperscalers partners we engage with to create. that ship, basically to create that compute system, call it that way, and it comprises the model, the software model, working closely with the compute engine, the XPU and the networking that binds together the clusters, those multiple XPUs as a whole to train those large frontier models.

Speaker Change: Thanks for your kind wishes, but.

Speaker Change: What we've seen.

Speaker Change: And by the way this fall are not yet customers as we define it.

Speaker Change: Yes.

Speaker Change: As I've always said.

Speaker Change: In developing and creating XP use.

Speaker Change: We are not really that we are not really the creator of.

Speaker Change: Those XP used to be honest we in April.

Speaker Change: Each of those Hyperscale as partners, we engaged with two.

Speaker Change: To create.

Speaker Change: That's true.

To create that compute system call it that way and it comprises.

Speaker Change: The model the software model working closely we've Uh huh.

Speaker Change: The compute engine the SBU in the networking.

Speaker Change: Together the clusters.

Speaker Change: Those multiple <unk> as a whole to train those lunch front.

Speaker Change: Frontier models.

Hock Tan: The fact that we create the hardware, it still has to work with the software models and algorithms of those partners of ours before it becomes fully deployable. Key, which is why we define customers in this case as those where we know they have deployed a skill and will receive the production volume to enable this. And for that, we only have... Just to read through it. for a I call it. partners who are trying to create the same thing as the first. and to run their own frontier model. And as I also said, it doesn't happen overnight.

Speaker Change: So in the end.

Speaker Change: The fact that we create the hardware.

Speaker Change: <unk> has worked.

Speaker Change: With the software models and algorithms of all of those partners of ours before it becomes fully deployable and SKU.

Speaker Change: Which is why we defined customers in this case as those where we know they are deployed at scale and we received the production volume to enable them to run and for that we only.

Speaker Change: Just to read through the <unk>.

Speaker Change: For.

Speaker Change: I call it.

Speaker Change: Partners, who are trying to create the same thing as the first three.

Speaker Change: And to run their own frontier models.

Speaker Change: Is it on you to train their own frontier models.

Speaker Change: And as I also said it doesn't happen overnight to do the first chip could take would take typically a year in enhance.

Hock Tan: To do the first chip would take typically a year and a half. And that's very accelerated, and which we could accelerate, given that we essentially have a framework and a methodology that works right now. It works for the three customers, no reason for it to not work for the four. But we still need those four partners. create and to develop the software. which we don't do to make it work. And to answer your question, there's no reason why these four guys. would not create a demand in the range of what we're seeing with the first three guys, but probably later.

Speaker Change: And thats very accelerated and which we could accelerate given that we essentially have a framework and a methodology that works right now and what's sort of the three customers no reason for it to not work for the fall.

Speaker Change: Still need those four partners to create and to develop the software.

Which we don't do to make it work.

Speaker Change: To answer your question, there's no reason why this fault guys.

Speaker Change: Well not creed.

Speaker Change: Demand in the range of what we've seen with the first street guidance bump probably later.

Hock Tan: Journey. They started it later and so they will probably get there later.

Speaker Change: It's a journey that started it later and so they will probably get this later.

Unknown Executive: Thank you very much. Thank you.

Speaker Change: Thank you very much.

Speaker Change: Thank you one moment for our next question and that will come from the line of Harlan sur with Jpmorgan. Your line is open.

Harlan Sur: One moment for our next question and that will come from the line of Harlan Sur with JP Morgan. Your line is open. Good afternoon and great job on the strong quarterly execution, Hock and team. Great to see the continual momentum in the AI business here in the first half of your fiscal year and the continued broadening out of your AI ASIC customers. I know, Hock, last earnings, you did call out a strong ramp in the second half of the fiscal year driven by new three nanometer AI accelerated programs kind of ramping.

Harlan Sur: Good afternoon, and great job on the strong quarterly execution Hopkinton grid.

Harlan Sur: Great to see the continued momentum in the business here in the first half of your fiscal year and <unk>.

Continuing broadening out of your AI Asa customers I know Hawk last earnings you did call out a strong ramp in the second half of the fiscal year driven by new three nanometer.

Harlan Sur: Related programs kind of ramping can you just help us either qualitatively or quantitatively profile.

Harlan Sur: Can you just help us either qualitatively, quantitatively profile this second half step up relative to what the team just delivered here in the first half? Has the profile changed either favorably, less favorably versus what you thought maybe 90 days ago? Because quite frankly, I mean, a lot has happened since last earnings, right? You've had the dynamics like DeepSeek and focus on AI model efficiency, but on the flip side, you've had strong CapEx outlooks by your cloud and hyperscale customers. So any color on the second half AI profile would be helpful.

Speaker Change: Can have step up relative to what the team just delivered here in the first half has a profile change either favorably less favorably versus what you thought maybe 90 days ago, because quite frankly, I mean, a lot has happened since last earnings right <unk> had the dynamics like deep seek.

Harlan Sur: Focus on.

Harlan Sur: <unk> modeled efficiency, but on the flip side, you've had strong capex outlooks by your cloud and Hyperscale customers. So any color on the second half profile would be helpful.

Speaker Change: You're asking me to look into the minds of my customers and I hate to tell me. They don't tell you. They don't show me then.

Hock Tan: You're asking me to look into the minds of my customers and I hate to tell you they don't show me their entire mindset here, but why are we beating the numbers so far in Q1? seems to be encouraging in Q2. And partly from improved networking shipments, as I indicated, to those XPUs and AI accelerators. Even in some cases, GPUs together for the hyperscalers, and that's good. And partly, also, we think there is some pull-in. of commissioned literature. And on the second half. that you talked about 90 days ago, the second half three nanometer ramp, is that still very much on track?

Speaker Change: <unk> mindset here, but one why we're beating the numbers so far in Q1.

Speaker Change: And it seems to be encouraging in Q2.

Speaker Change: Police from improved networking shipments as I indicated to the.

Speaker Change: Those sp use.

Speaker Change: Salary this.

Speaker Change: Even in some cases gpus together.

Speaker Change: For the Hyperscale is.

Speaker Change: And thats, good and partly also thing.

Dave.

Speaker Change: Pull ins.

Shipments and acceleration call it that way of shipments.

Speaker Change: In fiscal 2005.

Speaker Change: And on the second half.

Speaker Change: You talked about 90 days ago, the second half nanometer ramp is still very much on track.

Harlan Sur: Harlan, thank you.

Harlan Sur: Harlan Thank you.

Hock Tan: I only got you two, sorry, let's not speculate on the second half. Okay. Thank you, Hock. Thank you.

Speaker Change: Only guide Q2, sorry, let me, let's not speculate on the second half.

Hock Tan: Okay. Thank you hock.

Speaker Change: Thank you one moment, our next question and that will come from the line of William Stein with <unk> Securities. Your line is open.

William Stein: And that will come from the line of William Stein with Truist Securities. Your line is open. Great, thank you for taking my question. Congrats on these pretty great results. It, you know, it seems from the news headlines about tariffs and about deep seek that there may be some disruptions, some customers and some other complimentary suppliers are seem to feel a bit paralyzed, perhaps and have difficulty making tough decisions. Those tend to be you know, really useful times for great companies to sort of emerge as something bigger and better than they were in the past. You've, you know, grown this company in a tremendous way over the last decade plus.

William Stein: Great. Thank you for taking my question Congrats on these pretty great results.

Speaker Change: Got.

Speaker Change: It seems from the.

Speaker Change: The news headlines about tariffs and about deep seek that.

Speaker Change: Maybe some disruption.

Speaker Change: Disruption some customers and some other complementary suppliers or seem to feel a bit paralyzed perhaps.

Speaker Change: Difficulty, making tough decisions.

Speaker Change: Those tend to be.

Speaker Change: Really useful times for great companies to sort of emerge as something bigger is better than they were in the past.

Speaker Change: Grown this company.

Speaker Change: Mendes way over the last decade plus.

William Stein: And you're doing great now, especially in this AI area. But I wonder if you're seeing that sort of disruption from from these dynamics that we suspect are happening based on, you know, headlines, what we see from other companies, and how, aside from adding these customers in AI, I'm sure there's other great stuff going on.

Speaker Change: And Youre doing great now, especially in this AI area, but I wonder if you're seeing that sort of disruption from from these dynamics that we see.

Speaker Change: Suspect are happening based on headlines of what we see from other companies and how aside from adding these customers and AI I'm sure. There's other great stuff going on but should we expect some bigger changes to come from Broadcom as a result of this.

Hock Tan: But should we expect some bigger changes to come from Broadcom as a result of this? You pose a very interesting set of issues and questions, and those are very relevant, The only issue, the only problem we have at this point... It's really two worlds. to know why we're all lame. I mean, that's the threat, the noise of terrorism. Special. Chip. that hasn't materialized yet, nor do we know how it will be structured. So we don't know, but we do experience and we are living it. Now is the disruption that pause in a positive way, I should add a very positive disruption.

Speaker Change: Mhm.

Speaker Change: You posted a very interesting set of issues and questions.

Speaker Change: And those are very relevant and interesting issues.

Speaker Change: The only issue the only problem we have at this point is.

Speaker Change: I'd say it is.

Speaker Change: It's really too early.

Speaker Change: To Norway will all land I mean, thats the threat the noise of territories.

Speaker Change: Especially on chips.

Speaker Change: That hasn't materialized, yet nor do we know how it would be structured so we don't know, but we do experience and we are leaving it now is the disruption.

Speaker Change: That is.

Speaker Change: In a positive way I should add a very positive disruption.

Hock Tan: in Semiconductors on a Generative A. Workers, Generated AI. for watching. really accelerating the development of semiconductor technology, both process and packaging as well as design. Higher and higher performance accelerators and networking functionality. We've seen that innovation, that those upgrades occur every month as we face new interesting challenges. And when, particularly with XPUs, we're trying, we've been asked... OptiMinds to Frontier Models of... Our partners, our customers, as well as our hyperscale partners. And we, it's a lot of. I mean it's a privilege almost for us to participate in it and try to optimize. And by optimized, I mean you look at an accelerator.

Speaker Change: In semiconductors.

Speaker Change: <unk> generated.

Speaker Change: Generator of AI.

Speaker Change: Sure.

Speaker Change: <unk> also had the resale repeating.

Speaker Change: But we feel it more than ever is.

Speaker Change: Really is celebrating the development of.

The technology both.

Speaker Change: Process.

Speaker Change: And packaging as well as design.

Speaker Change: Towards higher and higher performance.

Speaker Change: Accelerators and networking functionality.

Speaker Change: We're seeing that innovation that.

Speaker Change: Those upgrades.

Speaker Change: All occur.

Speaker Change: Every month as we face new interesting challenges and win and particularly with Xps, Rick trying we've been asked to.

Speaker Change: To optimize.

Speaker Change: Two fronts.

Speaker Change: <unk> frontier models of.

Speaker Change: Our partners, our customers as well as our Hyperscale partners and there is.

Speaker Change: Is a lot of.

Speaker Change: I mean.

Speaker Change: It's a privilege almost for us to be to participate in it and try to optimize.

Speaker Change: Optimize I mean, you look at them accelerate.

Speaker Change: You can look at is simple high level to meant to perform to want to maybe mention not just on one single mentoring which is.

Hock Tan: You can look at it for simple terms, high level, to perform, to want to make the immersion, not just on one single metric, which is... Compute Capacity, how many teraflops? It's more than that. It's also tied to the fact that... This is a distributed computing problem. It's not just the compute capacity of a single... be you or GPU. It's also the network bandwidth. ties itself to the next adjacent XPU. So that has an impact. So you're doing that, you have to balance with that. Then you decide, are you doing training or you're doing pre-feeling, post-training, fine-tuning.

Speaker Change: <unk> capacity, how many teraflops.

Speaker Change: <unk> is also tied to the fact that.

Speaker Change: This is a distributed computing problem.

Speaker Change: It's not just the <unk> the compute capacity of a single <unk>.

B U R. GPU is also the network bandwidth.

Speaker Change: <unk> itself to the next adjacent GP XP or GPU. So that has an impact so you're doing that you will have to balance that.

Speaker Change: You decide how you're doing training or youre doing pretty feeling.

Speaker Change: Post training fine tuning and again then comes how much memory do you balance against that and Richard how much latency you can afford we just memory bandwidth. So you look at at least four variables maybe even five.

Hock Tan: And again, then comes how much memory do you balance against that? And with it, how much latency you can afford, which is memory bandwidth. So you look at at least four variables, maybe even five, if you include in memory bandwidth, not just memory capacity, when you go straight to inference. So we have all these variables to play with, and we try to optimize it. So all this is very, very, I mean, it's a great experience for our engineers to push their envelope on how to create all those chips. And so that's the biggest disruption we see right now from sheer trying to create.

Speaker Change: In crude in memory bandwidth not just memory capacity when you go straight to inference.

Speaker Change: All these variables to play with and.

Speaker Change: We try to optimize it so all of this is very very.

Speaker Change: I mean, it's a great experience for our engineers to push the envelope on how to create all those chips and so that's the biggest disruption we see right now from sheer trying to create.

Hock Tan: and Push the Envelope on Generative AI, trying to create the best hardware infrastructure to run it. Beyond that, yeah, there are other things, too, that come into play, because with AI, as I indicated, it does not just drive hardware for enterprises, it drives the way they architect their data centers. You know, data requirement, keeping data private and under control becomes important, so suddenly push of workloads towards public cloud. may take a little pause as The large enterprises, particularly, have to take... to recognize that you want to run AI workloads. You probably think very hard about running them on press.

Speaker Change: And push the envelope on generative trying to create a best hardware infrastructure to run it.

Speaker Change: Beyond that yes, there are there other things that come into play because we are as I indicated.

Speaker Change: Not just drive hardware for enterprises it.

Speaker Change: <unk> the way the architect.

Speaker Change: The data centers.

Speaker Change: Data requirement.

Speaker Change: Keeping data private under control becomes important.

Speaker Change: Suddenly.

Speaker Change: The push of workloads.

Speaker Change: Public cloud may.

Speaker Change: It may take a little pause.

Speaker Change: Yes.

Speaker Change: Large enterprises, particularly after than to have to take.

Speaker Change: To recognize that.

Speaker Change: Wanted to run.

AI workloads.

Speaker Change: <unk> think very hard about running them on Prem.

Hock Tan: And suddenly, you push yourself towards saying, you've got to upgrade your own data set. to hangar. and hence my comments on VMware Private AI Foundation. This is true, especially enterprises pushing direction are quickly recognizing that how, where do they run their AI workloads. So those are. Transcripts provided by Transcription Outsourcing, LLC. Sensitive Rules on Sovereignty in Cloud and Data. On as far as you mentioning tariffs is concerned, I think that's too early for us to figure out where you are.

Speaker Change: And suddenly push yourself towards saying you've got to upgrade your own data centers to do.

Speaker Change: Manage that your own data to run it on Prem.

Speaker Change: And that's also pushing a trend that.

Speaker Change: We have been seeing now over the past.

Speaker Change: Two one months.

Speaker Change: Hence my comments on Vmware private AI Foundation.

Speaker Change: It's too, especially enterprises pushing direction quickly recognizing that.

Speaker Change: Where do they run the AI workloads. So those are.

Speaker Change: Trends, we see today and a lot of it coming out of AI a lot of it coming out of.

Speaker Change: Sensitive rules on sovereignty.

Speaker Change: Cloud and.

Speaker Change: David.

Speaker Change: As far as you mentioned in tariffs is concerned I think that's too early for us to figure out a way to online and probably maybe give it another three to six months, we'll probably have a better idea of where to go.

Unknown Executive: and probably maybe give it another three, six months, we'll probably have a better idea where to go. Thank you. One moment for our next question.

Speaker Change: Thank you.

Thank you our next question.

Ross Seymour: And that will come from the line of Ross Seymour with Deutsche Bank. Your line is open. Thanks for the answer to the question.

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

Speaker Change: Yes. Thanks for let me ask the question Hock I wanted to go back to the SBU side of things and going from the fourth.

Ross Seymour: Hock, I want to go back to the XTU side of things and going from the four new engagements, not yet named customers, two last quarter and two more today that you announced, I want to talk about going from kind of design win to deployment. How do you judge that? Because there is some debate about, you know, tons of design wins, but the deployments actually don't happen, either that they never occur or that the volume is never what is originally promised. How do you view that kind of conversion ratio? Is there a wide range around it?

Speaker Change: Our new engagements not yet named customers two last quarter and two more today that you announced I wanted to talk about going from kind of design win to deployment.

Speaker Change: How do you judge that because there is some debate about tons of design wins, but the deployment is actually don't happen either that they never occur or that the volume.

Speaker Change: Is never what is originally promised how do you view that kind of conversion ratio is there a wide range around it or is there. Some way you could help us kind of understand how that works.

Hock Tan: Or is there some way you could help us kind of understand how that works? Well, Ross, that's an interesting question. I'll take the opportunity to say the way we look at design is probably very different from the way many of our peers look at it out there. Number one, to begin with, we believe design when we know our product is produced in scale, at scale and is actually deployed. Literary Deployment in Production So that takes a long lead time because from taping out Getting in the product, it takes a year, easily, from the product in the hand.

Speaker Change: Well each.

Speaker Change: The interesting question I'll take the opportunity to say.

Speaker Change: The way, we look at design rates, probably very different from the way many of our peers look at these other out there number one to begin with.

Speaker Change: We believe design win when we know our product is.

Speaker Change: In producing scale at scale and is actually deployed.

Speaker Change: Literally deploying in production so that takes a long lead time because from taping out.

Speaker Change: Getting into product it takes a year easily from the product in the hands.

Hock Tan: of our partner to when it goes into scale production it will take six months to a year is our experience that we're number one and number two. I mean, producing and deploying 5,000 XPUs, that's a joke, that's not real production. And so we also limit ourselves in selecting partners. to people who really need that large volume. You need that large volume from our viewpoint in scale right now in mostly training, training of large language models, frontier models in a continuing trajectory. So we limit ourselves to how many customers or how many potential customers that exist out there, Ross, and we tend to be very selective.

Speaker Change: Of our partner do when it goes into scale production. It would take six months to a year is our experience that we're seeing.

Speaker Change: Number one and number two.

I mean <unk>.

Speaker Change: <unk> seeing in deploying 5000 X peers.

Speaker Change: That's a joke that's not real production.

Speaker Change: Our view and so we also limit ourselves in selecting partners.

Speaker Change: Two people, who really need that large volume.

Speaker Change: Need their launch volumes from our viewpoint in scale right now in mostly framing trimming of launch language models frontier models in the continuing trajectory so.

Speaker Change: So we eliminate ourselves to how many customers.

Speaker Change: Many potential customers that exist out there and we tend to be very selective.

Hock Tan: who we pick to begin with. So when we say design means. It really is at scale. It's not something that starts in six months and die or a year and die again. Basically, it's a selection of customers. It's just the way we run our ASIC business in general for the last 15 years. We pick and choose the customers because we know this guy and we do multi-year roadmaps with these customers because we know these customers are sustainable. I put it bluntly, we don't do it for startups. Thank you.

Speaker Change: Who is big to begin with so when we say design win it.

Speaker Change: It really is at scale is not something that starts in six months and die all year and die again.

Speaker Change: Is the selection of customers. It's just the way we run our ASIC business in general for the last 15 years.

Speaker Change: You can choose the customers because we know discount and we do multiple multi year roadmaps with this customers because we know discussed mers are sustainable.

Speaker Change: I put it bluntly, we don't do it for start ups.

Speaker Change: Okay.

Speaker Change: Thank you.

Unknown Executive: And one moment for our next question.

Speaker Change: And one moment our next question.

Stacy Rasgon: And that will come from the line of Stacy Rasgon with Bernstein Research. Your line is open. Hi, guys. Thanks for taking my question. I wanted to go to the three customers that you do have in volume today. And what I wanted to ask was, is there any concern about some of the new regulations or the AI diffusion rules that are going to get put in place, supposedly in May, impacting any of those design wins or shipments? It sounds like you think all three of those are still on at this point. But anything you could tell us about worries about new regulations or AI diffusion rules, impacting any of those wins would be Thank you.

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

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

Stacy: I wanted to go to the three customers that you do have in volume today.

Stacy: What I wanted to ask was is there any concern about some of the new regulations or the AI diffusion rules that are going to get put in place supposedly in may.

Stacy: <unk> any of those design wins are shipments it sounds like you think.

Stacy: All three of those are still on at this point, but anything you could tell us about.

Stacy: It's about new regulations or AIG fusion was impacting of those wins would be helpful.

Stacy: Thank you in this era or this current era of geopolitical tensions.

Hock Tan: In this era, or this current era of geopolitical tensions, and fairly dramatic actions all around by governments... There's always some concern at the back of everybody's mind. But to answer your question directly, no. We don't have any... All right, so none of those are going into China or to Chinese customers then? No comment. Are you trying to look beautiful there, huh? Okay, that's helpful. Thank you.

Stacy: And fairly dramatic actions all around by governments.

Stacy: Yes, it's always.

Some concern at the back of everybody's mind.

Stacy: But to answer your question directly.

Stacy: Sure.

Stacy: We don't have any concerns.

Stacy: So none of those are going into China or to Chinese customers then.

Stacy: No comment.

Stacy: Yeah.

Stacy: Who they are.

Stacy: Uh huh.

Speaker Change: Okay. That's helpful. Thank you. Thank you one moment our next question.

Unknown Executive: One moment for our next question.

Vivek Arya: And that will come from the line of Vivek Arya with Bank of America. Your line is open. Thanks for taking my question. Hock, whenever you have described your AI opportunity, you have always emphasized the training workload. But the perception is that the AI market could be dominated by the inference workload, especially with these new reasoning models. So what happens to your opportunity and share if the mix moves more towards inference? Does it, you know, does it create a bigger spam for you than the 60 to 90 billion? You know, does it keep it the same, but there's a different mix of products?

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

Vivek Arya: Thanks for taking my question Hock whenever you have described your opportunity you've always emphasized the training workload, but.

Vivek Arya: But the perception is that the market could be dominated by the inference workloads, especially with these new.

Vivek Arya: Reasoning model, so what happens to your opportunity and share if the mix moves more towards insurance does it.

Does it create a bigger Sam for you than the $60 billion to $90 billion does it keep it the same but there is a different mix of product or does it more entrenched heavy market favorite GPU over the next to you.

Vivek Arya: Or does a more inference heavy market favor a GPU over an XPU? Thank you.

Hock Tan: That's a good question, interesting question. By the way, I never, I do talk a lot about training. We do, our XPUs are... also focus on inference. as a separate product line. They do, and that's why I can say the architecture of those chips are very different from the architecture of the training chips. And so it's a combination of those two, I should add, that adds up to this. to 90 billion dollars. So if I have not been clear, I do apologize. It's a combination of both. The larger part of the dollars come from training. non-inference within the service of the same that we have talked about so far.

Speaker Change: That's a good question interesting question by the way I nervous I do talk a lot about training.

Speaker Change: Our chip our XP use us.

Speaker Change: Also focus on inference.

Speaker Change: As a separate product line.

Speaker Change: They do and Thats why I can say the architecture of those chips are very different from the architecture of the training chips.

And then.

Speaker Change: And so it's a combination of those two I should add that adds up to this $60 billion to $90 billion. So if I had not been clear I do apologize, it's a combination of both but having said that the.

Speaker Change: The larger part of the dollars come from training.

Speaker Change: Not inference if within this.

Speaker Change: So.

Speaker Change: The same that we've talked about so far.

Speaker Change: Thank you.

Speaker Change: One moment for our next question.

Unknown Executive: One moment for our next question.

Harsh Kumar: And that will come from the line of Harsh Kumar with Piper Sandler. Your line is open. Thanks, Broadcom team. And again, great execution. Just talk, had a quick question. We've been hearing that almost all of the large clusters that are 100k plus, they're all going to ethernet.

Speaker Change: And that will come from the line of harsh Kumar with Piper Sandler Your line is open.

Thanks, Broadcom team and again great execution.

Speaker Change: Talk kind of quick question, we've been hearing that almost all of the large clusters that are 100, K plus they're all going to Ethernet.

Hock Tan: I was wondering if you could help us understand the importance of when the customer is making a selection, choosing between a guy that has the best switch ASIC, such as you, versus a guy that might have to compute there. Can you talk about what the customer is thinking and what are the final points that they want to hit upon when they make that selection for the NIC card? Okay, I'm sick. No, it's a yeah, it's down to In the case of the hyperscalers, now very much so. very driven by Performance. and its performance. what you're mentioning about on connecting Scaling up and scaling out those AI accelerators, be they XPU or GPU, among hyperscalers.

Speaker Change: I was wondering if you could help us understand the importance of when the customer is making a selection choosing between a guy that has the best switching snakes, such as U verse is a guy that might have the compute there can you talk about what the customers thinking and what are the final points that they wanted to hit upon when they make that selection for the mic.

Speaker Change: Cards.

Speaker Change: Okay.

Speaker Change: Actually.

Speaker Change: No. It's yes, it's down too.

Speaker Change: In the in the case of the cost in the case of the Hyperscale is now very much so it's very driven.

Speaker Change: Fine.

Speaker Change: Performance.

Speaker Change: And its performance.

Speaker Change: What you're mentioning about connecting.

Speaker Change: Connecting.

Speaker Change: Scaling up and scaling out dose AI accelerators, BD SBU, all GPU among hyperscale and in most cases, among those hyperscale as we engage with when it comes to connecting.

Hock Tan: And in most cases among those hyperscalers we engage with when it comes to connecting and Jose Costa are reviewed by the RIPP Presidency Committee. They are very driven by performance. I mean, if you are in a race to really get the best performance out of... your hardware as you train and continue to train your frontier models. That matters more than anything else. So the basic first thing they go for is proven. That's a proven piece of hardware. It's a proven system, subsystem in our case, that makes it work. And in that case, we tend to have a big advantage.

Speaker Change: Those clusters.

Speaker Change: They are very driven by.

Speaker Change: By performance I mean, if we are in a race to really get the best performance.

Speaker Change: Out of.

Speaker Change: <unk>.

Speaker Change: Hardware as you train.

Speaker Change: And continue to train yields.

Speaker Change: Yes.

<unk> models that matters more than anything else. So.

Speaker Change: Basically first thing they go for is proven that's a proven piece.

Speaker Change: Piece of hardware as a proven.

Uh huh.

Speaker Change: He is a proven system sub system in our case that makes it work.

Speaker Change: And in that case, we tend to have a big advantage because.

Hock Tan: I mean, networking RRs, you know, switching and routing RRs for the last 10 years at least. And the fact that it's AI. Just makes it more interesting for our engineers to work. But it's basically based on proven technology and experience in pushing the envelope on going from 800 gigabit per second bandwidth to 1.6 and moving on 3.2, which is exactly why we keep stepping up this rate of investment in coming up with a product where we take the Tomahawk-5, we double the radix to deal with just one hyperscaler because they want high radix to create larger clusters while running.

Speaker Change: Networking or us.

Speaker Change: Switching and routing for the last 10 years at least.

Speaker Change: In fact, there is AI.

Speaker Change: Just makes it more interesting for our engineers to work on.

Speaker Change: But it's basically based on proven technology and experience in pushing that and pushing the envelope ongoing from 800 gigabit per second bandwidth to one six and moving on three two which is exactly why we keep stepping on this.

Speaker Change: Rate of investment in coming up with our products.

Speaker Change: We take our tomo five.

Speaker Change: <unk> doubled the greater mix to deal with just one hyperscale because they won high ratings to create larger clusters, while running.

Hock Tan: Ben Weave that are smaller but that doesn't stop us from moving ahead to the next generation of Tomahawk 6 and I dare say we're even planning Tomahawk 7 and 8 right now and we're speeding up the rate of development and it's all largely for that few guys by the way. So we're making a lot of investment for very few customers, hopefully with very large... served available months. But, if nothing else, that's the Big Best we're playing.

Speaker Change: When we have that are smaller but that doesn't stop us from moving ahead to the next generation of Tomahawk six.

Speaker Change: And I Daresay, we even planning some of seven and eight right now and we're speeding up the rate of development and it's all Luckily for that few guys by the way.

Speaker Change: So we're making a lot of investment for very few customers hopefully with very large.

Speaker Change: So served available markets.

Speaker Change: That's even nothing else, that's a big bet to embracing.

Hock Tan: Thank you, Hock. Thank you.

Hock Tan: Thank you hock.

Thank you one moment, our next question and that will come from the line of Timothy Arcuri with UBS. Your line is open.

Unknown Executive: 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. Um, Hock, in the past, you have mentioned XPU units growing from about 2 million last year to about 7 million, you said, in the 2027-2028 timeframe. My question is, do these four new customers, do they add to that 7 million unit number? I know in the past, you sort of talked about an ASP of, you know, 20 grand by then. So, so, so those three, the first three customers are clearly a subset of that 7 million unit.

Timothy Arcuri: Thanks, a lot hock in the past you have mentioned <unk> units growing from about $2 million last year to about $7 million. You said in the 2027 2028 timeframe. My question is do these four new customers do they had to that 7 million unit number I know in the past you've sort of talked about Asps 'twenty.

Timothy Arcuri: 20 Grand by then so so so those first three customers are clearly a subset of that 7 million units. So do these new four engagements drive that seven higher or do they just fill in to get to that $7 million.

Timothy Arcuri: So do these new four engagements drive that 7 higher? Or do they just fill in to get to that 7 million? Thanks.

Hock Tan: And thanks Tim for asking that. To clarify, as I made, I thought I made it clear in my comments, no, the market we are talking about, including when you translate the units, is only among the three. Customers we have today. The other four, we talk about engagement partners. We don't consider that as customers yet and therefore are not in our served available market. Okay, so they would add to that number. Okay. Thanks, Hock.

Timothy Arcuri: Thanks, Tim for asking that to clarify as I made I thought I made it clear in my comments.

Timothy Arcuri: No.

Timothy Arcuri: The market we are talking about when you translate the unit is only among the three.

Timothy Arcuri: Customers we have today.

Timothy Arcuri: The other four we talk about engagement partners.

Timothy Arcuri: We don't consider that as customers yet.

Timothy Arcuri: And therefore are not in our served available market.

Timothy Arcuri: Okay. So they would add to that number okay. Thanks, a lot. Thanks.

CJ Muse: One moment for our next question. And that will come from the line of CJ Muse with Cancer Fitzgerald. Your line is open. Yeah, good afternoon. Thank you for taking the question. I guess I'll have to follow up on your prepared remarks and comments earlier around optimization with your best hardware and hyperscalers with their great software. I'm curious how you're expanding your portfolio now to six, you know, mega scale kind of frontier models will enable you to, you know, in one blush, you know, share tremendous information, but at the same time, a world where these six truly want to differentiate.

Timothy Arcuri: One moment our next question.

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

Speaker Change: Yes, good afternoon, and thank you for taking the question I guess talk to follow up on your prepared remarks and comments earlier around optimization.

Speaker Change: Your best hardware and hybrid sailors with a great software.

Speaker Change: Curious.

Speaker Change: Now you are expanding your portfolio now to six.

Speaker Change: Mega scale kind of frontier models will enable you to.

Speaker Change: Bush share tremendous information, but at the same time, a world where these six truly want to differentiate so obviously.

Hock Tan: So, obviously, you know, the goal for all of these players is exaflops per second per dollar of capex per watt. And I guess, to what degree are you aiding them in this efforts? And where does maybe the Chinese wall kind of start where they want to kind of differentiate and not share with you kind of some of the work that you're doing? Thank you.

Speaker Change: The goal for all of these players as extra flops per second per dollar of Capex per watt.

Speaker Change: And I guess to what degree are you aiding them in this effort and where it is maybe the Chinese wall kind of start.

Speaker Change: They want to kind of differentiate in and not share with you some of the work that youre doing thank you.

Hock Tan: Oh, you know, we only provide very basic fundamental technology in semiconductors to enable these guys to use what we have and optimize it to their own particular model. and the algorithms that relate to those models. That's it. That's all we do. So that's the level of a lot of that optimization we do for each of them. And as I mentioned earlier, there are maybe five degrees of freedom that we do. And we play with that. And so even if there are five degrees of freedom, there's only so much we can do at that point. But it is, and how they basically, how we optimize it, it's all tied to the partner telling us how they want to do it.

Speaker Change: We only provide.

Speaker Change: Very base.

It's the fundamental technology.

Speaker Change: In semiconductors.

Speaker Change: Enable these guys to use what we have.

Speaker Change: And optimize it to their own particular.

Speaker Change: Model.

Speaker Change: And algorithms that relate to those models.

Speaker Change: All we do so that's the level of a lot of optimization, we do for each of them and as I mentioned earlier.

Speaker Change: Five degrees of freedom that we do and we play with that and so even if the five degrees of freedom freedom is only so much we can do at that point, but it is and how the two and basically how we optimize it is all tied to the partner telling us how they want to do.

Hock Tan: So there's only so much. We also have visibility. But it's what we do now is what the XPU model is. Shear optimization translating to performance, but also It is about power. That is very important how they play. It is not just cost, power can translate into total cost of ownership eventually. It is how designed in power and how we balance it in terms of … the size of the cluster and whether they use it for training. I don't know, pre-training? post-training in France. You know, test time scaling. All of them have their own characteristics. And that's the advantage of doing that XPU and working closely with them to create that stuff.

Speaker Change: So theres only so much we also have visibility on.

Speaker Change: It's but it's what we do know is what the XP model share optimization translating to performance, but also.

Speaker Change: Power.

Speaker Change: That's very important how they play it.

Speaker Change: It's not just cost power.

Speaker Change: Power couple translates into a total cost of ownership eventually.

Speaker Change: It's how design at empower high and how we balance it in terms of.

Speaker Change: The size of the cluster.

Speaker Change: They use it for training.

Speaker Change: No pre training.

Speaker Change: Post training.

Speaker Change: In France.

Speaker Change: Test times scaling all of them have their own characteristics and thats the advantage of doing that SBU and working closely with them to create that stuff now as far as your question on.

Hock Tan: Now, as far as your question on. Cut China and all that. Frankly, I don't have any opinion on that at all. To us, it's a technical game. Thank you very much.

Speaker Change: China and all of that frankly, I don't have any opinion on that at all.

Speaker Change: To us it's a technical game.

Speaker Change: Thank you very much.

Christopher Rolland: One moment for our next question. And that will come from the line of Christopher Rolland with Susquehanna. Your line is open. Hey, thanks so much for the question. And, and this one's maybe for Hock and for Kirsten.

Speaker Change: One moment our next question.

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

Speaker Change: Hey, thanks, so much for the question.

Speaker Change: And.

And this one is maybe for Hawk and for Kirsten.

Hock Tan: I'd love to know just because you have kind of the complete connectivity portfolio, how you see new greenfield scale up opportunities playing out here between, you know, could be optical or copper or really anything and what additive this could be for your company. And then, and then Kirsten, I think OPEX is up. Maybe just talk about where those OPEX dollars are going towards within the AI opportunity and whether they relate. Thanks so much. Your question is very broad-reaching and our portfolio, yeah, we deploy, we have the advantage and a lot of the hyperscale customers we deal with.

Speaker Change: I'd Love to know just because you have kind of a complete connectivity portfolio, how you see new greenfield scale up opportunities playing out here.

Speaker Change: Could we optical or copper or really anything in additive this could be for your company.

Speaker Change: And then and then Kirsten I think Opex is up maybe just talk about where those opex dollars are.

Speaker Change: Our going towards within the AI opportunity and whether they relate thanks so much.

Speaker Change: Okay.

Speaker Change: Uh huh.

Speaker Change: Your question is very broad, reaching and our portfolio yeah, we deploy we have the advantage.

Speaker Change: And a lot of the cost.

Speaker Change: Hyperscale customers we deal with.

Hock Tan: They're talking about a lot of expansion, but it's almost all greenfield. You know, less so brownfield. It's very greenfield, it's all expansion, and it's all tends to be next generation that we do it, which is very exciting. So the opportunity is very, very high. And we deploy, I mean, we're both, we can do it in copper, but what we see a lot of opportunity from is when you provide the networking connectivity through optical. So there are a lot of active elements, including either, you know, multi-mode. lasers, which are called Vixels, or X-ray printing lasers, for basically single mode.

Speaker Change: We're talking about a lot of expansion.

Speaker Change: It's all almost all greenfield.

Speaker Change: Less so brownfield is very greenfield is all expansion and its all times to be next generation that we do it which is very exciting. So the opportunity is very very high and we deploy I.

Speaker Change: I mean, we are both we can do it in copper.

Speaker Change: While we see a lot of opportunity if the opportunity is.

Wayne: This is Wayne.

Wayne: And can they provide the networking connectivity through obstacle, but there are a lot of development, including either multi mode.

Wayne: Lasers, which are called <unk> or edge emitting lasers for us basically single mode, and we do both so theres a lot of opportunity just as in <unk>.

Hock Tan: And we do both. So there's a lot of opportunity, just as in scale up versus scale out. We used to do, we still do, a lot of other protocols beyond Ethernet to consider PCI Express, where we on the leading edge of that PCI Express. and the architecture on networking, switching, so to speak, we offer both. One is a very intelligent switch. and which is like our Jericho family with a dumb nick. or a very smart Nick with a dumb switch which is a tomahawk. We offer both architectures as well. So yeah, we have a lot of opportunities from it.

Wayne: In scale.

Wayne: Scale up versus scale out we.

Wayne: We used to do we still do a lot of us.

Wayne: The protocols beyond Ethernet to consider PCI Express, where we on the leading edge of that PC Express.

Wayne: And that architecture on networking switching so to speak.

Wayne: Offered both one is.

Wayne: A very intelligent switch.

Wayne: And which is why our Jericho family.

Wayne: We've a dumb Nic.

Wayne: All are very smart Nic, we got demonstrates <unk>, which is the tomahawk we offer both.

Wayne: Our architectures as well so yes, we have a lot of opportunities for me all things said and done all this nice wide portfolio and all of that.

Hock Tan: All things said and done, all this nice white portfolio and all that adds up to probably, as I said in prior quarters, about 20% of our total AI revenue, maybe going to 30%. Though last quarter we hit almost 40%, but that's not the norm. I would say typically, all those other portfolio products still add up to a nice, decent amount of revenue for us, but within the sphere of AI, they add up to, I would say on average, be close to 30%, and XPUs, the accelerators, is 75%. If that's what you're driving at, perhaps I can shed some light on how one matters over the other.

Wayne: Up to.

Wayne: Probably.

Wayne: As I said in prior quarters.

Wayne: 20% of our total <unk> revenue, maybe going to 30%, though last quarter, we hit almost 40%, but that's not the norm.

Wayne: I would say typically all those other ports.

Portfolio products still end up toward 90 decent amount of revenue for us, but within the sphere of AI. They add up to I would say on average close to 30% and XP used accelerators is 70%.

Wayne: Thats, what youre driving at perhaps that gives you some shed some light on towards way, how one method over the other but we have a wide range of products in the connectivity networking side of it.

Kirsten Spears: But we have a wide range of products in the connectivity, networking side. They just add up, though, to that 30%.

Wayne: I just add up those 230%.

Speaker Change: Thanks, So much Frank and then on the R&D front.

Kirsten Spears: And then on the R&D front, as I outlined, on a consolidated basis, we spent $1.4 billion in R&D in Q1, and I stated that it would be going up in Q2. Hock clearly outlined in his script the two areas where we're focusing on. Now, I would tell you, as a company, we focus on R&D across all of our product lines so that we can stay competitive with next-generation product offerings, but he did line out that we are focusing on taping out the industry's first two-nanometer AI XPU packaged in 3D. That was one in his script, and that's an area that we're focusing on.

Wayne: As I outlined.

Wayne: Holiday the basis, we spent $1 4 billion in R&D in Q1, and I've stated that it would be going up in Q2, clearly outlined in his script. The two areas, where we're focusing on I would tell you that the company we focus on R&D across all of our product lines. So that we can stay competitive with next generation product offerings, but he did line out that.

Wayne: We are focusing on taping out the industry's first two nanometer aix's du packaged in <unk> I was one in his script and that's an area that we're focusing on and then you mentioned that we've doubled the ratings capacity of existing Tomahawk fives.

Kirsten Spears: And then he mentioned that we doubled the RADIX capacity of existing Tomahawk 5s to enable our AI customers to scale up on Ethernet towards the one million XPUs, so I mean, that's a huge focus of the company. Yeah, thank you very much, Kirsten.

Wayne: To enable our customers to scale up on Ethernet towards the $1 million extra use naphtha huge focus on the company.

Wayne: Yes, thank you very much.

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. Hi, Hock. Thanks. Just a quick question on the networking side. Just wondering how much it was up sequentially on the AI side. And any thoughts on M&A going forward? There's been a lot of headlines around the Intel products group, etc. So thanks. Okay, on the networking side, it's indicated Q1 short bit of a surge, but I don't expect that to be that mix of 60-40, 60-80. Compute and 40% networking to be something that Normal.

Wayne: And one moment our next question.

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

Speaker Change: Yes, hi.

Vijay Rakesh: Thanks. Thanks, So just a quick question on the networking side, just wondering how much it was up sequentially on the AI side and any thoughts on M&A going forward.

Speaker Change: It's been a lot of headlines around.

Vijay Rakesh: Okay. Thanks.

Vijay Rakesh: Thanks.

Vijay Rakesh: Okay on the networking side as indicated Q1 short.

Vijay Rakesh: A bit of a search, but I don't expect that to be.

Vijay Rakesh: That makes of 60 40, 60 is it a compute and 40% net we're going to be something that is normal I think the Nam it's closer to 70 30, maybe at best 30%.

Hock Tan: I think the norm is closer to 70-30 maybe, at best 30%. And so, who knows what Q2 is. We kind of see Q2 as continuing, but that's just, in my mind, a temporary blip. The norm will be 70-30, and if you take it across a period of time like six months, a year. That's your question.

Vijay Rakesh: And so who knows what Q2 is we cannot see Q2 as continuing but that's just my mind a temporary blip.

Vijay Rakesh: The Nam will be 70, 30, if you take it across a period of time like six months a year.

Vijay Rakesh: Thanks for your question.

Hock Tan: M&A? No, I'm too busy. We're too busy doing AI and VMware at this point. We're not thinking of it at this point. Thank you.

Speaker Change: M&A no I'm too busy to busy doing AI and <unk> at this point.

Vijay Rakesh: We're not thinking of it.

Speaker Change: At this point.

Vijay Rakesh: Thanks Mark.

Thank you that is all the time, we have for a question and answer session I would now like to turn the call back over to you for any closing remarks. Thank you sorry, Broadcom currently plans to reported earnings for the second quarter of fiscal year 2025, after close of market on Thursday June five 2025.

Unknown Executive: That is all the time we have for our question and answer session.

Ji Yoo: I would now like to turn the call back over to Ji Yoo for any closing remarks. Thank you, Cherie. Broadcom currently plans to report its earnings for the second quarter of fiscal year 2025. After close of market on Thursday, June 5 2025. A public webcast of Broadcom's earnings conference call will follow at 2pm Pacific.

Vijay Rakesh: Our public webcast at Broadcom as earnings Conference call will follow at two P M Pacific that.

Unknown Executive: That will conclude our earnings call today. Thank you all for joining. Shree, you may end the call. Ladies and gentlemen, thank you for participating.

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

Speaker Change: Ladies and gentlemen, thank you for participating. This concludes today's program you may now disconnect.

Unknown Executive: This concludes today's program. You may now disconnect. Welcome to the Broadcom Inc.'s first quarter fiscal year 2025 financial results conference call.

Speaker Change: Okay.

Speaker Change: [music].

Speaker Change: Okay.

Speaker Change: [music].

Speaker Change: Okay.

Speaker Change: [music].

Speaker Change: Okay.

Speaker Change: Yes.

Speaker Change: [music].

Speaker Change: Okay.

Speaker Change: [music].

Speaker Change: Yes.

Speaker Change: [music].

Speaker Change: Okay.

Okay.

Speaker Change: [music].

Speaker Change: Okay.

Speaker Change: [music].

Speaker Change: Welcome to the Broadcom, Inc. 's first quarter fiscal year 2025 financial results conference call. 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.

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, Sheree. 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 Group. Broadcom distributed a press release and financial tables after the market closed, describing our financial performance for the first quarter of fiscal year 2025. If you did not receive a copy, you may obtain the information from the Investor section of Broadcom's website at Broadcom.com.

Speaker Change: Thank you Sherry and good afternoon, everyone. Joining me on today's call are Hock Tan President and CEO.

Speaker Change: <unk> Spirit's, Chief Financial Officer, and Charlie Clark President Semiconductor solutions group.

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

Speaker Change: If you did not receive a copy you may obtain the information from the investors section of <unk> 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 first quarter fiscal year 2025 results. Guidance for our second quarter of fiscal year 2025, as well as commentary regarding the business environment. We will take questions after the end of our prepared comments. 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: This conference call is being webcast live and then audio replay of the call can be accessed for one year.

Speaker Change: The investors section of Broadcom website.

Speaker Change: During the prepared comments Hawkins Kirsten will be providing details of our first quarter fiscal year 2025 result.

Speaker Change: <unk> for our second quarter of fiscal year, 2025, as well as commentary regarding the business environment. We.

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.

Ji Yoo: in addition to U.S. Gap reporting. Broadcom reports certain financial measures on a non-gap basis. A reconciliation between GAAP and non-GAAP measures is included in the tables attached to today's press release. The comments made during today's call will primarily refer to our non-GAAP financial results.

Speaker Change: In addition to U S GAAP reporting.

Speaker Change: <unk> report certain financial measures on a non-GAAP basis.

Speaker Change: A reconciliation between GAAP and non-GAAP measures is included in the tables attached to todays press release.

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

Hock Tan: I'll now turn the call over to Hock. Thank you, Ji. and thank you everyone for joining today. in a fiscal Q1 2025. Total revenue was a record $14.9 billion. Up 25% year-on-year. and Consolidated Adjusted EBITDA was a record again, $10.1 billion, up 41% year-on-year.

Håkan: I'll now turn the call over to Hakan.

Thank you Jay.

Speaker Change: And thank you everyone for joining today.

Speaker Change: In our fiscal Q1 2025.

Speaker Change: Total revenue was a record $14 $9 billion up 25% year on year.

Speaker Change: And consolidated adjusted EBITDA was a record gain $10 1 billion.

Speaker Change: Up 41% year on year.

Hock Tan: So let me first provide color on our semiconductor business. Q1 Semiconductor Revenue was $8.2 billion, up 11% year-on-year. Growth was driven by AI, as AI revenue of $4.1 billion was up 77% year-on-year. We bid our guidance for AI revenue of $3.8 billion. Due to stronger shipments of networking solutions to hyperscalers on AI. Our hyperscale partners. continue to invest aggressively in the next generation frontier model. which do require high performance accelerators. as well as AI data centers with larger clusters.

Speaker Change: So let me first provide color on our semiconductor business.

Speaker Change: Q1 semiconductor revenue was $8 2 billion up 11% year on year.

Speaker Change: Growth was driven by AI.

Speaker Change: <unk> revenue of $4 1 billion was up 77% year on year.

Speaker Change: We beat our guidance for AI revenue of $3 8 billion.

Speaker Change: Due to stronger shipments of networking solutions to Hyperscale is on AI.

Speaker Change: Our Hyperscale partners Couldnt.

Speaker Change: Continue to invest aggressively in the next generation frontier models.

Speaker Change: We still require higher performance accelerators.

Speaker Change: As well as AI data centers with larger clusters.

Hock Tan: Consistent with the We are stepping up our R&D investment on two fronts. One, we're pushing the envelope of technology in creating the next generation of accelerators. We're taping out the industry's first 2-nanometer AI XPU package in 3.5D as we drive towards 10,000 Teraflops. Secondly, we have a view towards scaling clusters of 500,000 accelerators for hyperscale clusters. We have doubled the radix capacity of the existing Tomahawk flight. Beyond to enable AI clusters to scale up on the Internet towards 1 million XPUs. We have taped out our next generation 100 terabit Tomahawk 6 switch running 200G studies at 1.6 terabit bandwidth.

Speaker Change: And consistent with this.

Speaker Change: We are stepping up our R&D investment on two fronts.

Speaker Change: One we're pushing the envelope of technology in creating the next generation of accelerators.

Speaker Change: With taping out the industry's first two nanometer AI XP you packaging tree three.

Speaker Change: Three five D as we drive towards a 10.

Speaker Change: Teraflops SP X Bu.

Speaker Change: Secondly, we have a view towards scaling clusters of 500000 accelerators for a hyperscale customer.

Speaker Change: We have doubled the reach.

Speaker Change: <unk> capacity of this of the existing amongst <unk>.

Speaker Change: And beyond this.

Speaker Change: To enable AI clusters to scale up on Ethernet towards 1 million next be use.

Speaker Change: We have to tape out our next generation 100 therapy Tomo six switch running 200 G studies and one six therapy therapy.

Speaker Change: Bandwidth.

Hock Tan: We will be delivering samples to customers within the next few months. These R&D investments are very aligned with the roadmap of our three hyperscale customers. as they each race towards 1 million XPU clusters by the end of 2020. And accordingly, we do reaffirm what we said last quarter, that we expect these three hyperscale customers will generate a serviceable, addressable market of SEM in the range of 60 to $90 billion in fiscal 2020. Beyond these three customers, we had also mentioned previous... that we are deeply engaged with two other hyperscalers. enabling them to create their own customized AI accelerators.

Speaker Change: We will be delivering samples to customers within the next few months.

Speaker Change: These R&D investments are very in line with the road map of our three hyperscale customers as they each race towards 1 million SBU clusters by the end of 2027.

Speaker Change: And accordingly.

Speaker Change: We do reaffirm what we said last quarter.

Speaker Change: We expect these three hyperscale customers would generate a serviceable addressable market our Sam in the range of $60 billion to $90 billion in fiscal 2027.

Speaker Change: Beyond these three customers. We had also mentioned previously that we are deeply engaged with two other hyperscale is.

Speaker Change: Enabling them to create their own customize AI accelerators.

Hock Tan: We on track. take out their XPUs this year. In the process of working with the hyperscalers It has become very clear that while they are excellent in software, Broadcom is the best in hub. Working together is what optimizes. We are a large language mall. It is therefore no surprise to us. Since our last earnings call that two additional hyperscalers have selected Broadcom to develop custom accelerators to train their next generation frontier models. So even as we have three hyperscale customers, we are shipping XPUs. volume today, there are now four more who are deeply engaged with us to create their own accelerator.

Speaker Change: We are on track to tape out.

Speaker Change: SP use.

Speaker Change: This year.

Speaker Change: In the process of working with the Hyperscale is.

It has become very clear.

Speaker Change: While they're excellent in software.

Speaker Change: Broadcom is the best in hardware.

Speaker Change: Working together is what optimizes.

Speaker Change: We are launch language models.

Speaker Change: It is therefore, no surprise to us.

Speaker Change: Since our last earnings call.

Speaker Change: The two additional Hyperscale is have selected broadcom to develop custom accelerators to train the next generation frontier models.

Speaker Change: So even as we have three hyperscale customers, we are shipping sbu's in volume today.

Speaker Change: Now for more deeply engaged with us to create their own accelerators.

Hock Tan: And to be clear, of course, these four are not included in our estimated SAM of $60-90 billion in 2020.

Speaker Change: And to be clear cause this fall.

Speaker Change: Not included in our estimate the same of $60 billion to $90 billion.

Speaker Change: In 2027.

Hock Tan: So we do see an exciting trend. New Frontier Models and Techniques. put unexpected pressures on AI systems. It's difficult to serve all classes of models with a single system design point. And therefore, it is hard to imagine that a general purpose accelerator can be configured and optimized across multiple frontier models. And as I mentioned before, the trend towards XPU. is a multi-year journey. So coming back to 2025. We see a steady ramp in deployment of our XPUs and networking products. Q1 AI revenue was $4.1 billion and we expect Q2 AI revenue to grow to $4.4 billion.

Speaker Change: So we do see an exciting trend.

Speaker Change: New frontier models.

Speaker Change: Thanks, Paul.

Speaker Change: Unexpected pressures on <unk>.

Pressures on AI systems.

Speaker Change: It's difficult to serve all classes of models with us.

Speaker Change: Single system design point.

Speaker Change: And therefore, it is hard to imagine that a general purpose accelerator.

Speaker Change: Can be configured.

Speaker Change: And optimize across multiple frontier models.

Speaker Change: And as I mentioned before the trend towards XP use.

Speaker Change: Is a multiyear journey.

Speaker Change: Coming back to 2025.

Speaker Change: We see a steady ramp in deployment of our XP use in networking products.

Speaker Change: Q1 revenue was $4 1 billion and we expect Q2 revenue to grow to four 4 billion.

Hock Tan: which is up 44% year-on-year.

Speaker Change: It's up 44% year on year.

Hock Tan: Turning to non-AI semiconductors. Revenue of $4.1 billion was down 9% sequentially on a seasonal decline in wildlife. In aggregate, during Q1, the recovery in non-AI semiconductors continued to be slow. Broadband, which bought them in Q4 2024, showed a double digit sequential recovery in Q1. and is expected to be up similarly in Q2 as service providers and telcos step up spending. Server storage was down single digits sequentially in Q1. but it's expected to be up high single digits sequentially in Q2. Meanwhile, enterprise networking continues to remain flattish in the first half of Fiscal 25 as customers continue to work through channel inventory.

Speaker Change: Turning to non AI semiconductors revenue.

Speaker Change: Revenue of $4 1 billion was down 9% sequentially.

Speaker Change: The seasonal decline in wireless.

Speaker Change: In aggregate during Q1, the recovery in non AI semiconductors.

Speaker Change: <unk> to be slow.

Speaker Change: Broadband, which bottom in Q4 2024 showed a double digit sequential recovery in Q1.

Speaker Change: <unk> is expected to be up similarly in Q2.

Speaker Change: Service providers and telcos step up spending.

Server storage was down single digit sequentially in Q1.

Speaker Change: But it's expected to be up high single digits sequentially in Q2.

Speaker Change: Meanwhile, enterprise networking.

Speaker Change: The news to remain flattish in the first half of fiscal 'twenty five.

Speaker Change: As customers continue to work through channel inventory.

Hock Tan: One wireless was down sequentially due to a seasonal decline. Remain flat year on year. Q2, Wireless is expected to be the same. flat again year on year. Resales in industrial were down double digits in Q1 and are expected to be down in Q2. So reflecting the foregoing puts and takes. We expect non-AI semiconductor revenue in Q2 to be flattish sequentially. Even though we are seeing bookings continue to grow year on year.

Speaker Change: One wireless was down sequentially due to a seasonal decline.

Speaker Change: <unk> remained flat year on year.

Speaker Change: In Q2 wireless is expected to be the same.

Speaker Change: Slide again year on year.

Speaker Change: Resales and industrial were down double digits in Q1.

Speaker Change: And I expect it to be down in Q2.

Speaker Change: So, reflecting the foregoing puts and takes.

Speaker Change: We expect non AI semiconductor revenue in Q2 to be flattish sequentially.

Even though we are seeing bookings continue to grow year on year.

Speaker Change: Yes.

Hock Tan: In summary for Q2. We expect total semiconductor revenue to grow 2% sequentially and up 17% year-on-year to $8.4 billion.

Speaker Change: In summary for Q2.

Speaker Change: We expect total semiconductor revenue to grow 2% sequentially and up 17% year on year to $8 4 billion.

Hock Tan: Turning now to Infrastructure Software, the second. Q1 infrastructure software revenue of $6.7 billion was up 47% year-on-year and up 15% sequentially. Exaggerated, though, by deals which slip from Q4 to Q1. Now this is the first quarter, Q1 25, where the year-on-year comparables include VMware in both quarters. We're seeing significant growth in the software segment for two reasons. One, we're converting to a footprint of large, we're, sorry, we're converting. from a footprint of largely perpetualized. to one of full subscription. And as of today, we are over 60% done. Two. These perpetual licenses were only largely for compute virtualization.

Speaker Change: Turning now to infrastructure software.

Speaker Change: The segment.

Speaker Change: Q1 infrastructure software revenue of $6 $7 billion was up 47% year on year and up 15% sequentially.

Speaker Change: As they generated though by deals which slipped from Q to Q4 Q1.

Speaker Change: Now this is the first quarter Q1, 'twenty five where the year on year Comparables include Vmware.

Speaker Change: In both quarters.

Speaker Change: We're seeing sickness significant growth in our software segment for two reasons.

Speaker Change: One we're converting to a footprint of launch.

We have come sorry, we're converting.

Speaker Change: From a footprint of largely perpetual license.

Speaker Change: Two one off full subscription.

Speaker Change: And as of today.

Speaker Change: Over 60%.

Speaker Change: Two.

Speaker Change: This perpetual licenses.

Speaker Change: <unk> largely for compute virtualization.

Hock Tan: otherwise called The Sphere. We are. upselling customers to a full stack. VCS. which enables the entire data center to be virtualized. This enables customers to create their own private cloud environment on-prem. and that's the end of Q1. Approximately 70% of our largest 10,000 customers. have adopted. And these customers consume VCF. with to see a further opportunity for future growth. and large enterprises that don't AI. They have to run their AI workload. on their on-prem data center. which will include both GPU servers as well as traditional And just as VCF virtualizes these traditional data centers using CPUs.

Speaker Change: Why it's called the sphere.

Speaker Change: We are.

Speaker Change: Upselling customers to a full stack.

Speaker Change: Which enables the entire data centers to be Virtualized.

Speaker Change: And this enables customers to create their own private cloud environment on Prem.

Speaker Change: And then <unk>.

Speaker Change: The end of Q1.

Speaker Change: Approximately 70% of our launches 10000, new customers.

Speaker Change: <unk> adopted vcs.

Speaker Change: As these customers consume vcs.

Speaker Change: We still see a further opportunity for future growth.

Speaker Change: As large enterprises adopt AI.

Speaker Change: They have to run the AI workloads.

Speaker Change: On their on Prem data centers, which will include both GPU servers as well as traditional Cpus.

Speaker Change: And just as Vcs Virtualized us these traditional datacenters using Cpus.

Hock Tan: VCF will also virtualize GPUs on a common platform and enable enterprises to import AI models to run their own data on press. This platform, which virtualized the GPU, is called the VMware Private AI Foundation. And as of today, in collaboration with NVIDIA, we have 39 enterprise customers for the VMware Private AI Foundation. Customer demand has been driven by our open ecosystem. Superior Load Balancing and Automation Capabilities. that allows them to intelligently pull and run workloads across both GPU and CPU infrastructure and leading to very reduced costs.

Speaker Change: This year, we'll also virtualized.

Speaker Change: <unk>.

Speaker Change: On the on a common platform and enable enterprises to important AI models to run their own data on Prem.

Speaker Change: This platform, which virtualized. The GPU is called the Vmware private AI Foundation and as of today in collaboration with Nvidia we.

We have 39 enterprise customers for the Vmware private AI Foundation.

Speaker Change: Customer demand has been driven by our open ecosystem superior load balancing and automation capabilities.

Speaker Change: That allows them to intelligently pool and run workloads across both GPU and CPU infrastructure and leading to very reduce costs.

Hock Tan: Moving on to Q2 Outlook for software. We expect revenue of $6.5 billion. Up 23% year-on-year. So in total, we're getting Q2 consolidated revenue to be approximately $14.9 billion. up 19% year-on-year. This, we expect, this will drive Q2 adjusted EBITDA to approximately...

Speaker Change: Moving on to Q2 outlook for software.

Speaker Change: We expect revenue of $6 5 billion.

Speaker Change: Up 23% year on year.

Speaker Change: So in total we're guiding Q2 consolidated revenue to be approximately $14 $9 billion.

Speaker Change: 19% year on year end.

Speaker Change: And we expect this will drive Q2, adjusted EBITDA to approximately 66% of revenue.

Kirsten Spears: With that, let me turn the call over to Kirsten. Thank you, Hock.

Speaker Change: With that let me turn the call over to kiss them.

Speaker Change: [laughter].

Speaker Change: Thank you Hock, let me now provide additional detail on our Q1 financial performance.

Kirsten Spears: Let me now provide additional detail on our Q1 financial performance. From a year on year comparable basis, keep in mind that Q1 of fiscal 2024 was a 14 week quarter and Q1 of fiscal 2025 is a 13 week quarter. Consolidated revenue was $14.9 billion for the quarter, up 25% from a year ago. Gross margin was 79.1% of revenue in the quarter, better than we originally guided on higher infrastructure software revenue and more favorable semiconductor revenue mix. Consolidated operating expenses were $2 billion of which $1.4 billion was for R&D. Q1 operating income of $9.8 billion was up 44% from a year ago, with operating margin at 66% of revenue.

Speaker Change: On a year on year comparable basis keep in mind that Q1 of fiscal 2024 was a 14 week quarter in Q1 of fiscal 2025 is a 13 week quarter.

Speaker Change: Consolidated revenue was $14 9 billion for the quarter up 25% from a year ago.

Speaker Change: Gross margin was 79, 1% of revenue in the quarter better than we originally guided on higher infrastructure software revenue and more favorable semiconductor revenue mix.

Speaker Change: Consolidated operating expenses were $2 billion of which one 4 billion was for R&D.

Speaker Change: Q1 operating income of $9 8 billion was up 44% from a year ago with operating margin at 66% of revenue.

Kirsten Spears: Adjusted EBITDA was a record $10.1 billion, or 68% of revenue, above our guidance of 66%. This figure excludes $142 million of depreciation.

Speaker Change: Adjusted EBITDA was a record $10 1 billion or 68% of revenue above our guidance of 66%. This figure excludes 142 million of depreciation.

Kirsten Spears: Now a review of the PNL for our two segments. starting with Semiconductor. Revenue for our Semiconductor Solutions segment was $8.2 billion and represented 55% of total revenue in the quarter. This was up 11% year on year. Gross Margin for our Semiconductor Solutions segment was approximately 68%, up 70 basis points year on year, driven by revenue mix. Operating expenses increased 3% year-on-year to $890 million on increased investment in R&D for leading-edge AI semiconductors, resulting in semiconductor operating margin of 57%.

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

Starting with semiconductors.

Speaker Change: Revenue for our semiconductor solutions segment was $8 2 billion and represented 55% of total revenue in the quarter. This was up 11% year on year.

Speaker Change: Gross margin for our semiconductor solutions segment with approximately 68% up 70 basis points year on year driven by revenue mix.

Speaker Change: Operating expenses increased 3% year on year to 890 million on increased investment in R&D for leading edge AI semiconductors, resulting in semiconductor operating margin of 57%.

Kirsten Spears: Now moving on to infrastructure software. Revenue for infrastructure software of $6.7 billion was 45% of total revenue and up 47% year-on-year based primarily on increased revenue from VMware. Gross margin for infrastructure software was 92.5% in the quarter, compared to 88% a year ago. Operating expenses were approximately $1.1 billion in the quarter, resulting in infrastructure software operating margin of 76%. This compares to operating margin of 59% a year ago. This year on your improvement reflects our disciplined integration of VMware and sharp focus on deploying our VCF strategy.

Speaker Change: Now moving on to infrastructure software revenue.

Speaker Change: Revenue for infrastructure software up $6 7 billion was 45% of total revenue and up 47% year on year based primarily on increased revenue from Vmware.

Speaker Change: Gross margin for infrastructure software was 92, 5% in the quarter compared to 88% a year ago.

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

Speaker Change: This compares to operating margin of 59% a year ago.

Speaker Change: This year on year improvement reflects our disciplined integration of Vmware and sharp focus on deploying our vcs strategy.

Kirsten Spears: Moving on to cash flow. Free cash flow in the quarter was $6 billion and represented 40% of revenue. Free cash flow as a percentage of revenue continues to be impacted by cash interest expense from debt related to the VMware acquisition and cash taxes due to the mix of U.S. taxable income, the continued delay in the reenactment of Section 174, and the impact of corporate AMT. We spent $100 million on capital expenditures. Day sales outstanding were 30 days in the first quarter compared to 41 days a year ago. We ended the first quarter with inventory of $1.9 billion, up 8% sequentially, to support revenue in future quarters.

Speaker Change: Moving on to cash flow free cash flow in the quarter was 6 billion and represented 40% of revenue free cash flow as a percentage of revenue continues to be impacted by cash interest expense from debt related to the Vmware acquisition and cash taxes due to the mix of U S. Taxable income the continued delay.

Speaker Change: And the re enactment of section 174, and the impact of corporate AAM team.

Speaker Change: We spent 100 million on capital expenditures.

Speaker Change: Days sales outstanding were 30 days in the first quarter compared to 41 days a year ago. We ended the first quarter with inventory of $1 9 billion up 8% sequentially to support revenue in future quarters.

Kirsten Spears: Our days of inventory on hand were 65 days in Q1, as we continue to remain disciplined on how we manage inventory across the ecosystem. We ended the first quarter with $9.3 billion of cash and $68.8 billion of gross principal debt. During the quarter, we repaid $495 million of fixed rate debt and $7.6 billion of floating rate debt with new senior notes, commercial paper, and cash on hand, reducing debt by a net $1.1 billion. Following these actions, the weighted average coupon rate and years to maturity of our $58.8 billion in fixed rate debt is 3.8% and 7.3 years respectively.

Speaker Change: Our days of inventory on hand were 65 days in Q1 as we continue to remain disciplined on how we manage inventory across the ecosystem.

Speaker Change: We ended the first quarter was $9 3 billion of cash and $68 8 billion of gross principal debt.

Speaker Change: During the quarter, we repaid $495 million of fixed rate debt and $7 6 billion of floating rate debt with new senior note commercial paper and cash on hand, reducing debt by a net $1 1 billion.

Speaker Change: Following these actions the weighted average coupon rate in years to maturity of our $58 8 billion in fixed rate debt is three 8% and seven three years respectively.

Kirsten Spears: The weighted average coupon rate and years to maturity of our $6 billion in floating rate debt is 5.4% and 3.8 years respectively.

Speaker Change: The weighted average coupon rate in years to maturity of our $6 billion in floating rate debt is five 4% and three eight years, respectively, and our $4 billion of commercial paper is at an average rate of four 6%.

Kirsten Spears: And our $4 billion in commercial paper is at an average rate of $4.6 Turning to Capital Allocations. In Q1, we paid stockholders $2.8 billion of cash dividends based on a quarterly common stock cash dividend of $0.59 per share. We spent $2 billion to repurchase 8.7 million ABGO shares from employees as those shares vested for withholding tax. In Q2, we expect the non-GAAP diluted share count to be approximately 4.95 billion shares.

Speaker Change: Turning to capital allocation.

Speaker Change: Q1, we paid stockholders $2 8 billion of cash dividends based on our quarterly common stock cash dividend of <unk> 59 per share. We spent $2 billion to repurchase eight 7 million <unk> shares from employees as those shares vested for withholding taxes.

Speaker Change: In Q2, we expect our non-GAAP diluted share count to be approximately $4 95 billion shares now.

Kirsten Spears: Now moving on to guidance. Our guidance for Q2 is for consolidated revenue of $14.9 billion, with semiconductor revenue of approximately $8.4 billion, up 17% year-on-year. We expect Q2 AI revenue of $4.4 billion, up 44% year-on-year. For non-AI semiconductors, we expect Q2 revenue of $4 billion. We expect Q2 infrastructure software revenue of approximately $6.5 billion, up 23% year-on-year. We expect Q2 Adjusted EBITDA to be about 66%. For modeling purposes, we expect Q2 Consolidated Growth Margin to be down approximately 20 basis points sequentially on the revenue mix of infrastructure, software, and product mix within semiconductors. As Hock discussed earlier, we are increasing our R&D investment in leading-edge AI in Q2, and accordingly, we expect adjusted EBITDA to be approximately 66%.

Speaker Change: Now moving onto guidance our guidance for Q2 is for consolidated revenue of $14 9 billion with semiconductor revenue of approximately $8 4 billion up 17% year on year.

Speaker Change: We expect Q2, AI revenue of $4 4 billion up 44% year on year.

Speaker Change: Fernand AI semiconductors, we expect Q2 revenue of $4 billion.

Speaker Change: We expect Q2 infrastructure software revenue of approximately $6 5 billion up 23% year on year.

Speaker Change: We expect Q2, adjusted EBITDA to be about 66%.

Speaker Change: For modeling purposes, we expect Q2 consolidated gross margin to be down approximately 20 basis points sequentially on the revenue mix up infrastructure software and product mix within semiconductors.

Speaker Change: As Hock discussed earlier, we are increasing our R&D investment in leading edge AI in Q2, and accordingly, we expect adjusted EBITDA to be approximately 66%.

Kirsten Spears: We expect the non-GAAP tax rate for Q2 and fiscal year 2025 to be approximately 14%.

Speaker Change: We expect the non-GAAP tax rate for Q2 and fiscal year 2025 to be approximately 14%.

Unknown Executive: 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, 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: 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 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: Okay.

Benjamin Reitzes: And our first question will come from the line of Ben Reitzes with Mellius. Your line is open. Hey, guys, thanks a lot. And congrats on the results. Hock, you talked about four more customers coming online. Can you just talk a little bit more about, you know, the trend you're seeing? Can any of these customers be as big as the current three? And what what does this say about the custom silicon trend overall, and your optimism and upside to the business long term? Thanks. Well, very interesting question, Ben, and thanks for your kind wishes, but... What we think, and by the way, these four are not yet customers as we defined them.

Speaker Change: And our first question will come from the line of Ben Reitzes with Melius. Your line is open.

Ben Reitzes: Hey, guys. Thanks, a lot.

Ben Reitzes: And congrats on the results Hock, you talked about four more customers coming online.

Speaker Change: Can you just talk a little bit more about the.

Speaker Change: The trend Youre seeing can any of these customers.

Speaker Change: As big as the current three and.

Speaker Change: What does this say about the custom silicon trend overall, and your optimism and upside to the business long term. Thanks.

Speaker Change: Okay.

Ben Reitzes: Well very interesting question Ben.

Speaker Change: Thanks for your kind wishes, but.

Speaker Change: What we are seeing.

Speaker Change: And by the way this fall are not yet customers as we define it.

Hock Tan: As I've always said... You know, in developing and creating XPU. You know, we are not really the, we are not really the creator of those experiences, to be honest. We enable... These are those hyperscalers partners we engage with to create. that ship, basically to create that compute system, call it that way, and it comprises the model, the software model, working closely with the compute engine, the XPU and the networking that binds together the clusters, those multiple XPUs as a whole to train those large frontier models. The fact that we create the hardware, it still has to work with the software models and algorithms of those partners of ours before it becomes fully deployable and scalable.

Speaker Change: As I've always said.

Speaker Change: In developing and creating XP use.

Speaker Change: We are not really that we are not really the creator of.

Speaker Change: Those XP used to be honest we in April.

Each of those Hyperscale as partners, we engaged with.

Speaker Change: To create.

That chip.

Speaker Change: To create that compute system call it that way and it's a comprises.

Speaker Change: The model the software model working closely we've Uh huh.

Speaker Change: The compute engine the SBU into networking.

Speaker Change: Together the clusters.

Speaker Change: Those multiple <unk> as a whole to train those lunch front.

Speaker Change: Frontier models.

Speaker Change: So in the end.

Speaker Change: The fact that we create the hardware.

Speaker Change: <unk> has worked.

Speaker Change: With the software models and algorithms of all of those partners of ours before it becomes fully deployable and scale.

Hock Tan: which is why we define customers in this case as those where we know they've deployed a skill and will receive the production volume to enable it. And for that, we only have... just to reiterate. for a I call it... Partners who are trying to create the same thing as the first and to run their own frontier model. And as I also said, it doesn't happen overnight. To do the first trip would take typically a year and a half. And that's very accelerated, and which we could accelerate, given that we essentially have a framework and a methodology that works right now.

Speaker Change: Which is why we defined customers in this case as those where we know they are deployed at scale and we received the production volume to enable it to run and for that we only.

Just to read through the <unk>.

Speaker Change: Four.

Speaker Change: I call it.

Partners, who are trying to create the same thing as the first three.

Speaker Change: And to run their own frontier models.

Speaker Change: Is it on you to train their own frontier models.

Speaker Change: And as I also said it doesn't happen overnight to do the first chip could take would take typically a year and a half.

Speaker Change: And thats very accelerated and which we could accelerate given that we essentially have a framework and a methodology that works right now and also the three customers no reason for it to not work for the fall.

Hock Tan: It works for the three customers. No reason for it to not work for the four. But we still need those four partners. create and to develop the software. What we don't do to make it work. And to answer your question, there's no reason why these four guys. would not create a demand in the range of what we're seeing with the first three guys, but probably later. Journey, they started it later and so they will probably get there later. Thank you very much. Thank you.

Speaker Change: Still need those four partners to create and to develop the software.

Which we don't do to make it work.

Speaker Change: To answer your question, there's no reason why this fall guys.

Speaker Change: We're not creed.

Speaker Change: Demand in the range of what we're seeing with the first street guidance bump probably later.

Speaker Change: It's a journey that started later and so they will probably get there later.

Speaker Change: Thank you very much.

Harlan Sur: One moment for our next question and that will come from the line of Harlan Sur with JP Morgan. Your line is open. Good afternoon and great job on the strong quarterly execution, Hock and team. Great to see the continual momentum in the AI business here in the first half of your fiscal year and the continued broadening out of your AI ASIC customers. I know, Hock, last earnings, you did call out a strong ramp in the second half of the fiscal year driven by new three nanometer AI accelerated programs kind of ramping. Can you just help us either qualitatively, quantitatively profile the second half step up relative to what the team just delivered here in the first half?

Speaker Change: Thank you one moment for our next question and that will come from the line of Harlan sur with Jpmorgan. Your line is open.

Harlan Sur: Good afternoon, and great job on the strong quarterly execution Hopkinton.

Harlan Sur: You see the continued momentum in the AI business here in the first half of your fiscal year and continue.

Harlan Sur: Continuing broadening out of your AI Asa customers I know Hock last earnings you did call out a strong ramp in the second half of the fiscal year driven by new three nanometer AI accelerated programs kind of ramping can you just help us either qualitatively or quantitatively profile.

Harlan Sur: Half step up relative to what the team just delivered here in the first half has a profile change either favorably less favorably versus what you thought maybe 90 days ago, because quite frankly, I mean, a lot has happened since last earnings right you've had.

Harlan Sur: Has the profile changed either favorably, less favorably versus what you thought maybe 90 days ago? Because quite frankly, I mean, a lot has happened since last earnings, right? You've had the dynamics like DeepSea can focus on AI model efficiency, but on the flip side, you've had strong CapEx outlooks by your cloud and hyperscale customers. So any color on the second half AI profile would be helpful. You're asking me to look into the minds of my customers and I hate to tell you they don't show me their entire mindset here, but why are we beating the numbers so far in Q1?

Harlan Sur: Now next leg deep sea can focus on.

Harlan Sur: AI model efficiency, but on the flip side, you've had on Capex outlooks by your cloud and Hyperscale customers. So any color on the second half profile would be helpful.

You're asking me to look into the minds of my customers and I hate to tell me. They don't tell you. They don't show me.

Harlan Sur: <unk> mindset here, but one why we're beating the numbers so far in Q1.

Hock Tan: seems to be encouraging in Q2. And partly from improved networking shipments, as I indicated, to those XPUs and AI accelerators. Even in some cases you'd be used together for the hyperscalers and and that's good and partly also we think There is some pull in Thank you. And on the second half. that you talked about 90 days ago, the second half three nanometer ramp, is that still very much on track? Harlan, thank you. I only got you two, sorry, let's not speculate on the second half. Okay. Thank you, Hock.

Harlan Sur: And it seems to be encouraging in Q2.

Harlan Sur: Police from improved networking shipments as I indicated to the.

Harlan Sur: The dose XP use and AI accelerators.

Harlan Sur: Even in some cases gpus together.

Harlan Sur: For the Hyperscale is.

Harlan Sur: And thats, good and partly also thing.

Harlan Sur: There is some.

Harlan Sur: Some pull ins.

Harlan Sur: Shipments and acceleration call it that way of shipments yes in.

Harlan Sur: In fiscal 'twenty five.

Harlan Sur: And on the second half.

Harlan Sur: You talked about 90 days ago, the second half nanometer ramp is still very much on track pilot.

Speaker Change: Alan Thank you only guide Q2, sorry, let me, let's not speculate on the second half.

Hock Tan: Okay. Thank you hock.

Harlan Sur: Thank you.

William Stein: One moment for our next question and that will come from the line of William Stein with Truist Securities. Your line is open. Great, thank you for taking my question. Congrats on these pretty great results. You know, it seems from the news headlines about tariffs and about deep seek that there may be some disruptions, some customers and some other complementary suppliers seem to feel a bit useful times for great companies to sort of emerge as something bigger and better than they were in the past. You've grown this company in a tremendous way over the last decade plus.

Speaker Change: Thank you one moment, our next question and that will come from the line of William Stein with <unk> Securities. Your line is open.

William Stein: Okay, great. Thank you for taking my question Congrats on these pretty great results.

Speaker Change: Got it.

William Stein: It seems from.

William Stein: The news headlines about tariffs and about deep seek that.

William Stein: It may be some disruption.

William Stein: Disruption some customers and some other complementary suppliers or seem to feel a bit paralyzed perhaps.

William Stein: Difficulty, making tough decisions.

William Stein: Those tend to be.

William Stein: Really useful times for great companies to sort of emerge as something bigger is better than they were in the past.

William Stein: <unk> grown this company.

Speaker Change: Mendes way over the last decade plus.

William Stein: And you're doing great now, especially in this AI area. But I wonder if you're seeing that sort of disruption from these dynamics that we suspect are happening based on, you know, headlines, what we see from other companies. And how, aside from adding these customers in AI, I'm sure there's other great stuff going on. But should we expect some bigger changes to come from Broadcom as a result of this? You post a very interesting set of issues and questions, and those are very relevant, The only issue, the only problem we have at this point... I would say.

And Youre doing great now, especially in this AI area, but I wonder if you're seeing that sort of disruption from from these dynamics that we see.

Speaker Change: Suspect are happening based on headlines of what we see from other companies and how aside from adding these customers and AI I'm sure. There's other great stuff going on but should we expect some bigger changes to come from Broadcom as a result of this.

Speaker Change: Mhm.

Speaker Change: You posted a very interesting set of issues and questions.

Speaker Change: And those are very relevant and interesting issues.

Speaker Change: The only issue the only problem we have at this point is.

Speaker Change: I'd say it is.

Hock Tan: It's really two worlds. to know why we're all lame. I mean, there's the threat, the noise of Terry. Special. Chip. that has not materialized yet, nor do we know how it will be structured. So, we don't know. But we do experience and we are living it now, is the disruption that is occurring in the pause in a positive way, I should add a very positive disruption. in semiconductors on a generative AC. generate the for watching. really accelerating the development of semiconductor technology, both process and packaging as well as design. It was higher and higher performance accelerators and networking functionality.

Speaker Change: Israeli tour.

Speaker Change: To Norway will all land I mean, there's the threat the noise of territories.

Speaker Change: Especially on chips.

Speaker Change: That hasn't materialized, yet nor do we know how it would be structured so we don't know, but we do experience and we are leaving it now is the disruption.

Speaker Change: That is.

Speaker Change: In a positive way I should add a very positive disruption.

Speaker Change: In semiconductors.

Speaker Change: <unk> generated.

Generator of AI.

Speaker Change: Sure.

Speaker Change: <unk> also had the resale repeating.

Speaker Change: But we feel it more than ever is.

Speaker Change: Really is celebrating the development of semiconductor technology both.

Speaker Change: Process.

Speaker Change: And packaging as well as design.

Speaker Change: Higher and higher performance.

Speaker Change: Accelerators and networking functionality.

Hock Tan: We've seen that innovation, that those upgrades occur every month as we face new interesting challenges. And when, particularly with XPUs, we're trying, we've been asked... OptiMind. to frontier models of Our partners, our customers, as well as our hyperscale partners. And we, it's a lot of... I mean it's a privilege almost for us to participate in it and try to optimize. And by optimized, I mean you look at an accelerator. You can look at it for simple terms, high-level, to perform, to want to be mentioned, not just on one single metric, which is... Compute Capacity, how many teraflops?

Speaker Change: We're seeing the innovation that.

Speaker Change: Those upgrades.

Speaker Change: All occur.

Speaker Change: Every month as we face new interesting challenges and win and particularly with XP use rig trying we've been asked to.

Speaker Change: To optimize.

Speaker Change: Two frontier models of.

Speaker Change: Our partners, our customers as well as our Hyperscale partners and there is.

Speaker Change: Is a lot of.

I mean.

Speaker Change: It's a privilege almost for us to be to participate in it and try to optimize.

Speaker Change: And by optimize I mean, you look at an accelerator.

Speaker Change: You can look at is simple terms high level to meant to perform to want to maybe mention not just on one single mentoring.

Speaker Change: Compute capacity, how many teraflops is Mona net is also tied to the fact that.

Hock Tan: It's more than that. It's also tied to the fact that... This is a distributed computing problem. It's not just the compute capacity of a single... ties itself to the next adjacent GPU. So that has an impact. So you're doing that, you have to balance with that. Then you decide, are you doing training or you're doing pre-fueling, post-training, fine-tuning. And again, then comes how much memory do you balance against that? And with it, how much latency you can afford, which is memory bandwidth. So you look at at least four variables, maybe even five, if you include in memory bandwidth, not just memory capacity, when you go straight to inference.

Speaker Change: This is a distributed computing problem, it's not just the <unk> the compute capacity of a single XP.

Speaker Change: <unk>. Our GPU is also the network bandwidth is tied to sell to the next adjacent GP XP or GPU. So that has an impact so you're doing that you will have the balance of that then you decide how you're doing training or youre doing pretty feeling.

Speaker Change: Post training fine tuning and again then comes how much memory do you balance against that and Richard how much latency you can afford we just memory bandwidth. So you look at at least four variables maybe even five.

Speaker Change: In crude in memory bandwidth not just memory capacity when you go straight to inference. So we have all these variables to play with and we try to optimize it. So all of this is very very.

Hock Tan: So we have all these variables to play with, and we try to optimize it. So all this is very, very, I mean, it's a great experience for our engineers to push their envelope on how to create all those chips. And so that's the biggest disruption we see right now from sheer trying to create. and Push the Envelope on Generative AI, trying to create the best hardware infrastructure to run it. Beyond that, yeah, there are other things, too, that come into play, because with AI, as I indicated, it does not just drive hardware for enterprises, it drives the way the architect their data centers.

I mean, it's a great experience for our engineers to push the envelope on how to create all those chips and so that's the biggest disruption we see right now from sheer trying to create.

Speaker Change: And push the envelope on generative trying.

Speaker Change: Trying to create a best hardware infrastructure to run it.

Speaker Change: Beyond that yes. There are there are other things that come into play because we are as I indicated just not just drive hardware for enterprises. It drives the way the architect.

Speaker Change: Our data centers.

Hock Tan: you know data requirement they have keeping data private on under control becomes important so suddenly push of workloads towards public cloud. may take a little pause as The large enterprises, particularly, have to take... to recognize that you want to run AI workloads. You probably think very hard about running them on press. and suddenly you push yourself towards saying you've got to upgrade your own data set. Delta Vagas, Vaibhavi Kkahani, Madhu Shikha from Harnarayan. Traummann has my comments on VMware Private AI Foundation. This is true, especially enterprises pushing direction are quickly recognizing that how, where do they run their AI workloads.

Speaker Change: Data requirement.

Speaker Change: Keeping data private <unk>.

Speaker Change: On the control becomes important.

Speaker Change: Suddenly.

Speaker Change: The push of workloads towards public cloud.

Speaker Change: May take a little pause.

Speaker Change: Yes.

Speaker Change: Large enterprises, particularly after then they have to take.

Speaker Change: To recognize that.

Speaker Change: You wanted to run AI workloads.

Speaker Change: Probably thinking very hard about running them on Prem.

Speaker Change: And suddenly push yourself towards saying you've got to upgrade your own data centers.

Speaker Change: To do and manage their own data to run it on Prem.

Speaker Change: And that's also pushing a trend that we have been seeing now over the past.

Speaker Change: Two one months.

Speaker Change: Hence my comment on Vmware private AI Foundation.

Speaker Change: It's too, especially enterprises pushing direction quickly recognizing that how well do they run the AI workloads. So those.

Hock Tan: So those are. Transcripts provided by Transcription Outsourcing, LLC. Sensitive Rules on Sovereignty in Cloud and Data. On as far as you mentioning tariffs is concerned, I think that's too early for us to figure out where you are. and probably maybe give it another three, six months, we'll probably have a better idea where to go. Thank you.

Speaker Change: Trends, we see today and a lot of it coming out of AI a lot of it coming out of.

Speaker Change: Uh huh.

Speaker Change: Sensitive rules on sovereignty.

Speaker Change: Cloud and.

Speaker Change: David.

As far as you mentioning tariffs is concerned I think that's too early for us to figure out a way to online and probably maybe give it another three to six months, we'll probably have a better idea of where to go.

Speaker Change: Thank you.

Unknown Executive: One moment for our next question. And that will come from the line of Ross Seymour with Deutsche Bank. Your line is open. They certainly asked a question. Hock, I want to go back to the XTU side of things and going from the four new engagements, not yet named customers, two last quarter and two more today that you announced, I want to talk about going from kind of design win to deployment. How do you judge that? Because there is some debate about, you know, tons of design wins, but the deployments actually don't happen, either that they never occur or that the volume is never what is originally promised.

Speaker Change: Thank you our next question.

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

Hock Tan: Yes. Thanks for let me ask the question Hock I wanted to go back to the <unk> side of things and going from the ore.

Hock Tan: New engagements not yet named customers two last quarter and two more today that you announced I wanted to talk about going from kind of design win to deployment.

Hock Tan: How do you judge that because there is some debate about tons of design wins, but the deployment is actually don't happen either that they never occur or that the volume.

Hock Tan: Is never what is originally promised how do you view that kind of conversion ratio is there a wide range around it or is there. Some way you could help us kind of understand how that works.

Ross Seymour: How do you view that kind of conversion ratio? Is there a wide range around it? Or is there some way you could help us kind of understand how that works? Ross, that's an interesting question. I'll take the opportunity to say, the way we look at design is probably very different from the way many of our peers look at it out there. Number one, to begin with, we believe design when we know our product is produced in scale, at scale and is actually deployed. Literary Deployment in Production. Getting in the product, it takes a year, easily, from the product in the hand.

Hock Tan: Well each.

Hock Tan: The interesting question I'll take the opportunity to say.

Hock Tan: The way, we look at design rates, probably very different from the way many of our peers look at these other out there number one to begin with.

Hock Tan: We believe design win when we know our product is in producing scale at scale and is actually deployed.

Hock Tan: Literally deploying in production so that takes a long lead time because from taping out.

Hock Tan: Getting into product it takes a year easily from the product in the hands.

Hock Tan: of our partner to when it goes into scale production, it will take six months to a year is our experience. number one and number two. I mean, producing and deploying 5,000 XPUs, that's a joke, that's not real production in our view. And so we also limit ourselves in selecting partners. to people who really need that large volume. You need that large volume from our viewpoint in scale right now in mostly training, training of large language models, frontier models in a continuing trajectory. So we limit ourselves to how many customers or how many potential customers that exist out there, Ross, and we tend to be very selective.

Hock Tan: Of our partner to when it goes in the scaled production it would take six months to a year is our experience that we're seeing.

Hock Tan: Number one and number two.

Hock Tan: I mean, producing and deploying 5000 X peers.

Hock Tan: That's a joke that's not real production.

Hock Tan: In our view and so we also limit ourselves in selecting partners.

Hock Tan: Two people, who really need that large volume.

Hock Tan: Need their launch volume from our viewpoint in scale right now in mostly training training of large language models from GM models in the continuing trajectory.

Hock Tan: So we eliminate ourselves to hanmi customers.

Hock Tan: How many potential customers that exist out there and we tend to be very selective.

Hock Tan: who we pick to begin with. So when we say design means. Relay is at scale. It's not something that starts in six months and die or a year and die again. Basically, it's a selection of customers. It's just the way we run our ASIC business in general for the last 15 years. We pick and choose the customers because we know this guy and we do multi-year roadmaps with these customers because we know these customers are sustainable. I put it bluntly, we don't do it for startups. Thank you.

Hock Tan: Who would pick to begin with so when we say design win it.

Hock Tan: It really is at scale is not something that starts in six months and die all year and die again.

Hock Tan: Basically is the selection of customers is just the way we run our ASIC business in general for the last 15 years, we pick and choose the customers because we know discount and we do multiple multi year roadmaps with this customers because we know discussed mers are sustainable.

Hock Tan: I put it bluntly, we don't do it for start ups.

Hock Tan: Okay.

Hock Tan: Thank you.

Stacy Rasgon: And one moment for our next question. And that will come from the line of Stacy Rasgon with Bernstein Research. Your line is open. Hi, guys, thanks for taking my question. I wanted to go to the three customers that you do have in volume today. And what I wanted to ask was, is there any concern about some of the new regulations or the AI diffusion rules that are going to get put in place, supposedly in May, impacting any of those design wins or shipments? It sounds like you think all three of those are still on at this point.

Hock Tan: And one moment our next question.

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

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

Stacy: I wanted to go to the three customers that you do have in volume today.

Stacy: What I wanted to ask was is there any concern about some of the new regulations or the AI diffusion rules that are going to get put in place supposedly on may <unk>.

Stacy: Acting any of those design wins are shipments it sounds like you think.

Stacy: All three of those are still on at this point, but anything you could tell us about.

Hock Tan: But anything you could tell us about, worries about new regulations or AI diffusion rules impacting any of those wins would be Thank you. In this era, or this current era of geopolitical tensions, and fairly dramatic actions all around by governments... There's always some concern at the back of everybody's mind. But to answer your question directly, no. We don't have any good. So none of those are going into China or to Chinese customers then? No comment. Are you trying to locate who they are? Okay, that's helpful. Thank you.

Stacy: It's about new regulations or AIG fusion was impacting of those wins would be helpful.

Stacy: Thank you in this era or this current era of geopolitical tensions.

Stacy: And fairly dramatic actions all around by governments.

Stacy: Yes, it's always.

Stacy: Some concern at the back of everybody's mind.

Stacy: But to answer your question directly.

Stacy: Sure.

Stacy: We don't have any concerns.

Speaker Change: So none of those are going into China or Chinese customers then.

Stacy: No comment.

Speaker Change: Yes.

Uh huh.

Speaker Change: Uh huh.

Okay. That's helpful. Thank you. Thank you one moment our next question.

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. Hock, whenever you have described your AI opportunity, you have always emphasized the training workload. But the perception is that the AI market could be dominated by the inference workload, especially with these new reasoning models. So what happens to your opportunity and share if the mix moves more towards inference? Does it, you know, does it create a bigger spam for you than the 60 to 90 billion? You know, does it keep it the same, but there is a different mix of product?

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

Vivek Arya: Thanks for taking my question Hock whenever you have described your opportunity you've always emphasized the training workload, but.

Speaker Change: But the perception is that the market could be dominated by the inference workloads, especially with these new.

Speaker Change: Reasoning model, so what happens to your opportunity and sure if the mix moves more towards insurance does it.

It does it create a bigger Sam for you than the $60 billion to $90 billion does it keep it the same but there is a different mix of products or does it more entrenched heavy market favorite GPU over that next to you.

Hock Tan: Or does a more inference heavy market favor a GPU over an XPU? Thank you. That's a good question, interesting question. By the way, I never, I do talk a lot about training. We do, our XPUs are... also focus on inference. as a separate product line. They do, and that's why I can say the architecture of those chips are very different from the architecture of the training chips. and . And so it's a combination of those two, I should add, that adds up to this. Time is Most Valued! Press Questions The larger part of the dollars come from training.

Speaker Change: That's a good question interesting question by the way I nervous I do talk a lot about training, we do our chip our XP used us.

Speaker Change: Also focus on inference.

Speaker Change: As a separate product line they.

Speaker Change: They do and Thats why I can say the architecture of those chips are very different from the architecture of the training chips.

Speaker Change: And.

Speaker Change: And so it's a combination of those two I should add that adds up to this 60% to $90 billion.

Speaker Change: So if I had not been clear I do apologize is a combination of both but having said that.

Speaker Change: The larger part of the dollars come from training.

Unknown Executive: non-inference within the service of the same that we have talked about so far.

Not inference.

Speaker Change: Within this service.

Speaker Change: Sam talked about so far.

Speaker Change: Thank you.

Harsh Kumar: One moment for our next question. and that will come from the line of Harsh Kumar with Piper Sandler. Your line is open. Thanks, Broadcom team. And again, great execution. Just had a quick question. We've been hearing that almost all of the large clusters that are 100k plus, they're all going to ethernet. I was wondering if you could help us understand the importance of when the customer is making a selection, choosing between a guy that has the best switch ASIC, such as you, versus a guy that might have to compute there. Can you talk about what the customer is thinking and what are the final points that they want to hit upon when they make that selection for the NIC card?

Speaker Change: One moment for our next question.

Operator: And that will come from the line of harsh Kumar with Piper Sandler Your line is open.

Speaker Change: Thanks, Broadcom team and again great execution.

Speaker Change: Talk had a quick question, we've been hearing that almost all of the large clusters that are 100, K plus they are all going to Ethernet.

Speaker Change: I was wondering if you could help us understand the importance of when the customer is making a selection choosing between a guy that has the best fit change takes such as U verse is the guy that might have the compute there can you talk about what the customers thinking and what are the final points that they want to hit upon when they make that selection.

Speaker Change: Nic cards.

Hock Tan: Okay, I'm sick. No, yeah, it's down to In the case of the hyperscalers, now very much so. very driven, fine. Perform and its performance. what you're mentioning about on connecting Scaling up and scaling out those AI accelerators, be they XPU or GPU, among hyperscalers. And in most cases among those hyperscalers we engage with when it comes to connecting please those clusters They are very driven by performance. I mean, if you are in a race to really get the best performance out of... your hardware as you train and continue to train your frontier models. That matters more than anything else.

Speaker Change: Okay.

Speaker Change: Great.

Speaker Change: No.

Speaker Change: Yes, it's down too.

Speaker Change: In the in the case of the curve in the case of the Hyperscale is now very much so.

Speaker Change: It's very driven.

<unk>.

Speaker Change: Performance.

Speaker Change: <unk> performance was.

Speaker Change: Are you mentioning about on connecting.

Speaker Change: Scaling up and scaling out dose AI accelerators, BD SBU, all GPU among hyperscale and in most cases, among those hyperscale as we engage with when it comes to connecting.

Speaker Change: Those clusters.

They are very driven.

Speaker Change: By performance I mean, if we are in a race to really get the best performance.

Speaker Change: Out of.

Speaker Change: <unk>.

Speaker Change: Hardware as you train.

Speaker Change: And continued training yields.

Speaker Change: Yes.

Speaker Change: <unk> models that matters more than anything else. So.

Hock Tan: So the basic first thing they go for is proven. That's a proven piece of hardware. It's a proven system, subsystem in our case, that makes it work. And in that case, we tend to have a big advantage. I mean, networking, R.S. You know, switching and routing for the last 10 years at least, and the fact that it's AI. Just makes it more interesting for our engineers to work on. But it's basically based on proven technology and experience in pushing the envelope on going from 800 gigabit per second bandwidth to 1.6 and moving on 3.2, which is exactly why we keep stepping up this rate of investment in coming up with a product where we take the Tomahawk 5, we double the radix to deal with just one hyperscaler because they want high radix to create larger clusters while running.

Speaker Change: Basically first thing they go for is proven that's the.

Speaker Change: Proven.

Speaker Change: Piece of hardware as a proven.

Speaker Change: Uh huh.

Speaker Change: It is.

Speaker Change: <unk> system sub system in our case that makes it work.

Speaker Change: And in that case, we tend to have a big advantage because.

In networking, our us and us.

Switching and routing on us for the last 10 years at least.

Speaker Change: The fact that is AI.

Speaker Change: Just makes it more interesting for our engineers to work on.

Speaker Change: But it's basically based on proven technology and experience in pushing that.

Speaker Change: Pushing the envelope ongoing from 800 gigabit per second bandwidth to one six and moving on three two which is exactly why we keep stepping up this rate of investment in coming up with our products.

We take our tomo five.

Speaker Change: <unk> doubled the greater mix to deal with just one hyperscale because they won high ratings to create larger clusters, while running.

Hock Tan: then we've got a smaller but that doesn't stop us from moving ahead to the next generation of Tomahawk 6 and I dare say we're even planning Tomahawk 7 and 8 right now and we're speeding up the rate of development and it's all largely for that few guys by the way. So we're making a lot of investment for very few customers, hopefully with very large. Cooperatives Phase 3 Page of Nth But, if anything else, that's the big bets we're playing. Thank you, Hock.

Speaker Change: When we have that are smaller but that doesn't stop us from moving ahead to the next generation of Tomahawk steaks.

Speaker Change: And I Daresay, we even planning amongst seven and eight right now and we're speeding up the rate of development and it's all Luckily for that few guys by the way.

Speaker Change: So we're making a lot of investment for very few customers hopefully with very large.

Speaker Change: So served available markets.

Speaker Change: Nothing else, that's a big Bachelor pricing.

Hock Tan: Thank you hock.

Hock Tan: Thank you.

Speaker Change: Thank you one moment, our next question and that will come from the line of Timothy Arcuri with UBS. Your line is open.

Unknown Executive: 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. Um, Hock, in the past, you have mentioned XPU units growing from about 2 million last year to about 7 million, you said, in the 2027-2028 timeframe. My question is, do these four new customers, do they add to that 7 million unit number? I know, in the past, you sort of talked about an ASP of, you know, 20 grand by then. So, so, so those three, the first three customers are clearly a subset of that 7 million unit.

Timothy Arcuri: Thanks, a lot hock in the past you have mentioned <unk> units growing from about $2 million last year to about $7 million. You said in the 2027 2028 timeframe. My question is do these four new customers do they had to that 7 million unit number I know in the past you've sort of talked about Asps 'twenty.

Timothy Arcuri: 20 Grand by then so so so those first three customers are clearly a subset of that 7 million units. So do these new four engagements drive that seven higher or do they just fill in to get to that $7 million.

Timothy Arcuri: So do these new four engagements drive that 7 higher? Or do they just fill in to get to that 7 million? Thanks. And thanks Tim for asking that. To clarify, as I made, I thought I made it clear in my comments, no, duh, the market we are talking about, including when you translate the unit, is only among the three. customers we have today. The other four, we talk about engagement partners. We don't consider that as customers yet and therefore are not in our served available market. Okay, so they would add to that number. Okay. Thanks.

Timothy Arcuri: Thanks, Tim for asking that to clarify as I made I thought I made it clear in my comments.

Timothy Arcuri: No.

Speaker Change: The market, we are talking about including when you translate the unit is only among the three.

Timothy Arcuri: Customers we have today.

Timothy Arcuri: The other four we talk about engagement partners.

Timothy Arcuri: We don't consider that as customers yet and.

Timothy Arcuri: And therefore are not in our served available market.

Timothy Arcuri: Okay. So they would add to that number okay. Thanks a lot.

Timothy Arcuri: Thanks.

CJ Muse: One moment for our next question. And that will come from the line of CJ Muse with Cancer Fitzgerald. Your line is open. Yeah, good afternoon. Thank you for taking the question. I guess I'll have to follow up on your prepared remarks and comments earlier around optimization with your best hardware and hyperscalers with their great software. I'm curious how you're expanding your portfolio now to six, you know, mega scale kind of frontier models will enable you to, you know, in one blush, you know, share tremendous information, but at the same time, a world where these six truly want to differentiate.

Speaker Change: Our next question.

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

Speaker Change: Yes, good afternoon, and thank you for taking the question I guess talk to follow up on your prepared remarks and <unk>.

Speaker Change: Comments earlier around optimization with your best hardware and hybrid <unk> with a great software.

Speaker Change: I'm curious, how you're expanding your portfolio now to six.

Speaker Change: Mega scale kind of frontier models.

Speaker Change: We will enable you to.

Speaker Change: Bosch share.

Speaker Change: This information, but at the same time or world, where these six truly want to differentiate so obviously.

Hock Tan: So obviously, you know, the goal for all of these players is exaflops per second per dollar of capex per watt. And I guess to what degree are you aiding them in this effort? And and where does maybe the Chinese wall kind of start where they want to kind of differentiate and and and not share with you kind of some of the work that you're doing? Thank you. Oh, you know, we only provide very basic fundamental technology in semiconductors to enable these guys to use what we have and optimize it to their own particular model. and the algorithms that relate to those models.

Speaker Change: The goal for all of these players as extra flops per second per dollar of Capex per watt.

Speaker Change: And I guess to what degree are you aiding them in this effort and where it is maybe the Chinese wall kind of start where.

Speaker Change: Where they want to kind of differentiate in and not share with you. Some of the work that youre doing thank you.

Speaker Change: Oh.

Speaker Change: We only provide.

Very base.

Speaker Change: Our basic fundamental technology in semiconductors to enable these guys to use what we have.

Speaker Change: And optimize it to their own particular models and algorithms that relate to those models. That's it that's all we do so that's the level of a lot of that optimization, we do for each of them and as I mentioned earlier, there are maybe five degrees of freedom.

Hock Tan: That's it, that's all we do. So that's the level of a lot of that optimization we do for each of them. And as I mentioned earlier, there are maybe five degrees of freedom that we do. And we play with that. And so even if there are five degrees of freedom, there's only so much we can do at that point. But it is, and how they basically, how we optimize it is all tied to the partner telling us how they want you to do it. So that's only so much, we also have visibility. But it's and but it's what we do now is what the XPU model is.

Speaker Change: That we do and we play with that and so even if their degrees of freedom freedom is only so much we can do at that point, but it is and how the two <unk> basically how we optimize it is all tied to the partner telling us how they wanted to do.

Speaker Change: To do it. So there is only so much we also have visibility on.

Speaker Change: But it's.

Speaker Change: It's what we do know is what the XP model share optimization translating to performance, but also.

Hock Tan: Sheer optimization, translating to performance, but also for the listening. Thank you for listening. the size of the cluster and whether they use it for training. No pre-training. post-training in France. You know, test time scaling. All of them have their own characteristics. And that's the advantage of doing that XPU and working closely with them to create that stuff. Now, as far as your question on. Cut China and all that. Frankly, I don't have any opinion on that at all. To us, it's a technical game. Thank you very much.

Speaker Change: Power.

Speaker Change: That's very important how they play it.

Speaker Change: It's not just cost dock power couple translates into a total cost of ownership. Eventually is it's how designing empower high and how we balance it in terms of.

Speaker Change: The size of the cluster and whether they use it for training.

Speaker Change: No pre training.

Speaker Change: Post training.

Speaker Change: In France.

Speaker Change: Test time scaling all of them have their own characteristics and thats the advantage of doing that SBU and working closely with them to create that now.

Speaker Change: Now as far as your question on.

Speaker Change: China and all of that frankly, I don't have any opinion on that at all.

Speaker Change: To us it's a technical game.

Speaker Change: Thank you very much.

Hock Tan: One moment for our next question. And that will come from the line of Christopher Rolland with Susquehanna. Your line is open. Hey, thanks so much for the question. And this one's maybe for Hock and for Kirsten. I'd love to know, just because you have kind of the complete connectivity portfolio, how you see new Greenfield scale up opportunities playing out here between, you know, could be or copper or really anything and what additive this could be for your company. And then Kirsten, I think OPEX is up. Maybe just talk about where those OPEX dollars are going towards within the AI opportunity and whether they relate.

Speaker Change: One moment our next question.

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

Speaker Change: Hey, thanks, so much for the question.

Speaker Change: And.

Speaker Change: And this one is maybe for Hawk and for Kirsten.

Speaker Change: I'd Love to know just because you have kind of the complete connectivity portfolio, how you see new greenfield scale up opportunities playing out here.

Speaker Change: Could we optical or copper or really anything in what additive this could be for your company.

Speaker Change: And then and then Chris I think Opex is up maybe just talk about where those opex dollars are.

Speaker Change: Our going towards with them.

Speaker Change: AI opportunity and whether they relate thanks so much.

Hock Tan: Thanks so much. Your question is very broad-reaching and our portfolio, yeah, we deploy, we have the advantage and a lot of the hyperscale customers we deal with. They're talking about a lot of expansion, but it's almost all greenfield. You know, less so brownfield. It's very greenfield, it's all expansion, and it's all tends to be next generation that we do it, which is very exciting. So the opportunity is very, very high. And we deploy, I mean, we're both, we can do it in copper, but what we see a lot of opportunity from is when you connect, provide the networking connectivity through optical, because there are a lot of active elements including either, you know, multi-mode.

Speaker Change: Okay.

Speaker Change: Uh huh.

Speaker Change: Your question is very broad, reaching and our portfolio yeah, we deploy we have the advantage.

Speaker Change: And a lot of the cost.

Speaker Change: Hyperscale customers we deal with.

Speaker Change: We're talking about a lot of expansion.

Speaker Change: But it's all almost all greenfield.

So brownfield is very Greenfield is all expansion and its all tends to be next generation that we do it which is very exciting. So the opportunity is very very high and we deploy.

Speaker Change: We are both we can do it in copper, but while we see a lot of opportunity if the opportunity is when we can.

Speaker Change: And they provide the networking connectivity through obstacle, but there are a lot of development, including either multi mode.

Hock Tan: lasers, which are called Vixels, or edge-minting lasers, for basically single mode, and we do both. So there is a lot of opportunity, just as in scale up versus scale out, we used to do, we still do, a lot of other protocols beyond Ethernet to consider PCI Express, where we were on the leading edge of that PCI Express. and the architecture on networking, switching, so to speak, we offered both. One is a very intelligent switch. and which is like our Jericho family with a dumb nick. or a very smart nick with a dumb switch which is a tomahawk.

Speaker Change: Lasers, which are called <unk> or edge emitting lasers for us basically single mode and we do both so there is a lot of opportunity just as in scale.

Speaker Change: Scale up versus scale out.

Speaker Change: We used to do we still do a lot of us.

The protocols beyond Ethernet to consider PCI Express, where we on the leading edge of that PC Express.

Speaker Change: And that architecture on networking switching so to speak.

Speaker Change: Offered both one is the.

Speaker Change: A very intelligent switch.

Speaker Change: And which is why our Jericho family.

Speaker Change: We have a dumb nic.

Speaker Change: All are very smart Nic, we got demonstrates <unk>, which is the tomahawk we offer both.

Hock Tan: We offer both architectures as well. So yeah, we have a lot of opportunities from it. All things said and done, all this nice white portfolio and all that adds up to probably, as I said in prior quarters, about 20% of our total AI revenue, maybe going to 30%. Though last quarter we hit almost 40%, but that's not the norm. I would say typically, all those other portfolio products still add up to a nice, decent amount of revenue for us, but within the sphere of AI, they add up to, I would say on average, be close to 30%, and XPUs, the accelerators, is 75%.

Speaker Change: Our architectures as well so yes, we have a lot of opportunities for me all things said and done all this nice wide portfolio and all of that.

Speaker Change: Up to.

Speaker Change: Probably.

Speaker Change: As I said in prior quarters about 20% of our total revenue maybe going to 30%, though last quarter, we hit almost 40%, but that's not the norm.

Speaker Change: I'd say typically all those other ports.

Speaker Change: <unk> portfolio of products still end up toward 90 decent amount of revenue for us, but within the sphere of AI. They add up to I would say on average close to 30% and XP used accelerators is 70%.

Kirsten Spears: If that's what you're driving at, perhaps I'll give you some, shed some light on, to us, where, how one matters over the other. But we have a wide range of products in the connectivity, networking side. They just add up, though, to that 30%. And then on the R&D front, as I outlined, on a consolidated basis, we spent $1.4 billion in R&D in Q1, and I stated that it would be going up in Q2. Hock clearly outlined in his script the two areas where we're focusing on. Now, I would tell you, as a company, we focus on R&D across all of our product lines so that we can stay competitive with next-generation product offerings, but he did line out that we are focusing on taping out the industry's first two-nanometer AI XPU packaged in 3D.

Speaker Change: Thats, what youre driving at perhaps that gives you some shed some light on towards way, how one method over the other but we have a wide range of products in the connectivity networking side of it there.

Speaker Change: They just add up to that 30%.

Speaker Change: Thanks, So much rock and then on the R&D front.

Speaker Change: As I outlined on a consolidated basis, we spent $1 4 billion in R&D in Q1, and I've stated that it would be going up in Q2, clearly outlined in his script. The two areas, where we're focusing on I would tell you that the company we focus on R&D across all of our product lines. So that we can stay competitive with next generation product offer.

Speaker Change: But he did line out that we are focusing on taping out the industry's first two nanometer aix's du packaged in <unk> I was one in his script and that's an area that we're focusing on and then you mentioned that we doubled the ratings capacity of existing Tomahawk fives.

Kirsten Spears: That was one in his script, and that's an area that we're focusing on. And then he mentioned that we doubled the rate of capacity of existing Tomahawk 5s to enable our AI customers to scale up on Ethernet towards the one million XPUs. So, I mean, that's a huge focus of the company. Yeah, thank you very much, Kirsten.

Speaker Change: To enable our AI customers to scale up our Ethernet towards the 1 million extra year Napa huge focus of the company.

Speaker Change: Yes, thank you very much.

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, hi, Hock. Thanks. Just a quick question on the networking side. Just wondering how much does up sequentially on the AI side, and any thoughts on M&A going forward? It's been a lot of headlines around the Intel products group, etc. So thanks. Okay, on the networking side, it's indicated Q1 short bit of a surge, but I don't expect that to be that mix of 60-40, 60-80. Compute and 40% networking to be something that Normal.

Speaker Change: And one moment our next question.

And that will come from the line of Vijay Rakesh with Mizuho. Your line is open.

Vijay Rakesh: Yes, hi.

Vijay Rakesh: Thanks. Thanks, So just a quick question on the networking side, just wondering how much it goes up sequentially on the AI side and any thoughts on M&A going forward.

Vijay Rakesh: It's been a lot of headlines around.

Vijay Rakesh: Okay.

Vijay Rakesh: Yes.

Vijay Rakesh: Okay on the networking side as indicated Q1 short.

Vijay Rakesh: A bit of a search, but I don't expect that to be.

Vijay Rakesh: That makes of 60 40, 60 is it a compute and 40% net we're going to be something that is normal I think the Nam it's closer to 70 30, maybe at best 30% and so who knows what Q2 is we cannot see Q2 as kantar.

Hock Tan: I think the norm is closer to 70-30 maybe, at best 30%. And so, who knows what Q2 is. We kind of see Q2 as continuing, but that's just, in my mind, a temporary blip. The norm will be 70-30, if you take it across a period of time like six months, a year. That's a good question. M&A? No, I'm too busy. We're too busy doing AI and VMware at this point. We're not thinking of it at this point. Thanks, Harc. Thank you.

Vijay Rakesh: <unk>, but that's just.

Vijay Rakesh: My mind, a temporary blip.

Vijay Rakesh: Nam will be 70 30.

Vijay Rakesh: Across a period of time like six months a year.

Speaker Change: Thanks for your question M&A now I'm too busy we're too busy doing AI and <unk> at this point.

Vijay Rakesh: We're not thinking of it.

Speaker Change: At this point.

Speaker Change: Thanks Mark.

Ji Yoo: That is all the time we have for our question and answer session. I would now like to turn the call back over to Ji Yoo for any closing remarks. Thank you, Cherie. Broadcom currently plans to report its earnings for the second quarter of fiscal year 2025. After close of market on Thursday, June 5 2025. A public webcast of Broadcom's earnings conference call will follow at 2pm Pacific.

Speaker Change: Thank you that is all the time, we have for a question and answer session I would now like to turn the call back over to you for any closing remarks.

Thank you Sheree Broadcom currently plans to reported earnings for the second quarter of fiscal year 2025, after close of market on Thursday June five 2025 public.

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

Unknown Executive: That will conclude our earnings call today. Thank you all for joining. Shree, you may end the call. Ladies and gentlemen, thank you for participating.

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

Speaker Change: Thank you Les.

Speaker Change: Ladies and gentlemen, thank you for participating. This concludes today's program you may now disconnect.

This concludes today's program. You may now disconnect.