Q3 2026 Broadcom Inc Earnings Call

September 2, 2026

Speaker #1: Thank you, Sheree, and good afternoon, everyone. Joining me on today's call are Hock Tan, President and CEO; Amie Thuener, Chief Financial Officer; and Charlie Kawwas, President, Semiconductor Solutions Group.

Speaker #1: Broadcom Distributed a press release and financial tables after the market closed, describing our financial performance for the third quarter fiscal year 2026. If you did not receive a copy, you may obtain the information from the investor section of Broadcom's website at broadcom.com.

Speaker #1: This conference call is being webcast live, and an audio replay of the call can be accessed for 1 year, through the investor section of Broadcom's website.

Speaker #1: During the prepared comments, Hock and Amie will be providing details of our third quarter fiscal year 2026 results, guidance for our fourth quarter of fiscal year 2026, as well as commentary regarding the business environment.

Speaker #1: We'll 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 risk factors that could cause our actual results to differ materially from the forward-looking statements made on this call.

Speaker #1: In addition to U.S. gap reporting, Broadcom reports certain financial measures on a non-gap basis. A reconciliation between gap and non-gap measures to the extent possible is included in the tables attached to today's press release.

Speaker #1: Comments made during today's call will primarily refer to our non-gap financial results. I will now turn the call over to Hock.

Speaker #2: Well, thank you, Ji and thank you, everyone here, for joining us today. We delivered an exceptional quarter with revenue operating income and free cash flow all exceeding prior records.

Operator: Welcome to Broadcom Inc.'s third quarter fiscal year 2026 financial results conference call. 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. Please go ahead.

Speaker #2: Inc.'s Q3 fiscal year 2026 financial results conference call. At this time, for opening remarks and introductions, I would like to turn the call over to Ji Yoo, Head of Investor Relations at Broadcom Inc. Please go ahead.

Speaker #2: And driving this was our Q3 AI semiconductor revenue, which grew 221% year on year, and up 54% sequentially. This brought our consolidated revenue to 29.6 billion, which was up 86% year on year.

Speaker #2: Please go ahead.

Speaker #1: Thank you, Sherri, 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.

Ji Yoo: Thank you, Sherry, and good afternoon, everyone. Joining me on today's call are Hock Tan, President and CEO, Amie Thuener, Chief Financial Officer, and Charlie Kawwas, President, Semiconductor Solutions Group. Broadcom distributed a press release and financial tables after the market close describing our financial performance for the third quarter fiscal year 2026. 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 Amie will be providing details of our third quarter fiscal year 2026 results, guidance for our fourth quarter of fiscal year 2026, as well as commentary regarding the business environment. We will take questions after the end of our prepared comments.

Speaker #1: Broadcom distributed a press release and financial tables after the market closed, describing our financial performance for the third quarter of fiscal year 2026. If you did not receive a copy, you may obtain the information from the investor section of Broadcom's website at broadcom.com.

Speaker #2: Operating income grew even faster at 92% year on year, with operating margin at a record 68% of revenue, reflecting strong operating leverage. Q3 demand was simply hot, and we're just getting started.

Speaker #1: 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.

Speaker #2: Our 6 XPU customers are accelerating the adoption of custom accelerators, and AI semiconductor revenue more than tripled year over year to 16.7 billion. During the quarter, we delivered Ironwood TPU v7, version 7, in high volume to both Anthropic and Google.

Speaker #1: During the prepared comments, Hock and Amy will be providing details of our third quarter fiscal year 2026 results, guidance for our fourth quarter of fiscal year 2026, as well as commentary regarding the business environment.

Ji Yoo: Please refer to our press release today and our recent filings with the SEC for information on 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 to the extent possible 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. I will now turn the call over to Hock.

Speaker #2: At the same time, we began production shipments of the next-generation TPU, version 8i, for Google. This new TPU generation has been designed with more memory and bandwidth compared to Ironwood, and just like Ironwood is optimized for inference workloads.

Speaker #1: 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, to the extent possible, is included in the tables attached to today's press release.

Speaker #2: And in performance, it is comparable, if not surpasses, the Vera Rubin, GPU. Just to reiterate the major technology challenges of developing these complex accelerators, Broadcom is now shipping the TPU, version 8i, ahead of the MediaTek, version v8T, which in fact was initiated earlier.

Speaker #1: Comments made during today's call will primarily refer to our non-GAAP financial results. I will now turn the call over to Hock.

Speaker #3: Well, thank you, Ji. And thank you, everyone here, for joining us today. We delivered an exceptional quarter, with revenue, operating income, and free cash flow all exceeding prior records.

Hock Tan: Well, thank you, Ji, and thank you everyone here for joining us today. We delivered an exceptional quarter with revenue, operating income, and free cash flow all exceeding prior records. Driving this was our Q3 AI semiconductor revenue, which grew 221% year on year and up 54% sequentially. This brought our consolidated revenue to $29.6 billion, which was up 86% year on year. Operating income grew even faster at 92% year on year, with operating margin at a record 68% of revenue, reflecting strong operating leverage. Q3 demand was simply hot and we are just getting started. Our six XPU customers are accelerating the adoption of custom accelerators and AI semiconductor revenue more than tripled year over year to $16.7 billion. During the quarter, we delivered Ironwood TPU v7, version 7 in high volume to both Anthropic and Google.

Speaker #3: And driving this was our Q3 AI semiconductor revenue, which grew 221% year on year and was up 54% sequentially. This brought our consolidated revenue to $29.6 billion, which was up 86% year on year.

Speaker #2: In Q3, we also shipped Jalapeno, OpenAI's first-generation custom accelerator, which outperforms Grace Blackwell's GPU for inference workloads. Overall, our XPU shipments for the third quarter were up over 3.5 times year on year, and represented 73% of AI revenue during the quarter.

Speaker #3: Operating income grew even faster, at 92% year on year, with operating margin at a record 68% of revenue, reflecting strong operating leverage. Q3 demand was simply hot, and we're just getting started.

Speaker #2: Our AI networking 2.5 times year on year. And this strong momentum continues into Q4. During the quarter, we expect to accelerate shipments of Ironwood to Anthropic, we also expect to ramp up high-volume shipments of the TPU, version 8i, to Google.

Speaker #3: Our six XPU customers are accelerating the adoption of custom accelerators, and AI Semiconductor revenue more than tripled year over year to $16.7 billion. During the quarter, we delivered Ironwood TPU v7, version 7, in high volume to both Anthropic and Google.

Speaker #2: Shipments of Jalapeno for OpenAI will continue, and for Matter we expect production shipments of their custom NTIA accelerator optimized for inference and recommendation at scale.

Speaker #2: In Q4, we expect both XPUs and AI networking revenue to triple year on year, and together we expect these deployments to drive our Q4 AI revenue to 21.7 billion, which is up 236% year on year.

Speaker #3: At the same time, we began production shipments of the next-generation TPU, version 8i, for Google. This new TPU generation has been designed with more memory and bandwidth compared to Ironwood, and, just like Ironwood, is optimized for inference workloads.

Hock Tan: At the same time, we began production shipments of the next generation TPU v8i for Google. This new TPU generation has been designed with more memory and bandwidth compared to Ironwood, and just like Ironwood, is optimized for inference workloads. In performance, it is comparable, if not surpasses the Vera Rubin GPU. Just to reiterate the major technology challenges of developing these complex accelerators, Broadcom is now shipping the TPU v8i ahead of the MediaTek v8t, which in fact was initiated earlier. In Q3, we also shipped Jalapeño, OpenAI's first generation custom accelerator, which outperforms Grace Blackwell GPU for inference workloads. Overall, our XPU shipments for the third quarter were up over 3.5x year on year and represented 73% of AI revenue during the quarter. Our AI networking revenue was up over 2.5x year on year.

Speaker #3: And in performance, it is comparable—if not surpasses—the Vera Rubin GPU. Just to reiterate the major technology challenges of developing these complex accelerators, Broadcom is now shipping the TPU version 8i ahead of the MediaTek version v8T, which, in fact, was initiated earlier.

Speaker #2: Based on all this, Q4 on this Q4 guidance, I should say, we expect our fiscal 2026 AI revenue to be 58 billion. For the year, up 186% year on year upon year.

Speaker #2: And above our prior guidance of 56 million. We are continuing to see exponential growth in demand from our XPU customers. We believe the vast majority of compute demand for AI workloads today originates from these concentrated groups who develop state-of-the-art frontier models.

Speaker #3: In Q3, we also shipped Jalapeño, OpenAI's first-generation custom accelerator, which outperforms Grace Blackwell's GPU for inference workloads. Overall, our XPU shipments for the third quarter were up over 3.5 times year on year, and represented 73% of AI revenue during the quarter.

Speaker #2: And we expect their need for compute infrastructure to inflect even more in 2027 and 2028. And they are all growing with XPUs. To achieve superior performance costs and power.

Speaker #3: Our AI networking revenue was up over 2.5 times year-on-year, and this strong momentum continues into Q4. During the quarter, we expect to accelerate shipments of Ironwood to Anthropic. We also expect to ramp up high-volume shipments of the TPU, version 8i, to Google.

Hock Tan: This strong momentum continues into Q4. During the quarter, we expect to accelerate shipments of Ironwood to Anthropic. We also expect to ramp up high volume shipments of the TPU v8i to Google. Shipments of Jalapeño for OpenAI will continue, and for Meta, we expect production shipments of their custom MTIA accelerator optimized for inference and recommendation at scale. In Q4, we expect both XPUs and AI networking revenue to triple year on year, and together we expect these deployments to drive our Q4 AI revenue to $21.7 billion, which is up 236% year on year. Based on this Q4 guidance, we expect our fiscal 2026 AI revenue to be $58 billion for the year, up 186% year upon year, and above our prior guidance of $56 million. We are continuing to see exponential growth in demand from our XPU customers.

Speaker #2: Let me now walk you through each of their journey, each of our customers' journey, towards using XPUs at scale to run their frontier models worldwide, workloads, sorry.

Speaker #2: Our engagement with Google has never been stronger. This has been recently strengthened by a long-term agreement to develop and supply future-generations of TPUs and AI networking.

Speaker #3: Shipments of Jalapeño for OpenAI will continue, and for Matter, we expect production shipments of their custom NTIA accelerator optimized for inference and recommendation at scale.

Speaker #2: Under this agreement, we're planning to deliver multi-tens of billions of dollars of TPUs annually over the next several years. We expect this growing demand in 2028 and 2029 to be fulfilled through successive generations of the increasingly complex TPUs we are developing today.

Speaker #3: In Q4, we expect both XPU and AI networking revenue to triple year on year, and together we expect these deployments to drive our Q4 AI revenue to $21.7 billion, which is up 236% year on year.

Speaker #3: Based on all this, on this Q4 guidance, I should say, we expect our fiscal 2026 AI revenue to be $58 billion for the year, up 186% year on year.

Speaker #2: With Google. Our partnership with Google will continue to sustain because we have the strongest IP portfolio in semiconductor design, including industry-leading 30s chip-to-chip interconnect, leading-edge HBM and SRAM integration, and simply differentiated advanced packaging.

Speaker #3: And above our prior guidance of $56 million. We are continuing to see exponential growth in demand from our XPU customers. We believe the vast majority of compute demand for AI workloads today originates from these concentrated groups who develop state-of-the-art frontier models.

Speaker #2: Most of all, we've been consistently we have consistently, I should say, delivered the fastest time-to-market for TPUs from product definition to production. We found the need for re-smiths, we believe, these are very deep modes for any competitor to cross.

Hock Tan: We believe the vast majority of compute demand for AI workloads today originates from this concentrated group who develops state-of-the-art frontier models. We expect their need for compute infrastructure to inflate even more in 2027 and 2028. They are all growing with XPUs to achieve superior performance, cost, and power. Let me now walk you through each of our customers' journey towards using XPUs at scale to run their frontier workloads. Our engagement with Google has never been stronger. This has been recently strengthened by a long-term agreement to develop and supply future generations of TPUs and AI networking. Under this agreement, we are planning to deliver multi-tens of billions of dollars of TPUs annually over the next several years.

Speaker #3: And we expect their need for compute infrastructure to inflect even more in 2027 and 2028. And they are all growing with XPUs to achieve superior performance, cost, and power.

Speaker #2: Moving on to Anthropic, starting with the 1-gigawatt of Ironwood, we are deploying in 2026. We expect Anthropic to deploy another 5-gigawatts of TPU, version 8i, in 2027.

Speaker #3: Let me now walk you through each of their journeys—each of our customers' journeys—towards using XPUs at scale to run their frontier models worldwide; workloads, sorry.

Speaker #2: And in 2028, we have clear line of sight to deliver another and incremental 10-gigawatts even as we expect Google to grow for us. Anthropic is on track to become our largest XPU customer in 2027 and sustain that in 2028.

Speaker #3: Our engagement with Google has never been stronger. This has been recently strengthened by a long-term agreement to develop and supply future generations of TPUs and AI networking.

Speaker #2: For OpenAI, Jalapeno is on track for the planned deployment of 1.3-gigawatts in 2027. Together with OpenAI, we are deep in development of the next-generation XPU beyond Jalapeno which is approaching with Taper.

Speaker #3: Under this agreement, we're planning to deliver multi-tens of billions of dollars of TPUs annually over the next several years. We expect this growing demand in 2028 and 2029 to be fulfilled through successive generations of the increasingly complex TPUs we are developing today.

Hock Tan: We expect this growing demand in 2028 and 2029 to be fulfilled through successive generations of the increasingly complex TPUs we are developing today with Google. Our partnership with Google will continue to sustain because we have the strongest IP portfolio in semiconductor design, including industry-leading SerDes, chip-to-chip interconnect, leading-edge HBM and SRAM integration, and simply differentiated advanced packaging. Most of all, we have consistently delivered the fastest time to market for TPUs from product definition to production without the need for resubmits. We believe these are very deep moats for any competitor to cross. Moving on to Anthropic. Starting with the 1 gigawatt of Ironwood we are deploying in 2026, we expect Anthropic to deploy another 5 gigawatts of TPU v8i in 2027.

Speaker #2: In 2028, we have line of sight for OpenAI to deploy over 5-gigawatts of Jalapeno and its successor generation of XPU. Which would make OpenAI our second-largest XPU customer.

Speaker #3: With Google, our partnership will continue to sustain because we have the strongest IP portfolio in semiconductor design, including industry-leading studies in chip-to-chip interconnect, leading-edge HBM and SRAM integration, and simply differentiated advanced packaging.

Speaker #2: In addition, we are in development development with OpenAI on their third-generation XPU. As OpenAI announced last week, Jalapeno Jalapeno outperforms the grays, Blackwell Ultra in performance, but what latency, throughput, and power and it's actually comparable to Vera Rubin GPUs in running OpenAI workloads.

Speaker #3: Most of all, we've been consistently we have consistently, I should say, delivered the fastest time-to-market for TPUs from product definition to production. We found the need for re-smiths, we believe, these are very deep modes for any competitor to cross.

Speaker #2: The lesson here is when you co-develop a chip that is optimized for your particular LLM workloads, you will outperform any GPU. Like TPUs, Jalapeno demonstrates that it can also run other frontier models.

Speaker #3: Moving on to Anthropic. Starting with the 1 gigawatt of Ironwood we are deploying in 2026, we expect Anthropic to deploy another 5 gigawatts of TPU version 8i in 2027.

Speaker #2: And you can do all this at half the cost of a GPU. Our partnership with Matter to deliver multiple-generations of MTI XPUs remains on track.

Speaker #3: And in 2028, we have clear line of sight to deliver another incremental 10 gigawatts, even as we expect Google to grow for us. Anthropic is on track to become our largest XPU customer in 2027 and sustain that in 2028.

Hock Tan: In 2028, we have clear line of sight to deliver an incremental 10 gigawatts, even as we expect Google to grow for us. Anthropic is on track to become our largest XPU customer in 2027 and sustain that in 2028. For OpenAI, Jalapeño is on track for the planned deployment of 1.3 gigawatts in 2027. Together with OpenAI, we are deep in development of the next generation XPU beyond Jalapeño, which is approaching tape-out. In 2028, we have line of sight for OpenAI to deploy over 5 gigawatts of Jalapeño and its successor generation of XPU, which would make OpenAI our second-largest XPU customer. In addition, we are in development with OpenAI on their third-generation XPU. As OpenAI announced last week, Jalapeño outperforms the Grace Blackwell Ultra in performance by one latency, throughput, and power, and is actually comparable to very robust GPUs in running OpenAI workloads.

Speaker #2: Between 2027, we will be delivering three generations of MTI accelerators to Matter. Across these three generations, we have line of sight to to deploy 3-gigawatts through 2028.

Speaker #3: For OpenAI, Jalapeño is on track for the planned deployment of 1.3 gigawatts in 2027. Together with OpenAI, we are deep in development of the next-generation XPU beyond Jalapeño, which is approaching in 2028.

Speaker #2: Our content in AI, as you know, goes beyond XPUs. We're the leader in AI networking. And we continue to extend our lead. In Ethernet switching, for scale-up and scale-out, we were first to market with our 100-terabit Tomahawk 6, and we just taped out Tomahawk 7.

Speaker #3: We have line of sight for OpenAI to deploy over 5 gigawatts of Jalapeño and its successor generation of XPU, which would make OpenAI our second-largest XPU customer.

Speaker #2: The industry's first 200-terabit per second Ethernet switch. For scaling in, we continue to be the leader in every generation of PCI Express switching. We're now the leader in leading-edge optical DSPs and are rapidly expanding our capacity and market position as a leader in EMLs, VXLs, and CW lasers, for optical interconnects.

Speaker #3: In addition, we are in development with OpenAI on their third-generation XPU. As OpenAI announced last week, Jalapeño outperforms the Grays, Blackwell Ultra in performance, but what latency, throughput, and power, and it's actually comparable to Vera Rubin GPUs in running OpenAI workloads.

Speaker #2: In sum, we continue to invest and invest heavily to provide the broadest and most leading-edge AI portfolio. In fact, our AI network networking revenue is expected to grow just as fast as XPUs over the next few years.

Speaker #3: The lesson here is that when you co-develop a chip optimized for your particular LLM workloads, you will outperform any GPU. Like TPUs, Jalapeño demonstrates that it can also run other frontier models.

Hock Tan: The lesson here is when you co-develop a chip that is optimized for your particular LLM workloads, you will outperform any GPU. Like TPUs, Jalapeño demonstrates that it can also run other frontier models, and you can do all this at half the cost of a GPU. Our partnership with Meta to deliver multiple generations of MTIA XPUs remains on track. Between now and the end of 2027, we will be delivering 3 generations of MTIA accelerators to Meta. Across these 3 generations, we have line of sight to deploy 3 gigawatts through 2028. Our content in AI, as you know, goes beyond XPUs. We are the leader in AI networking, and we continue to extend our lead.

Speaker #2: Reflecting our excellent progress with these key group of LLM customers, here is our outlook for our AI semiconductor revenue. In 2027, we have secured the supply to again double AI revenue to approximately $115 billion.

Speaker #3: And you can do all this at half the cost of a GPU. Our partnership with Matter to deliver multiple generations of MTI XPUs remains on track.

Speaker #3: Between now and the end of 2027, we will be delivering three generations of MTI accelerators to Matter. Across these three generations, we have line of sight to deploy 3 gigawatts through 2028.

Speaker #2: Our demand actually exceeds this outlook and we will work to improve supply. In 2028, we expect the trajectory of growth to continue. We have line of sight for our fiscal 2028 AI semiconductor revenue growth to again double, to $230 billion.

Speaker #3: Our content in AI, as you know, goes beyond XPUs. We're the leader in AI networking, and we continue to extend our lead. In Ethernet switching, for scale-up and scale-out, we were first to market with our 100-terabit Tomahawk 6, and we just taped out Tomahawk 7.

Speaker #2: Here again, we have secured the supply to meet this outlook. This AI revenue guidance through 2028 is being provided to to give you the trajectory of our growth.

Hock Tan: In Ethernet switching for scale up and scale out, we were first to market with our 100 terabit Tomahawk 6, and we just taped out Tomahawk 7, the industry's first 200 terabit per second Ethernet switch. For scaling in, we continue to be the leader in every generation of PCI Express switching. We are now the leader in leading-edge optical DSPs and are rapidly expanding our capacity and market position as a leader in EMLs, VCSELs, and CW lasers for optical interconnects. In sum, we continue to invest and invest heavily to provide the broadest and most leading-edge AI portfolio. In fact, our AI networking revenue is expected to grow just as fast as XPUs over the next few years. Reflecting our excellent progress with this key group of LLM customers, here is our outlook for our AI semiconductor revenue.

Speaker #3: The industry's first 200-terabit-per-second Ethernet switch. For scaling in, we continue to be the leader in every generation of PCI Express switching. We're now the leader in leading-edge optical DSPs and are rapidly expanding our capacity and market position as a leader in EMLs, VCELs, and CW lasers for optical interconnects.

Speaker #2: That demand for compute continues to be extremely strong. As a result, I got to say we are very much on target to exceed $30 in earnings per share in fiscal 2028.

Speaker #2: Now, turning to non-AI semiconductors, Q3 revenue of $4.2 billion was up 5% year on year, and flat sequentially. Broadband and server storage together were up, partially offset by a decline in wireless.

Speaker #3: In sum, we continue to invest—and invest heavily—to provide the broadest and most leading-edge AI portfolio. In fact, our AI networking revenue is expected to grow just as fast as XPUs over the next few years.

Speaker #2: In Q4, we forecast non-AI semiconductor revenue to be approximately $4.3 billion, again up 5% sequentially. Talking about infrastructure software, Q3 revenue of $8.8 billion was up 29% year on year, and what when we sustained ARR growth of 15% year on year.

Speaker #3: Reflecting our excellent progress with these key groups of LLM customers, here is our outlook for our AI semiconductor revenue. In 2027, we have secured the supply to double AI revenue to approximately $115 billion.

Speaker #2: For Q4, we forecast infrastructure software revenue to stabilize at approximately $8.7 billion. We announced VMware Private AI Cloud giving enterprises a secure, cost-effective platform to build and run AI alongside their existing applications.

Hock Tan: In 2027, we have secured the supply to again double AI revenue to approximately $115 billion. Our demand actually exceeds this outlook, and we will work to improve supply. In 2028, we expect the trajectory of growth to continue. We have line of sight for fiscal 2028 AI semiconductor revenue growth to again double to $230 billion. Here again, we have secured the supply to meet this outlook. This AI revenue guidance through 2028 is being provided to give you the trajectory of our growth. That demand continues to be extremely strong. As a result, I got to say, we are very much on target to exceed $30 in earnings per share in fiscal 2028. Turning to non-AI semiconductors, Q3 revenue of $4.2 billion was up 5% year-on-year and flat sequentially. Broadband and server storage together were up, partially offset by a decline in wireless.

Speaker #3: Our demand actually exceeds this outlook, and we will work to improve supply. In 2028, we expect the trajectory of growth to continue. We have line of sight for our fiscal 2028 AI semiconductor revenue growth to again double, to $230 billion.

Speaker #2: It brings together AI infrastructure, security, and compliance, and the tools to build and operate trusted AI agents, all while protecting enterprise data. VCF VMware Cloud, that is, is also making it easier for customers to repatriate workloads from public cloud to private cloud, where they can gain greater control and significantly improve infrastructure economics.

Speaker #3: Here again, we have secured the supply to meet this AI revenue guidance through 2028. This is being provided to give you the trajectory of our growth.

Speaker #2: Enterprise consumption of AI is, in fact, opening a new opportunity for our infrastructure software business. So to sum it all, for Q4 2026, we expect consolidated revenue to grow to $34.8 billion up 93% year on year.

Speaker #3: That demand continues to be extremely strong. As a result, I have to say we are very much on target to exceed $30 in earnings per share in fiscal 2028.

Speaker #2: We expect Q4 AI revenue to be $21.7 billion up $236% year on year. And we expect operating margin to be approximately $66% of revenue.

Speaker #3: Now, turning to non-AI semiconductors—Q3 revenue of $4.2 billion was up 5% year over year, and flat sequentially. Broadband and server storage together were up, partially offset by a decline in wireless.

Speaker #2: And with that, let me turn it over to Amie.

Speaker #1: Thank you, Hock. Let me now provide additional detail on our Q3 financial performance. Consolidated revenue was a record 29.6 billion for the quarter, up 86% year on year.

Speaker #3: In Q4, we forecast non-AI semiconductor revenue to be approximately $4.3 billion, again up 5% sequentially. Talking about infrastructure software, Q3 revenue of $8.8 billion was up 29% year on year, and we sustained ARR growth of 15% year on year.

Hock Tan: In Q4, we forecast non-AI semiconductor revenue to be approximately $4.3 billion, up 5% sequentially. Talking about infrastructure software, Q3 revenue of $8.8 billion was up 29% year on year, and we sustain ARR growth of 15% year on year. For Q4, we forecast infrastructure software revenue to stabilize at approximately $8.7 billion. We announced VMware Private AI Cloud, giving enterprises a secure, cost-effective platform to build and run AI alongside their existing applications. It brings together AI infrastructure, security, and compliance, and the tools to build and operate trusted AI agents, all while protecting enterprise data. VCF, VMware Cloud that is also making it easier for customers to repatriate workloads from public cloud to private cloud, where they can gain greater control and significantly improve infrastructure economics. Enterprise consumption of AI is, in fact, opening a new opportunity for our infrastructure software business.

Speaker #1: Gross margin was 75% of revenue in the quarter, down 210 basis points sequentially, as AI semiconductor revenue was a greater proportion of our total revenue mix.

Speaker #1: This was better than our guidance of 74%. Q3 operating income was a record 20.1 billion, up 92% from a year ago. Even with the declining gross margin due to revenue mix, operating margin increased 240 basis points year over year, to 67.9% because of the phenomenal operating leverage we are achieving.

Speaker #3: For Q4, we forecast infrastructure software revenue to stabilize at approximately $8.7 billion. We announced VMware Private AI Cloud, giving enterprises a secure, cost-effective platform to build and run AI alongside their existing applications.

Speaker #1: Aligned with this, Q3 non-GAAP EPS of $3.32 was up 96% year on year. Now I'll review the P&L for our two segments. Starting with semiconductors.

Speaker #3: It brings together AI infrastructure, security, and compliance, and the tools to build and operate trusted AI agents, all while protecting enterprise data. VCF—VMware Cloud, that is—is also making it easier for customers to repatriate workloads from public cloud to private cloud, where they can gain greater control and significantly improve infrastructure economics.

Speaker #1: Revenue for our semiconductor solutions segment was a record 20 20.8 billion, up 127% year on year, and represented 70% of our total revenue. AI semiconductor revenue of $16.7 billion represented 56% of total revenue, up from $49% in Q2.

Speaker #1: Gross margin for our semiconductor solutions segment was approximately 76%. Operating expenses of $1.2 billion reflected investments in R&D, with OpEx representing 6% of segment revenue.

Speaker #3: Enterprise consumption of AI is, in fact, opening a new opportunity for our infrastructure software business. So, to sum it all, for Q4 2026, we expect consolidated revenue to grow to $34.8 billion, up 93% year on year.

Hock Tan: To sum it all, for Q4 2026, we expect consolidated revenue to grow to $34.8 billion, up 93% year on year. We expect Q4 AI revenue to be $21.7 billion, up 236% year on year. We expect operating margin to be approximately 66% of revenue. With that, let me turn it over to Amie.

Speaker #1: Operating margin of $61% was up 440 basis points year on year, as revenue growth of $127% outpaced operating expenses, which grew 22% year on year.

Speaker #3: We expect Q4 AI revenue to be $21.7 billion, up 236% year on year. And we expect operating margin to be approximately 66% of revenue.

Speaker #1: Moving on to infrastructure software. Revenue of $8.8 billion was up 29% year on year, and represented 30% of our total revenue. Gross margins for infrastructure software was 94% in the quarter, and operating expenses were over $900 million.

Speaker #3: And with that, let me turn it over to Amy.

Speaker #1: Thank you, Hock. Let me now provide additional detail on our Q3 financial performance. Consolidated revenue was a record $29.6 billion for the quarter, up 86% year-on-year.

Amie Thuener: Thank you, Hock. Let me now provide additional detail on our Q3 financial performance. Consolidated revenue was a record $29.6 billion for the quarter, up 86% year on year. Gross margin was 75% of revenue in the quarter, down 210 basis points sequentially, as AI semiconductor revenue was a greater proportion of our total revenue mix. This was better than our guidance of 74%. Q3 operating income was a record $20.1 billion, up 92% from a year ago. Even with the decline in gross margin due to revenue mix, operating margin increased 240 basis points year over year to 67.9% because of the phenomenal operating leverage we are achieving. Aligned with this, Q3 non-GAAP EPS of $3.32 was up 96% year on year. Now I'll review the P&L for our two segments, starting with Semiconductors.

Speaker #1: Q3 software operating margin was up 650 basis points year on year, to approximately $84%. Moving on to the balance sheet. We ended the third quarter with $24 billion of cash, compared to $19.6 billion in the prior quarter, up 4.3 billion sequentially.

Speaker #1: Gross margin was 75% of revenue in the quarter, down 210 basis points sequentially, as AI semiconductor revenue was a greater proportion of our total revenue mix.

Speaker #1: This was better than our guidance of 74%. Q3 operating income was a record $20.1 billion, up 92% from a year ago. Even with the declining gross margin due to revenue mix, operating margin increased 240 basis points year over year to 67.9%, because of the phenomenal operating leverage we are achieving.

Speaker #1: We ended the third quarter with inventory of $4.5 billion, to support our strong semiconductor demand. Moving on to cash flow. Free cash flow in the quarter was a record 13.7 billion, and represented 46% of revenue.

Speaker #1: We spent $532 million on capital expenditures in the quarter. Turning to capital allocation. In Q3, we paid stockholders $3.1 billion of cash dividends, based on a quarterly common stock dividend of $65 per share.

Speaker #1: Aligned with this, Q3 non-GAAP EPS of $3.32 was up 96% year on year. Now, I'll review the P&L for our two segments, starting with semiconductors.

Speaker #1: In Q3, we also paid down $5.6 billion of long-term debt. Subsequent to quarter end, we paid down an additional $1.5 billion of senior notes upon maturity.

Speaker #1: Revenue for our Semiconductor Solutions segment was a record $20.8 billion, up 127% year on year, and represented 70% of our total revenue. AI semiconductor revenue of $16.7 billion represented 56% of total revenue, up from 49% in Q2.

Amie Thuener: Revenue for our Semiconductor Solutions segment was a record $20.8 billion, up 127% year on year and represented 70% of our total revenue. AI semiconductor revenue of $16.7 billion represented 56% of total revenue, up from 49% in Q2. Gross margin for our Semiconductor Solutions segment was approximately 76%. Operating expenses of $1.2 billion reflected investments in R&D, with OpEx representing 6% of segment revenue. Operating margin of 61% was up 440 basis points year on year, as revenue growth of 127% outpaced operating expenses, which grew 22% year on year. Moving on to infrastructure software. Revenue of $8.8 billion was up 29% year on year and represented 30% of our total revenue. Gross margins for infrastructure software was 94% in the quarter, and operating expenses were over $900 million. Q3 software operating margin was up 650 basis points year on year to approximately 84%. Moving on to the balance sheet.

Speaker #1: The weighted average coupon rate and years to maturity of our gross principal fixed rate debt of $59.6 billion is 4%, and $7.4 years respectively.

Speaker #1: In June, we established the AI XPV platform in partnership with Apollo and Blackstone, to enable more than 20 gigawatts of compute infrastructure for OpenAI and Anthropic by the end of 2028.

Speaker #1: Gross margin for our Semiconductor Solutions segment was approximately 76%. Operating expenses of $1.2 billion reflected investments in R&D, with OpEx representing 6% of segment revenue.

Speaker #1: We closed the first $35 billion tranche in June for Anthropic's 1 gigawatt deployment, which is already underway. While future tranches will include unique features tailored to specific lab and investor needs, our core strategy remains consistent.

Speaker #1: Operating margin of 61% was up 440 basis points year on year, as revenue growth of 127% outpaced operating expenses, which grew 22% year on year.

Speaker #1: Moving on to Infrastructure Software. Revenue of $8.8 billion was up 29% year on year, and represented 30% of our total revenue. Gross margins for Infrastructure Software were 94% in the quarter, and operating expenses were over $900 million.

Speaker #1: First, we are empowering two of our most strategic customers, the leading AI labs, to bridge the gap between their current cash flow and the significant upfront investments required for their businesses.

Speaker #1: Our XPUs enable cost-effective, sustainable growth at multiples of their infrastructure costs. Second, this XPV platform enables highly investable customers and facilitates execution where demand is already locked in.

Speaker #1: Q3 software operating margin was up 650 basis points year on year, to approximately 84%. Moving on to the balance sheet, we ended the third quarter with $24 billion of cash, compared to $19.6 billion in the prior quarter—up $4.3 billion sequentially.

Amie Thuener: We ended the third quarter with $24 billion of cash, compared to $19.6 billion in the prior quarter, up $4.3 billion sequentially. We ended the third quarter with inventory of $4.5 billion to support our strong semiconductor demand. Moving on to cash flow. Free cash flow in the quarter was a record $13.7 billion and represented 46% of revenue. We spent $532 million on capital expenditures in the quarter. Turning to capital allocation. In Q3, we paid stockholders $3.1 billion of cash dividends based on a quarterly common stock dividend of $0.65 per share. In Q3, we also paid down $5.6 billion of long-term debt. Subsequent to quarter end, we paid down an additional $1.5 billion of senior notes upon maturity. The weighted average coupon rate and years to maturity of our gross principal fixed rate debt of $59.6 billion is 4% and 7.4 years, respectively.

Speaker #1: And third, we review any strategic financings through a commercial as well as balance sheet lens, consistent with our existing capital allocation framework. Through the XPV platform, we partner with sophisticated third-party financial partners to independently underwrite and capitalize the assets, rather than providing the direct financing ourselves.

Speaker #1: We ended the third quarter with inventory of $4.5 billion to support our strong semiconductor demand. Moving on to cash flow, free cash flow in the quarter was a record $13.7 billion, and represented 46% of revenue.

Speaker #1: We spent $532 million on capital expenditures in the quarter. Turning to capital allocation, in Q3 we paid stockholders $3.1 billion of cash dividends, based on a quarterly common stock dividend of $0.65 per share.

Speaker #1: Where necessary, we may provide modest residual value guarantees, which are contingent liabilities we view as low risk, supported by the strong profitability trajectory of these labs and the sustaining value of the underlying assets.

Speaker #1: In Q3, we also paid down $5.6 billion of long-term debt. Subsequent to quarter end, we paid down an additional $1.5 billion of senior notes upon maturity.

Speaker #1: Moving on to guidance. Our guidance for Q4 is our guidance for Q4 is for consolidated revenue of $34.8 billion, up 93% year on year.

Speaker #1: The weighted average coupon rate and years to maturity of our gross principal fixed-rate debt of $59.6 billion is 4% and 7.4 years, respectively. In June, we established the AI XPV platform in partnership with Apollo and Blackstone to enable more than 20 gigawatts of compute infrastructure for OpenAI and Anthropic by the end of 2028.

Speaker #1: We forecast semiconductor revenue of approximately $26.1 billion, up 136% year on year. Within this, we expect Q4 AI semiconductor revenue of $21.7 billion, up over $236% year on year.

Amie Thuener: In June, we established the AI XPV Platform in partnership with Apollo and Blackstone to enable more than 20 gigawatts of compute infrastructure for OpenAI and Anthropic by the end of 2028. We closed the first $35 billion tranche in June for Anthropic's one gigawatt deployment, which is already underway. While future tranches will include unique features tailored to specific lab and investor needs, our core strategy remains consistent. First, we are empowering two of our most strategic customers, the leading AI labs, to bridge the gap between their current cash flow and the significant upfront investments required for their businesses. Our XPUs enable cost-effective, sustainable growth at multiples of their infrastructure costs. Second, this XPV Platform enables highly investable customers and facilitates execution where demand is already locked in.

Speaker #1: We expect Q4 infrastructure software revenue of approximately $8.7 billion, up 24% 25% year on year. Moving on to margins. As the proportion of AI revenue accelerates in Q4, we expect our Q4 consolidated gross margin to be approximately 73%, down from 78% a year ago.

Speaker #1: We closed the first $35 billion tranche in June for Anthropic's 1-gigawatt deployment, which is already underway. While future tranches will include unique features tailored to specific lab and investor needs, our core strategy remains consistent.

Speaker #1: First, we are empowering two of our most strategic customers—the leading AI labs—to bridge the gap between their current cash flow and the significant upfront investments required for their businesses.

Speaker #1: As we've discussed previously, this reflects the increasing mix of XPUs, with their increasing memory content, which is diluting our consolidated gross margin. Regardless, we expect Q4 operating margin to be approximately 66%, flat from a year ago, because our strong revenue growth drives substantial operating leverage.

Speaker #1: We expect the non-gap tax rate for Q4 and fiscal year 2026 rate to be approximately 16% due to the impact of the global minimum tax and the geographic mix of incomes compared to that of fiscal year 2025.

Speaker #1: And third, we review any strategic financings through a commercial as well as a balance sheet lens, consistent with our existing capital allocation framework. Through the XPV platform, we partner with sophisticated third-party financial partners to independently underwrite and capitalize the assets, rather than providing the direct financing ourselves.

Amie Thuener: Third, we review any strategic financings through a commercial as well as balance sheet lens, consistent with our existing capital allocation framework. Through the XPV Platform, we partner with sophisticated third-party financial partners to independently underwrite and capitalize the assets rather than providing the direct financing ourselves. Where necessary, we may provide modest residual value guarantees, which are contingent liabilities we view as low risk, supported by the strong profitability trajectory of these labs and the sustaining value of the underlying assets. Moving on to guidance. Our guidance for Q4 is for consolidated revenue of $34.8 billion, up 93% year-on-year. We forecast Semiconductor Solutions revenue of approximately $26.1 billion, up 136% year-on-year. Within this, we expect Q4 AI Semiconductor Solutions revenue of $21.7 billion, up over 236% year-on-year. We expect Q4 infrastructure software revenue of approximately $8.7 billion, up 25% year-on-year. Moving on to margins.

Speaker #1: We expect the Q4 non-gap diluted share count to be approximately 4.94 billion shares, excluding the impact of potential share repurchases. And in Q4, we expect capital expenditures of $1.4 billion, as we invest in capacity for semiconductors.

Speaker #1: As Hock mentioned in his remarks, we expect AI revenue to double again to approximately $115 billion in fiscal 2027 and double again in fiscal 2028 to $230 billion.

Speaker #1: Moving on to guidance. Our guidance for Q4 is our guidance for Q4 is for consolidated revenue of 34.8 billion, up 93% year on year.

Speaker #1: This AI revenue guidance is being provided to you to give you the trajectory of our growth, and we do not intend to update it on a quarterly basis.

Speaker #1: That concludes my prepared remarks. Operator, please open up the call for questions.

Speaker #1: We forecast semiconductor revenue of approximately $26.1 billion, up 136% year on year. Within this, we expect Q4 AI semiconductor revenue of $21.7 billion, up over 236% year on year.

Speaker #2: Thank you. As a reminder to ask a question, you will need to press star 11 on your telephone. To withdraw your question, press star 11 again.

Speaker #2: Due to time restraints, we ask that you please limit yourself to one question. One moment while we compile the Q&A roster. And that will come from the line of Joseph Moore with Morgan Stanley.

Speaker #1: We expect Q4 infrastructure software revenue of approximately $8.7 billion, up 24% to 25% year on year. Moving on to margins: As the proportion of AI revenue accelerates in Q4, we expect our Q4 consolidated gross margin to be approximately 73%, down from 78% a year ago.

Speaker #2: Your line is open.

Speaker #3: Great. Thank you. And congratulations on the results. You talked about the business doubling next year, and you said demand could be higher than that.

Amie Thuener: As the proportion of AI revenue accelerates in Q4, we expect our Q4 consolidated growth margin to be approximately 73%, down from 78% a year ago. As we've discussed previously, this reflects the increasing mix of XPUs with their increasing memory content, which is diluting our consolidated growth margin. Regardless, we expect Q4 operating margin to be approximately 66%, flat from a year ago, because our strong revenue growth drives substantial operating leverage. We expect the non-GAAP tax rate for Q4 and fiscal year 2026 rate to be approximately 16% due to the impact of the global minimum tax and the geographic mix of income compared to that of fiscal year 2025. We expect the Q4 non-GAAP diluted share count to be approximately 4.94 billion shares, excluding the impact of potential share repurchases. In Q4, we expect capital expenditures of $1.4 billion as we invest in capacity for semiconductors.

Speaker #3: Can you talk about the supply around that and kind of what are the supply bottlenecks and what are the variables that could drive that number higher, if you're able to resolve them?

Speaker #4: Well, Joe, that's a loaded question. Whatever I tell you, you guys go out and print a bigger number. I think that's funny. And we're very careful, and to be honest, we try to be conservative.

Speaker #4: So we're giving you an outlook, and yeah, demand we can ship significantly more. Question to some in our mind sometimes is, are they going to be even as we ship the chips, going to be deployed on a timely basis?

Speaker #1: We expect the non-GAAP tax rate for Q4 and fiscal year 2026 to be approximately 16%, due to the impact of the global minimum tax and the geographic mix of income, compared to that of fiscal year 2025.

Speaker #4: And that's always very much in our mind when we give you that outlook. But certainly, our customers want us to ship more. But we are we have secured supply and we think it's the right number to put it to 115.

Speaker #1: We expect the Q4 non-GAAP diluted share count to be approximately 4.94 billion shares, excluding the impact of potential share repurchases. In Q4, we expect capital expenditures of $1.4 billion, as we invest in capacity for semiconductors.

Speaker #4: And if times if circumstances change and our ability to scale more supply change, of course, we will uplift but at this point, that's our best outlook.

Speaker #1: As Hock mentioned in his remarks, we expect AI revenue to double again to approximately $115 billion in fiscal 2027, and to double again in fiscal 2028 to $230 billion.

Amie Thuener: As Hock mentioned in his remarks, we expect AI revenue to double again to approximately $115 billion in fiscal 2027, and double again in fiscal 2028 to $230 billion.

Speaker #4: And by the way, the same applies the same thinking applies to 2028 when we give you that outlook of 230 230 billion. This is real demand.

Speaker #1: This AI revenue guidance is being provided to give you the trajectory of our growth, and we do not intend to update it on a quarterly basis.

Amie Thuener: This AI revenue guidance is being provided to you to give you the trajectory of our growth, and we do not intend to update it on a quarterly basis. That concludes my prepared remarks. Operator, please open up the call for questions.

Speaker #4: We believe based on what's available what data center sites locations are ready 2028 with respect to our customers. The size of what we have and against the supply chain we have in leading edge wafers, substrates, and HBM memory.

Speaker #1: That concludes my prepared remarks. Operator, please open up the call for questions.

Speaker #2: Thank you. As a reminder, to ask a question, you will need to press star 11 on your telephone. To withdraw your question, press star 11 again.

Operator: Thank you. As a reminder, to ask a question, you will need to press star 1 1 on your telephone. To withdraw your question, press star 1 1 again. Due to time restraints, we ask that you please limit yourself to one question. One moment while we compile the Q&A roster. That will come from the line of Joseph Moore with Morgan Stanley. Your line is open.

Speaker #4: And so this is again a carefully structured outlook that we believe we can achieve.

Speaker #3: Great. Thank you.

Speaker #2: One moment for our next question. That will come from the line of Blaine Curtis with Jefferies. Your line is open.

Speaker #2: Your line is open.

Speaker #3: Great, thank you. And congratulations on the results. You talked about the business doubling next year, and you said demand could be higher than that.

Joseph Moore: Great. Thank you, and congratulations on the results. You talked about the business doubling next year, and you said demand could be higher than that. Can you talk about the supply around that and what are the supply bottlenecks and what are the variables that could drive that number higher if you are able to resolve them?

Speaker #5: Hey, guys. Thanks for taking my question. I actually want to follow up on Joe's on supply. Big key point. Can you just talk about from either a substrate or, you know, kind of the interposer co-ops replacement?

Speaker #3: Can you talk about the supply around that, and what are the supply bottlenecks? Also, what are the variables that could drive that number higher, if you're able to resolve them?

Speaker #5: Is any XPUs decided to use your Singapore capacity, and how does that fit into that supply picture you put together?

Speaker #4: Well, Joel, that's a loaded question. Whatever I tell you, you guys go on and print a bigger number. I think that's funny. And we're very careful, and to be honest, we try to be conservative.

Hock Tan: Well, Joe, that's a loaded question. Whatever I tell you guys go on and print a bigger number. I think that's funny. We are very careful, and to be honest, we try to be conservative. So we are giving you an outlook, and demand, we can ship significantly more. Question in our mind sometimes is, are they going to be, even as we ship the chips, going to be deployed on a timely basis? That's always very much in our mind when we give you that outlook. But certainly, our customers want us to ship more. But we have secured supply, and we think it's the right number to put it to 115. If times, if circumstances change and our ability to scale more supply change, of course, we will uplift. But at this point, that's our best outlook.

Speaker #4: Well, we are going to start deploying our Singapore fab for substrates. By the way, in starting fiscal 27. And that would, I guess, address a key part of our supply bottlenecks.

Speaker #4: So we're giving you an outlook, and yeah, demand — we can ship significantly more. The question, to some in our minds sometimes, is: Are they, even as we ship the chips, going to be deployed on a timely basis?

Speaker #4: Anything else?

Speaker #2: One moment.

Speaker #5: Maybe you can go.

Speaker #4: Yes. Finish your question.

Speaker #4: And that's always very much in our mind when we give you that outlook. But certainly, our customers want us to ship more. But we have secured supply and we think it's the right number to put it at 115.

Speaker #5: No. I was going to give another one because that was a short answer. I mean, just one clarification quickly. The Google number you gave, and then you said 10 gigawatts for Anthropic.

Speaker #5: Are those it's not one number, right? The 10 gigawatts is just for Anthropic?

Speaker #4: 10 gigawatts is just for Anthropic.

Speaker #4: And if times, if circumstances change and our ability to secure more supply changes, of course, we will uplift. But at this point, that's our best outlook.

Speaker #5: Thanks, Hock.

Speaker #2: One moment for our next question. And that will come from the line of Harlan Sur with JP Morgan. Your line is open.

Speaker #4: And by the way, the same applies—the same thinking applies—to 2028. When we give you that outlook of $230 billion, this is real demand.

Hock Tan: The same thinking applies to 2028 when we give you that outlook of 230 billion. This is real demand we believe, based on what's available, what data center sites, locations are ready 2028 with respect to our customers, the size of what we have, and against the supply chain we have in leading-edge wafers, substrates, and HBM memory. So this is again, a carefully structured outlook that we believe we can achieve.

Speaker #6: Yeah. Good afternoon. Thanks for taking my question. Hock, your customers are driving their XPUs, their GPUs higher in performance. The networking bandwidth is also scaling accordingly, right?

Speaker #6: Your customers are now moving from 100 gigabits per second per lane to 200 gigabits per second per lane. From Tom Hock 5 to Tom Hock 6.

Speaker #4: We believe, based on what's available, which data center sites and locations are ready in 2028 with respect to our customers. The size of what we have, and against the supply chain we have in leading-edge wafers, substrates, and HBM memory.

Speaker #6: We've heard that Tom Hock 6 ramp has been the fastest ramp of your any of your switching families. And we've also heard that you guys are almost sold out for Tom Hock 6 for next year.

Speaker #6: I'm not sure if you can give us an update on that. And it's also good to see the team kicked out its next-gen Tom Hock 7 with 200 gig servers.

Speaker #4: And so, this is again a carefully structured outlook that we believe we can achieve.

Speaker #6: Out of your six frontier model-based XPU customers, how many of them are using Tom Hock scale-out networking? And can you also just give us an update on the adoption of your Tom Hock Ultra scale-up platform as well?

Speaker #3: Great. Thank you.

Joseph Moore: Great. Thank you.

Speaker #2: One moment for our next question. That will come from the line of Blaine Curtis with Jefferies. Your line is open.

Operator: One moment for our next question. That will come from the line of Blayne Curtis with Jefferies. Your line is open.

Speaker #5: Hey, guys. Thanks for taking my question. I actually want to follow up on Joe's question on supply—big key point. Can you just talk about, from either a substrate or, you know, kind of the interposer co-ops replacement perspective?

Blayne Curtis: Hey, guys. Thanks for taking my question. I actually want to follow up on Joe's on supply, the key point. Can you just talk about from either a substrate or interposer CoWoS replacement, has any XPUs decided to use your Singapore capacity, and how does that fit into that supply picture you put together?

Speaker #4: Charlie will take this question. He knows everything here.

Speaker #6: Thank you, Hock. And thanks, Harlan as well. You're right. Our Tom Hock 6 has been a phenomenal success. It started really first in scale-out, as you mentioned.

Speaker #5: Has any XPUs decided to use your Singapore capacity, and how does that fit into the supply picture you put together?

Speaker #4: Well, we are going to start deploying our Singapore fab for substrates, by the way, starting in fiscal '27. And that would, I guess, address a key part of our supply bottlenecks.

Speaker #6: And it's available both in 100 gig and 200 gig servers. So we actually have two versions of Tom Hock 6. Both are very successful, and both are deployed in pretty much all of the AI hyperscalers that are building XPUs with us.

Hock Tan: Well, we are going to start deploying our Singapore fab for substrates, by the way, starting fiscal 2027. That would, I guess, address a key part of our supply bottlenecks. Anything else?

Speaker #6: And even those that are not using our XPUs are using the Tom Hock 6 both 100 gig and 200 gig. With respect to Tom Hock Ultra, we have quite innovated ahead of the market here by enabling scale-up using low-latency Ethernet.

Speaker #4: Anything else?

Speaker #2: One moment.

Operator: One moment.

Blayne Curtis: Maybe you can go.

Speaker #5: Maybe you can go.

Speaker #4: It finish your question.

Hock Tan: Finish your question.

Speaker #5: No, I was going to give another one because I was trying to answer— I mean, just one clarification quickly. The Google number you gave, and then you said 10 gigawatts for Anthropic. Are those— it's not one number, right?

Blayne Curtis: No, I was going to give another one because that was short answer. Just one clarification quickly. The Google number you gave, and then you said 10 gigawatts for Anthropic. It is not one number, right? The 10 gigawatts is just for Anthropic?

Speaker #6: The adoption also on this device has surprised us, and we're starting to see it deployed starting actually this quarter and in FY27, coming up in scale-up applications.

Speaker #5: The 10 gigawatts is just for Anthropic?

Hock Tan: 10 gigawatts is just for Anthropic.

Speaker #4: 10 gigawatts is just for Anthropic.

Speaker #4: Just to amplify, what Charlie is saying with Tom Hock Ultra is that we are enabling now scaling up within cluster of GPU, XPU in a rack on on basically Ethernet.

Speaker #5: Thanks, Hock.

Blayne Curtis: Thanks, Hock.

Speaker #2: One moment for our next question. That will come from the line of Harlan Sur with J.P. Morgan. Your line is open.

Operator: One moment for our next question. That will come from the line of Harlan Sur with JP Morgan. Your line is open.

Speaker #3: Yeah. Good afternoon. Thanks for taking my question. Hock, your customers are driving their XPUs, their GPUs, higher in performance. The networking bandwidth is also scaling accordingly, right?

Harlan Sur: Yeah, good afternoon. Thanks for taking my question. Hock, your customers are driving their XPUs, their GPUs higher in performance. The networking bandwidth is also scaling accordingly, right? Your customers are now moving from 100 gigabits per second per lane to 200 gigabits per second per lane, from Tomahawk 5 to Tomahawk 6. We have heard that Tomahawk 6 ramp has been the fastest ramp of any of your switching families, and we have also heard that you guys are almost sold out for Tomahawk 6 for next year. I am not sure if you can give us an update on that. It is also good to see the team peep out its next gen Tomahawk 7 with 200 gig SerDes. Out of your six frontier model-based XPU customers, how many of them are using Tomahawk scale-out networking?

Speaker #4: For the first time, because it won't Tom Hock Ultra will perform just as well as anything else that exists prior to that before Ethernet before being able to do it on Ethernet, particularly with respect to loss and latency.

Speaker #3: Your customers are now moving from 100 gigabits per second per lane to 200 gigabits per second per lane, from Tomahawk 5 to Tomahawk 6.

Speaker #3: We've heard that the Tomahawk 6 ramp has been the fastest ramp of any of your switching families. And we've also heard that you guys are almost sold out of Tomahawk 6 for next year.

Speaker #6: Thanks, Charlie. Thanks, Hock.

Speaker #3: I'm not sure if you can give us an update on that. And it's also good to see the team kicked out its next-gen Tomahawk 7 with 200-gig servers.

Speaker #2: One moment for our next question. And that will come from the line of Stacy Rasgon with Bernstein. Your line is open.

Speaker #3: Out of your six frontier model-based XPU customers, how many of them are using Tomahawk scale-out networking? And can you also just give us an update on the adoption of your Tomahawk Ultra scale-up platform as well?

Speaker #7: Hi, guys. Thanks for taking my question. I just wanted to to verify if I add up the gigawatts for 27 and 28, they have line of sight.

Harlan Sur: Can you also just give us an update on the adoption of your Tomahawk Ultra scale-up platform as well?

Speaker #7: I'm still getting roughly 10 for 27, about 6 of which are Anthropic and OpenAI, and roughly 20 for 28, roughly 15 of which are Anthropic and OpenAI.

Speaker #4: Charlie will take this question. He knows everything here.

Hock Tan: Charlie will take this question. He knows everything here.

Speaker #3: Thank you, Hock, and thanks, Harlan, as well. You're right. Our Tomahawk 6 has been a phenomenal success. It started really, first in scale-out, as you mentioned.

Charlie Kawwas: Thank you, Hock. Thanks, Harlan, as well. You are right. Our Tomahawk 6 has been a phenomenal success. It started really first in scale-out, as you mentioned. It is available both in 100 gig and 200 gig SerDes. So we actually have two versions of Tomahawk 6. Both are very successful, and both are deployed in pretty much all of the AI hyperscalers that are building XPUs with us. Even those that are not using our XPUs are using the Tomahawk 6, both 100 gig and 200 gig. With respect to Tomahawk Ultra, we have quite innovated ahead of the market here by enabling scale-up using low latency Ethernet.

Speaker #7: I just want to make sure that was the case. And and and if it is, if I sort of calculate the revenue per gigawatt, I come out with your revenue, guys, I come out somewhere between 11 and 12 billion per gigawatt.

Speaker #7: I mean, your your competitors suggested something like 40. And I'm just wondering, is is is this the right level that we should think about for your content as and and how should that, I guess, trend as we go forward, as you said, and and a more generations of XPUs again with with higher performance and higher memory?

Speaker #3: And it's available both in 100-gig and 200-gig servers. So we actually have two versions of Tomahawk 6. Both are very successful, and both are deployed in pretty much all of the AI hyperscalers that are building XPUs with us.

Speaker #7: Like, how do those trend? So I guess, first, just the gigawatt count, and second, the the the content.

Speaker #3: And even those that are not using our XPUs are using the Tomahawk 6, both 100-gig and 200-gig. With respect to Tomahawk Ultra, we have really innovated ahead of the market here by enabling scale-up using low-latency Ethernet.

Speaker #4: Right. Stacy, you're very clever. You put two tricky questions into one. So let me take it well, let me pass it one by one.

Speaker #4: Yes. You can count the number of gigawatts we are outlining. Not everyone, because we're focused on to be fair, we're six customers. Four of them are just simply going to be huge.

Speaker #3: The adoption also on this device has surprised us, and we're starting to see it deployed starting, actually, this quarter and in FY27, coming up in scale-up applications.

Charlie Kawwas: The adoption also on this device has surprised us, and we are starting to see it deployed, starting actually this quarter and in FY27 coming up in scale-up applications.

Speaker #4: And we outline. If we walk through with you guys, their journey into deploying XPUs as I said, it's a journey. Google has been doing for the last 10 years.

Speaker #4: Just to amplify what Charlie is saying with Tomahawk Ultra, we are now enabling scaling up within a cluster of GPU or XPU in a rack, basically over Ethernet.

Hock Tan: Just to amplify what Charlie is saying with Tomahawk Ultra, is that we are enabling now scaling up within cluster of GPU, XPU in a rack on basically Ethernet for the first time. Because Tomahawk Ultra will perform just as well as anything else that exists prior to that, before being able to do it on Ethernet, particularly with respect to loss and latency.

Speaker #4: Others OpenAI, last two, three years. Matter, last three years. Each one is doing it differently. And we like to take you guys through it.

Speaker #4: For the first time, because it won't, Tomahawk Ultra will perform just as well as anything else that existed prior to that—before Ethernet—before being able to do it on Ethernet, particularly with respect to loss and latency.

Speaker #4: But more importantly, what it means in their deployment and XPU for from 20 for the for these three years, 26, 27, 28. And so we laid out and you're right, Stacy, you add up the gigawatts.

Speaker #4: So we deploy we are showing where they are headed. What we didn't tell you specifically when we came up to the final number is how many of these gigawatts do actually come up for actual deployment?

Speaker #3: Yeah. Thanks, Charlie. Thanks, Hock.

Harlan Sur: Thanks, Charlie. Thanks, Hock.

Speaker #2: One moment for our next question. That will come from the line of Stacey Reskon with Bernstein. Your line is open.

Operator: One moment for our next question. That will come from the line of Stacy Rasgon with Bernstein. Your line is open.

Speaker #4: Because deployment doesn't just include getting the chips out there. Before you get the chips out there, you got to get the data center shell in place.

Speaker #6: Hi, guys. Thanks for taking my question. I just wanted to verify—if I add up the gigawatts for '27 and '28, they have line of sight.

Stacy Rasgon: Hi, guys. Thanks for taking my question. I just wanted to verify, if I add up the gigawatts for 2027 and 2028 that you have line item, I am still getting roughly 10 for 2027, about 6 of which are Anthropic and OpenAI, and roughly 20 for 2028, roughly 15 of which are Anthropic and OpenAI. I just want to make sure that was the case. If it is, if I sort of calculate the revenue per gigawatt, I come out with your revenue guides, I come out somewhere between 11 and 12 billion per gigawatts. I mean, your competitor suggested something like 40. I am just wondering, is this the right level that we should think about for your content, and how should that I guess trend as we go forward, as you said, into more generations of XPUs, again, with higher performance and higher memory?

Speaker #6: I'm still getting roughly 10 for '27, about 6 of which are Anthropic and OpenAI, and roughly 20 for '28, roughly 15 of which are Anthropic and OpenAI.

Speaker #4: And ready for literally production. And we're talking about a fiscal years. So what we're saying is, as you add up the gigawatts, we're not saying that over the next two years, 27, 28, that there are 30 gigawatts that will go into production.

Speaker #6: I just want to make sure that was the case. And if it is, if I sort of calculate the revenue per gigawatt, I come out with your revenue, guys—I come out somewhere between $11 and $12 billion per gigawatt.

Speaker #4: And therefore, we ship it. If we think we're judge it conservatively, to be somewhat less. But we see the demand that if they can get it all in place, and our products and we can ship those chips racks in place, it will be 30 gig 30 gigawatts between the customers we have, the six customers we have.

Speaker #6: I mean, your competitors suggested something like 40, and I'm just wondering, is this the right level that we should think about for your content?

Speaker #6: And how should that, I guess, trend as we go forward, as you said, into more generations of XPUs, again, with higher performance and higher memory?

Speaker #6: Like, how do those trend? So I guess, first, just the gigawatt count, and second, the content.

Stacy Rasgon: How do those trend? I guess first just the gigawatt count and second the content.

Speaker #4: Question is, will all 30 come into production within these two years, fiscal years? And we are giving you a judge number of 115 billion dollars of chips to these guys in 27 fiscal 27.

Speaker #4: Right. Stacey, you're very clever. You put two tricky questions into one, so let me take it—and let me pass it—one by one.

Hock Tan: Right. Stacy, you are very clever. You put two tricky questions into one. Let me take it. Let me parse it one by one. Yes, you can count the number of gigawatts we are outlining. Not every one, because we are focused on. To be fair, we have 6 customers. 4 of them are just simply going to be huge. We outline, and we walk through with you guys their journey into deploying XPUs. As I said, it is a journey. Google has been doing it for the last 10 years. Others, OpenAI, last 2, 3 years. Meta, last 3 years. Each one is doing it differently, and we like to take you guys through it. More importantly, what it means in the deployment of XPU for these 3 years, 2026, 2027, 2028. We lay down. You are right, Stacy, you add up the gigawatts.

Speaker #4: Yes, you can count the number of gigawatts we are outlining. Not everyone, because we're focused on—to be fair—six customers. Four of them are just simply going to be huge.

Speaker #4: Another 230 billion in 28, which would add up, I know, to about 350 billion. So another way of saying is, we believe pretty pretty with a pretty high degree of confidence, we will ship 350 billion dollars of AI semiconductors to this customers in the next two years.

Speaker #4: And we outline—if we walk through with you guys—their journey into deploying XPUs. As I said, it's a journey Google has been on for the last 10 years.

Speaker #4: That's the best way to look at it. It doesn't necessarily mean all 30 gigawatts need to have been deployed within that period. Turning to your more next interesting question, right, as I said, when you do an XPU, it's not only performance.

Speaker #4: Others OpenAI, last two, three years. Matter, last three years. Each one is doing it differently. And we like to take you guys through it.

Speaker #4: But more importantly, what it means in their deployment of XPU for— from '20 for the last— for these three years: '26, '27, '28. And so we laid out— and you're right, Stacey.

Speaker #4: As well, if not better, for your particular LLM workloads, for each of our customers, it's also half the cost. In fact, less than half the cost.

Speaker #4: You add up the gigawatts. So we deploy, we are showing where they are headed. What we didn't tell you specifically, when we came up to the final number, is how many of these gigawatts do actually come up for actual deployment?

Hock Tan: We deploy. We are showing where they are headed. What we did not tell you specifically when we came up to the final number is how many of these gigawatts do actually come up for actual deployment, because deployment does not just include getting the chips out there. Before you get the chips out there, you got to get the data center shell in place and ready for literary production. We are talking about fiscal years. What we are saying is, as you add up the gigawatts, we are not saying that over the next two years, 2027, 2028, that there are 30 gigawatts that will go into production, and therefore we ship it. We think, we judge it conservatively to be somewhat less.

Speaker #4: And that's you made my point exactly.

Speaker #2: Thank you. One moment for our next question. That will come from the line of Ben Reitzes with Melius. Your line is open.

Speaker #4: Because deployment doesn't just include getting the chips out there. Before you get the chips out there, you've got to get the data center shell in place.

Speaker #5: Hi. This is Jack Adair on for Ben. Thank you for the question. Can you shed a little bit more light on the maximum off balance sheet risk for the backstop agreements?

Speaker #4: And ready for, literally, production. And we're talking about fiscal years. So what we're saying is, as you add up the gigawatts, we're not saying that over the next two years—'27 and '28—that there are 30 gigawatts that will go into production.

Speaker #5: Your last Q showed that the first tranche had maximum exposure of about 29 billion. Is this about the order of magnitude for each of the Anthropic and OpenAI gigawatts over the next few years?

Speaker #4: And therefore, we ship it. We think—we judge it conservatively to be somewhat less. But we see the demand that, if they can get it all in place, and our products, and we can ship those chips, racks in place, it will be 30 gigawatts between the customers we have, the six customers we have.

Speaker #5: And could you add a little bit more color as well on the residual value and the risk associated? Thank you.

Hock Tan: But we see the demand that if they can get it all in place and our products, and we can ship those chips, racks in place, it will be 30 gigawatts between the customers we have, the six customers we have. Question is, will all 30 come into production within these two year fiscal years? We are giving you a judged number of $115 billion of chips to these guys in fiscal 2027, another $230 billion in 2028, which would add up, I know, to about $350 billion. So another way of saying is, we believe with a pretty high degree of confidence, we will ship $350 billion of AI semiconductors to these customers in the next two years. That is the best way to look at it. It does not necessarily mean all 30 gigawatts need to have been deployed within that period. Turning to your more next interesting question.

Speaker #8: Thanks, Jack. Listen, we don't have anything to announce today. On residual value guarantees or backstops. So there's nothing new to add. So the numbers that you outlined around what we've already done remain true.

Speaker #4: The question is, will all 30 come into production within these two fiscal years? And we are giving you a judge number of $115 billion of chips to these guys in fiscal '27.

Speaker #8: And as I mentioned in my prepared remarks, we're going to evaluate any strategic financing really on a deal-by-deal basis. And and we expect that any that we do in the future is they're going to have unique features and they're going to be tailored specifically to the lab and to the investor needs.

Speaker #4: Another $230 billion in '28, which would add up, I know, to about $350 billion. So another way of saying it is, we believe, with a pretty high degree of confidence, we will ship $350 billion of AI semiconductors to these customers in the next two years.

Speaker #8: So I can't give you an overarching look at, like, what's the max and what each one's going to look like. But we will tell you at the right time.

Speaker #8: We have nothing to announce today.

Speaker #6: Next question.

Speaker #2: Thank you. One moment for our next question. That will come from the line of Vivek Arya with Bank of America. Your line is open.

Speaker #4: That's the best way to look at it. It doesn't necessarily mean all 30 gigawatts need to have been deployed within that period. Turning to your next, more interesting question—right.

Speaker #7: thanks for taking my question. Amie, I just wanted to clarify any impact on, gross margins in 27 and 28 given this, XPU mix and and rising, memory cost.

Hock Tan: Right. As I said, when you do an XPU, it not only performs as well if not better for your particular LLM workloads for each of our customers, it is also half the cost. In fact, less than half the cost. That is, you made my point exactly.

Speaker #4: As I said, when you do an XPU, it's not only about performance. As well, if not better, for your particular LLM workloads, for each of our customers, it's also half the cost.

Speaker #7: And then Hock you know, you mentioned the two frontier labs will be your largest AI, you know, customers. You know, even though Google continues to say they are, supply constrained.

Speaker #7: So why aren't they you know, taking more? And in many cases, these frontier labs depend on land, power shell, from their cloud service providers.

Speaker #4: In fact, less than half the cost. And you just made my point exactly.

Speaker #7: You know, sometimes Nvidia-related entities. So how much of their user silicon is truly their choice versus what their CSP or funding partner, dictates? I'm I'm just curious.

Speaker #2: Thank you. One moment for our next question. That will come from the line of Ben Reitzes with Melius. Your line is open.

Operator: Thank you. One moment for our next question. That will come from the line of Ben Reitzes with Melius. Your line is open.

Speaker #7: you know, what gives you the certainty that they will give Broadcom, right, that specific business in in, you know, 27 or 28? I mean, so much of their funding depends on, other cloud service providers who might have a different view when it comes to the choice of silicon in that, specific facility.

Jack Adair: Hi, this is Jack Adair on for Ben. Thank you for the question. Could you shed a little bit more light on the maximum off-balance sheet risk for the backstop agreements? Your last Q showed that the first tranche had maximum exposure of about $29 billion. Is this about the order of magnitude for each of the Anthropic and OpenAI gigawatts over the next few years? Could you add a little bit more color as well on the residual value and the risk associated? Thank you.

Speaker #5: Hi, this is Jackie Darron for Ben. Thank you for the question. Can you shed a little bit more light on the maximum off-balance sheet risk for the backstop agreements?

Speaker #5: Your last question showed that the first tranche had maximum exposure of about $29 billion. Is this about the order of magnitude for each of the Anthropic and OpenAI gigawatts over the next few years?

Speaker #4: I know. The best way to answer that a couple of ways I have of answering, in that and couple of points. Don't forget the rate with the rate of growth, this, this, startups and we're talking about, you know, we have six customers.

Speaker #5: And could you add a little bit more color as well on the residual value and the risk associated? Thank you.

Speaker #7: Thanks, Jack. Listen, we don't have anything to announce today on residual value guarantees or backstops, so there's nothing new to add. The numbers that you outlined around what we've already done remain true.

Amie Thuener: Thanks, Jack. Listen, we don't have anything to announce today on residual value guarantees or backstops, so there's nothing new to add. The numbers that you outlined around what we've already done remain true. As I mentioned in my prepared remarks, we're going to evaluate any strategic financing really on a deal-by-deal basis. We expect that any that we do in the future is they're going to have unique features, and they're going to be tailored specifically to the lab and to the investor needs. So I can't give you an overarching look at what's the max and what each one's going to look like. But we will tell you at the right time. We have nothing to announce today.

Speaker #4: And we are creating this financial, vehicles as Amie described them for just two of them. Not all. They are the other four customers are ours, are financially secure, stable enough to be able to fund it themselves.

Speaker #7: And as I mentioned in my prepared remarks, we're going to evaluate any strategic financing really on a deal-by-deal basis. And we expect that any that we do in the future is they're going to have unique features and they're going to be tailored specifically to the lab and to the investor needs.

Speaker #4: And we're happy for them to do that and get them up to where they need to with technology, shared technology, which we have in plentiful supply.

Speaker #7: So I can't give you an overarching look at, like, what's the max and what each one's going to look like. But we will tell you at the right time.

Speaker #4: With these two guys Anthropic and OpenAI, I mean, these are is I thinking of I mean, you have two geniuses. In the middle of what?

Speaker #7: We have nothing to announce today.

Speaker #4: Out of Mongolia, say. And they need to go to college, to to fulfill where they want to. So we do what we can to help them.

Speaker #6: Next question.

Hock Tan: Next question.

Speaker #2: Thank you. One moment for our next question. That will come from the line of Vivek Arya with Bank of America. Your line is open.

Operator: Thank you. One moment for our next question. That will come from the line of Vivek Arya with Bank of America. Your line is open.

Speaker #4: And part of it is creating an ABAF creating sources of financing to help these companies with the leading edge, frontier models in the world be able to f to play in the same playing field and be able to offer this great technology products to the world.

Speaker #8: Thanks for taking my question. Amy, I just wanted to clarify any impact on gross margins in '27 and '28, given this XPU mix and rising memory cost.

Vivek Arya: Thanks for taking my question. Amy, I just wanted to clarify any impact on gross margins in 2027 and 2028, given this XPU mix and rise in memory cost. Hock, you mentioned the two frontier labs will be your largest AI customers, even though Google continues to say they are supply constrained. So why aren't they taking more? In many cases, these frontier labs depend on land, power, shell from their cloud service providers, sometimes Nvidia-related entities. So how much of their use of silicon is truly their choice versus what their CSP or funding partner dictates? I'm just curious, what gives you the certainty that they will give Broadcom that specific business in 2027 or 2028 when so much of their funding depends on other cloud service providers who might have a different view when it comes to the choice of silicon in that specific facility?

Speaker #8: And then Hock, you know, you mentioned the two frontier labs will be your largest AI, you know, customers. You know, even though Google continues to say they are supply constrained.

Speaker #4: And that's simply what we're doing. It's a great investment for us when you think about it, that every gigawatt of compute they deploy they could achieve 30 billion of ARR, annual re-revenue per gigawatt.

Speaker #8: So why aren't they taking more? And in many cases, these frontier labs depend on land, power, and shell from their cloud service providers, you know, sometimes NVIDIA-related entities.

Speaker #8: So how much of their use of silicon is truly their choice versus what their CSP or funding partner dictates? I'm just curious, you know, what gives you the certainty that they will give Broadcom right that specific business in, you know, 27 or 28?

Speaker #4: That's a hell of a business model so for us, that's a great investment to focus on doing. So to put that simply, that's how we look at this simple thing.

Speaker #4: It's makes economic sense for Broadcom to invest and enable these guys. Now, these guys are going to be hyperscalers in their own right. So what they do is they want first-party compute capacity.

Speaker #8: I mean, so much of their funding depends on other cloud service providers who might have a different view when it comes to the choice of silicon in that specific facility.

Speaker #4: I know. The best way to answer that—a couple of ways I have of answering that, and a couple of points. Don't forget that, with the rate of growth, these startups— and we're talking about, you know, we have six customers.

Hock Tan: I know. The best way to answer that, a couple of ways I have of answering that and a couple of points. Don't forget, with the rate of growth, these startups, and we're talking about we have six customers, and we are creating these financial vehicles, as Amy described them, for just two of them. Not all. The other four customers of ours are financially secure, stable enough to be able to fund it themselves. We're happy for them to do that and get them up to where they need to with technology, sheer technology, which we have in plentiful supply. With these two guys, Anthropic and OpenAI, it's like thinking of you have two geniuses in the middle of, what, outer Mongolia, say, and they need to go to college to fulfill where they want to.

Speaker #4: Just as they create their own they want to create silicon that is very cost performance optimized and they want to run their own data centers eventually first pass.

Speaker #4: And we are creating these financial vehicles, as Amy described them, for just two of them—not all. The other four customers of ours are financially secure and stable enough to be able to fund it themselves.

Speaker #4: Not even eventually. They want to run it as fast as they can. Short term, you're right. They are de they are using cloud services third-party services to run to deploy their models.

Speaker #4: Long term, we see these guys to be no different from a hyperscaler. And we'll run their own data centers and be first-party to offer AI generative AI APIs assess models to the world.

Speaker #4: And we're happy for them to do that and get them up to where they need to be with technology—shared technology—which we have in plentiful supply.

Speaker #4: So we see that happening. It's not speculation. It's actually happening. And we are in the midst of enabling that.

Speaker #4: With these two guys, Anthropic and OpenAI, I mean, these are who I'm thinking of. I mean, you have two geniuses in the middle of, what, Outer Mongolia, say.

Speaker #8: And I'm happy to take your gross margin question. first, I want to just make sure I correct, what I said before. Just to be super crisp.

Speaker #4: And they need to go to college to fulfill where they want to. So we do what we can to help them. And part of it is creating sources of financing to help these companies with the leading-edge frontier models in the world, to be able to play in the same playing field and be able to offer these great technology products to the world.

Speaker #8: Gross margin for our semiconductor solution segment was approximately 67% in Q4 or Q3. And to your question, gross margin, you know, really reflects revenue mix between both semiconductors and infrastructure software.

Hock Tan: So we do what we can to help them, and part of it is creating sources of financing to help these companies with the leading-edge frontier models in the world be able to play in the same playing field and be able to offer these great technology products to the world. That's simply what we're doing. It's a great investment for us when you think about it, that every gigawatt of compute they deploy, they could achieve $30 billion of ARR, annual revenue per gigawatt. That's a hell of a business model. So for us, that's a great investment to focus on doing. To put that simply, that's how we look at this simple thing. It makes economic sense for Broadcom to invest and enable these guys. Now, these guys are going to be hyperscalers in their own right.

Speaker #8: And then within semiconductors, it also is the product mix is reflected. as well as increasing memory content. And so as you can see, as XPUs become a larger proportion of our total revenue mix, it impacts our margin.

Speaker #4: And that's simply what we're doing. It's a great investment for us. When you think about it, every gigawatt of compute they deploy could achieve $30 billion of ARR—annual revenue per gigawatt.

Speaker #8: We guide one core at a time. So we'll we'll tell you each quarter. You know, how, what our margin's going to look like.

Speaker #4: Yeah. It's not the first time we've told you guys that. Because stop focusing on gross margin. It's what we're saying. Look at where it matters.

Speaker #4: That's a hell of a business model. So for us, that's a great investment to focus on doing. To put it simply, that's how we look at this simple thing.

Speaker #4: Operating margin at the bottom of the at the end of the day. Because the growth in revenue far out, surpasses out, their growth in OPEX, operating spending to support their growth in revenue.

Speaker #4: It makes economic sense for Broadcom to invest and enable these guys. Now, these guys are going to be hyperscalers in their own right. So what they do is they want first-party compute capacity.

Speaker #4: So we have a lot of accretion level operating leverage we call it on operating margin. So we expect to be able to sustain operating margin even as mix of products dilutes the gross margin.

Hock Tan: What they do is they want first-party compute capacity. Just as they want to create silicon that is very cost performance optimized, and they want to run their own data centers eventually first path, not even eventually. They want to run it as fast as they can. Short term, you're right. They are using cloud services, third-party services to deploy their models. Long term, we see these guys to be no different from a hyperscaler and will run their own data centers and be first party to offer AI, generative AI, APIs, assess models to the world. So we see that happening. It's not speculation. It's actually happening, and we are in the midst of enabling that.

Speaker #4: Just as they create their own, they want to create silicon that is very cost-performance optimized. And they want to run their own data centers eventually—first batch.

Speaker #2: Thank you. One moment for our next question. And that will come from the line of Tom O'Malley with Barclays. Your line is open.

Speaker #4: Not even eventually. They want to run it as fast as they can. Short term, you're right. They are using cloud services, third-party services, to run and deploy their models.

Speaker #5: Thanks for taking the question. This one's for Hock and Charlie. So you guys talked a bit about the Tomahawk Ultra. you're ramping a variety of ASICs over the next couple of years.

Speaker #5: How should we think about the attach rate of Tomahawk Ultra to the ASICs that you're developing? I would assume that, you know, when a customer goes down the road with you for an ASIC, they would decide to use your networking as well.

Speaker #4: Long term, we see these guys to be no different from a hyperscaler. And they'll run their own data centers and be first party to offer generative AI APIs, access models to the world.

Speaker #5: maybe some kind of forecast as to how many of those customers are using your Tomahawk Ultra. How many are using Ethernet? Just because I assume your penetration with your accelerators is going to lead to a lot of success on the networking side.

Speaker #4: So we see that happening. It's not speculation—it's actually happening. And we are in the midst of enabling that.

Speaker #5: Any help there would be great. Thank you.

Speaker #4: Go ahead, Charlie. Thank you, Hock. Yeah. on the attach rate for scale up, today we're actually seeing customers who build XPUs with us deploy either Tomahawk 6 that used to deploy Tomahawk 5.

Amie Thuener: I am happy to take your gross margin question. First, I want to make sure I correct what I said before, just to be super crisp. Gross margin for our Semiconductor Solutions segment was approximately 67% in Q3. To your question, gross margin really reflects revenue mix between both semiconductors and infrastructure software. Within semiconductors, the product mix is reflected as well as increasing memory content. As you can see, as XPUs become a larger proportion of our total revenue mix, it impacts our margin. We guide one quarter at a time, so we will tell you each quarter what our margin is going to look like.

Speaker #7: And I'm happy to take your gross margin question. First, I want to just make sure I correct what I said before. Just to be super crisp, gross margin for our Semiconductor Solutions segment was approximately 67% in Q4 or Q3.

Speaker #4: Now they're going to Tomahawk 6. And some of them are going to Tomahawk Ultra. So if you look at where we're seeing XPUs and even GPUs, we're seeing scale up solutions adopt both Tomahawk 6 and Tomahawk Ultra.

Speaker #7: And to your question, gross margin, you know, really reflects revenue mix between both semiconductors and infrastructure software. And then, within semiconductors, the product mix is also reflected.

Speaker #4: And the beauty and the reason why they're doing this, it's an as Hock, articulated earlier on, it's Ethernet based, which means it's open. Anybody can connect to it, especially with the standards that we have collaborated with the entire entire industry on.

Speaker #7: As well as increasing memory content. And so, as you can see, as XPUs become a larger proportion of our total revenue mix, it impacts our margin.

Speaker #7: We guide one quarter at a time. So we'll tell you each quarter, you know, what our margin is going to look like.

Speaker #4: So at this point in time, we're seeing it pretty much deployed in in both XPU clusters and in some GPU clusters.

Speaker #6: Yeah, it's not the first time we've told you guys that. Stop focusing on gross margin—it's what we're saying. Look at where it matters.

Hock Tan: Yeah. It is not the first time we have told you guys that, because stop focusing on gross margin. It is what we are saying. Look at R&D methods operating margin at the end of the day, because the growth in revenue far surpasses their growth in OpEx, operating spending to support their growth in revenue. So we have a lot of accretion, operating leverage, we call it, on operating margin. So we expect to be able to sustain operating margin even as mix of products dilutes the gross margin.

Speaker #6: Operating margin at the bottom, at the end of the day, is because the growth in revenue far surpasses their growth in opex—operating spending to support their growth in revenue.

Speaker #2: Thank you. One moment for our next question. That will come from the line of Will Stein with Truist Securities. Your line is open.

Speaker #3: Great. Thanks for taking my question. Congrats on the good results and the, very impressive longer-term outlook you gave. Hock, I hope you can tell us a little bit about, the constraints as it relates to land, power, and shell.

Speaker #6: So, we have a lot of accretion-level operating leverage, as we call it, on operating margin. So, we expect to be able to sustain operating margin even as the mix of products dilutes the gross margin.

Speaker #3: you mentioned this a little bit earlier, but I wonder to what degree the outlook is sensitized to, potential constraints in that area. Thank you.

Speaker #1: Thank you. One moment for our next question. That will come from the line of Tom O'Malley with Barclays. Your line is open.

Operator: Thank you. One moment for our next question. That will come from the line of Tom O'Malley with Barclays. Your line is open.

Speaker #4: good question, Will. Of course, we have to be realistic. And be and when we provide our outlook, we not just look at, just look at what our customer our end customer, one of those at our frontier models, just asks of us for compute capa compute capacity in the form of chips, or in some cases even in the form of racks.

Speaker #5: Thanks for taking the question. This one's for Hock and Charlie. So, you guys talked a bit about the Tomahawk Ultra. You're ramping a variety of ASICs over the next couple of years.

Tom O'Malley: Thanks for taking the question. This one is for Hock and Charlie. You guys talked a bit about the Tomahawk Ultra. You are ramping a variety of ASICs over the next couple of years. How should we think about the attach rate of Tomahawk Ultra to the ASICs that you are developing? I would assume that, when a customer goes down the road with you for an ASIC, they would decide to use your networking as well. Maybe some kind of forecast as to how many of those customers are using your Tomahawk Ultra, how many are using Ethernet, just because I assume your penetration with your accelerators is going to lead to a lot of success on the networking side. Any help there would be great. Thank you.

Speaker #5: Maybe some kind of forecast as to how many of those customers are using your Tomahawk Ultra, how many are using Ethernet? Just because I assume your penetration with your accelerators is going to lead to a lot of success on the networking side.

Speaker #4: We are very engaged with each of them on how much or which how much LPS, land, power, and sites, and, and systems they have.

Speaker #5: Any help there would be great. Thank you.

Speaker #4: Go ahead, Charlie.

Hock Tan: Go ahead, Charlie.

Charlie Kawwas: Okay. Thank you, Hock. On the attach rate for scale up, today we are actually seeing customers who build XPUs with us deploy either Tomahawk 6. They used to deploy Tomahawk 5, now they are going to Tomahawk 6, and some of them are going to Tomahawk Ultra. So if you look at where we are seeing XPUs and even GPUs, we are seeing scale-up solutions adopt both Tomahawk 6 and Tomahawk Ultra. And the beauty and the reason why they are doing this, it is, as Hock articulated earlier on, it is Ethernet based, which means it is open. Anybody can connect to it, especially with the standards that we have collaborated with the entire industry on. So at this point in time, we are seeing it pretty much deployed in both XPU clusters and in some GPU clusters.

Speaker #6: Okay. Thank you, Hock. Yeah, on the attach rate for scale-up, today we're actually seeing customers who build XPUs with us deploy either Tomahawk 6.

Speaker #4: Before, we actually believed it will happen. Because this some this part on, sites PowerShell has a long lead time. The construction project, you correctly indicate that.

Speaker #6: They used to deploy Tomahawk 5. Now they're going to Tomahawk 6, and some of them are going to Tomahawk Ultra. So if you look at where we're seeing XPUs and even GPUs, we're seeing scale-up solutions adopt both Tomahawk 6 and Tomahawk Ultra.

Speaker #4: So we do that. And, we do that dev not development, but analysis with our customer. And we reflect that in the forecast outlook we're giving you today.

Speaker #3: Thank you.

Speaker #6: And the beauty and the reason why they're doing this is, as Hock articulated earlier, it's Ethernet-based, which means it's open. Anybody can connect to it, especially with the standards that we have collaborated on with the entire industry.

Speaker #2: Thank you. One moment for our next question. And that will come from the line of Joshua Buchalter with TD Cowwen. Your line is open.

Speaker #6: Hey, guys. Thank you for taking my question. you know, with with the XPV, I think you have 35 billion of the financing secured. You know, as we think about the 10-gigawatts for Anthropic and, and the 5 for OpenAI in fiscal 2028, are you expecting most or all of this to be financed through the XPV?

Speaker #6: So at this point in time, we're seeing it pretty much deployed in both XPU clusters and in some GPU clusters.

Speaker #6: And, and can you give us sort of any help on the timeline and hurdles required to secure financing as we look forward to, to those deployments getting secured?

Speaker #1: Thank you. One moment for our next question. That will come from the line of Will Stein with Truist Securities. Your line is open.

Operator: Thank you. One moment for our next question. That will come from the line of Will Stein with Truist Securities. Your line is open.

Speaker #6: Thank you.

Speaker #4: Let me start broadly and Amie will give you a specifics. Not everyone not every site will be the same. Because don't forget, next two years, we'll see things happen between from an with Anthropic and OpenAI.

Speaker #2: Great. Thanks for taking my question. Congrats on the good results, and the very impressive longer-term outlook you gave. Hock, I hope you can tell us a little bit about the constraints as it relates to land, power, and shell.

Will Stein: Great. Thanks for taking my question. Congrats on the good results and the very impressive longer term outlook you gave. Hock, I hope you can tell us a little about the constraints as it relates to land, power, and shell. You mentioned this a little earlier, but I wonder to what degree the outlook is sensitized to potential constraints in that area. Thank you.

Speaker #4: And, you know, and it's open it's an open secret. Anthropic is well on its way to an IPO. And with that, it's investment credit will change.

Speaker #2: You mentioned this a little bit earlier, but I wonder to what degree the outlook is sensitized to potential constraints in that area. Thank you.

Speaker #4: OpenAI, we still don't have less clarity, but it's different. And so with that, let me pass it to Amie.

Speaker #6: Good question, Will. Of course, we have to be realistic and be and when we provide our outlook, we not just look at just look at what our customer our end customer, one of those out of frontier models, just asks of us for compute capacity in the form of chips, or in some cases even in the form of racks.

Hock Tan: Good question, Will. Of course, we have to be realistic and when we provide our outlook, we not just look at what our end customer, one of those LLM frontier models, just asks of us for compute capacity in the form of chips or in some cases even in form of racks. We are very engaged with each of them on how much LPS, land, power, and systems they have before we actually believe it will happen. Because this part on site's power shell has a long lead time, the construction project, you correctly indicate that. We do that, and we do that, not development, but analysis with our customer, and we reflect that in the forecast outlook we are giving you today.

Speaker #7: I, I mean, I think just to double down on that, I think in every scenario, our partners were going to be looking for financing from lots of different sources.

Speaker #7: In some cases, we may step in and help out if it makes sense to us and if it's within our, our, our framework. And in other cases, we may not.

Speaker #7: But we're going to look at it case by case. And evaluate, you know, evaluate the specific needs.

Speaker #6: We are very engaged with each of them on how much, or which, how much LPS, land, power, and sites, ITs, and systems they have.

Speaker #3: Thank you both.

Speaker #2: One moment for our next question. And that will come from the line of Jim Schneider with Goldman Sachs. Your line is open.

Speaker #8: Thanks. And thanks for taking my question. I was wondering if you could maybe address some of the other constraints that are impacting your outlook.

Speaker #6: Before we actually believe it will happen, because some of this, in parts on-site, PowerShell has a long lead time. The construction project, you correctly indicate that.

Speaker #8: You mentioned land, power, shell. Do you think that is the largest of the constraints that you face heading into fiscal 27? And specifically, can you address any other supply chain constraints?

Speaker #6: So we do that. And we do that not in development, but in analysis with our customer, and we reflect that in the forecast outlook we're giving you today.

Speaker #8: Whether it be memory, substrates, or other components that could be impacting your outlook. And to what extent you could expect those to get, ameliorated?

Speaker #8: Thank you.

Speaker #2: Thank you.

Will Stein: Thank you.

Speaker #4: Thank you, Jim. There's a hell of a question. And you're right in all counts. It's a multi-dimensional issue. That and to get to get a date to get a AI data center deployed, yeah, we now have almost sensitive about especially at scale.

Speaker #1: Thank you. One moment for our next question. And that will come from the line of Joshua Buckhalter with TD Cowen. Your line is open.

Operator: Thank you. One moment for our next question. That will come from the line of Joshua Buchalter with TD Cowen. Your line is open.

Speaker #5: Hey, guys. Thank you for taking my question. With the XPV, I think you have $35 billion of the financing secured. As we think about the 10 gigawatts for Anthropic and the 5 for OpenAI in fiscal 2028, are you expecting most or all of this to be financed through the XPV?

Joshua Buchalter: Hey, guys. Thank you for taking my question. With XPV, I think you have $35 billion of the financing secured. As we think about the 10 gigawatts for Anthropic and the five for OpenAI in fiscal 2028, are you expecting most or all of this to be financed through the XPV? Can you give us any help on the timeline and hurdles required to secure financing as we look forward to those deployments getting secured? Thank you.

Speaker #4: And the scale we are now delivering at scale to our customers. XPUs, XP data AI data center, built upon XPU computing accelerators. Land, power, and shell, as I indicated to question with Will, is a is a big concern.

Speaker #5: And can you give us any help on the timeline and hurdles required to secure financing as we look forward to those deployments getting secured?

Speaker #5: Thank you.

Speaker #6: Let me start broadly, and Amy will give you the specifics. Not everyone—not every site—will be the same. Because don't forget, over the next two years, we'll see things happen between, from, with Anthropic and OpenAI.

Hock Tan: Let me start broadly, and Amie will give you specifics. Not every site will be the same. Because don't forget, next 2 years, we'll see things happen with Anthropic and OpenAI. It's an open secret, Anthropic is well on its way to an IPO. With that, its investment credit will change. OpenAI, we still don't have less clarity, but it's different. With that, let me pass it to Amie.

Speaker #4: It more than a big concern, it de dictates specific timing. Of when this ca-capacity gets deployed and be available. But very much in the mix that we have to think about with our customers.

Speaker #6: And you know, it's an open secret—Anthropic is well on its way to an IPO. And with that, its investment credit will change.

Speaker #4: And it's a it's a joint collaborative effort. Not just one-directional. It's you said it correctly. Leading edge silicon. Because for one, to produce to get the chip produced, and availability of those chips, and the time it comes in.

Speaker #6: OpenAI, we still don't have more clarity, but it's different. And so, with that, let me pass it to Amy.

Amie Thuener: I think just to double down on that, I think in every scenario, our partners were going to be looking for financing from lots of different sources. In some cases, we may step in and help out if it makes sense to us and if it's within our framework. In other cases, we may not, but we're going to look at it case by case and evaluate the specific needs.

Speaker #3: I mean, I think, just to double down on that, in every scenario, our partners are going to be looking for financing from lots of different sources.

Speaker #4: Then even deeper than that, we talk about substrates. Which is a specific issues. Which is leading us to build our own substrate capacity at scale in s in our factory in s in Singapore.

Speaker #3: In some cases, we may step in and help out if it makes sense to us and if it's within our framework, and in other cases, we may not.

Speaker #3: But we're going to look at it case by case and evaluate, you know, evaluate the specific needs.

Speaker #4: Jointly with one of our partners. And of course, we all know about memory, HBM memory, and beyond HBM memory, the system memory that goes into a AI server.

Speaker #2: Thank you both.

Hock Tan: Thank you both.

Speaker #1: One moment for our next question. That will come from the line of Jim Schneider with Goldman Sachs. Your line is open.

Operator: One moment for our next question. That will come from the line of James Schneider with Goldman Sachs. Your line is open.

Speaker #4: Which we don't supply necessarily. But our customers have to secure too. So all these are multi it's not multi-dimensional problem. Coming from various sources.

Speaker #7: Thank you, and thanks for taking my question. I was wondering if you could maybe address some of the other constraints that are impacting your outlook.

James Schneider: Thanks, and thanks for taking my question. I was wondering if you could maybe address some of the other constraints that are impacting your outlook. You mentioned land, power, shell. Do you think that is the largest of the constraints that you face heading into fiscal 2027? Specifically, can you address any other supply chain constraints, whether it be memory, substrates, or other components that could be impacting your outlook? To what extent you could expect those to get ameliorated. Thank you.

Speaker #7: You mentioned land power shell. Do you think that is the largest of the constraints that you face heading into fiscal '27? And specifically, can you address any other supply chain constraints?

Speaker #4: And then depending on different times, each might become a bottleneck. And so it's a constant interesting challenge. As we work this through, which is why in some ways, I'm so glad we have only six customers to deal with.

Speaker #7: Whether it be memory, substrates, or other components that could be impacting your outlook, and to what extent you could expect those to get ameliorated?

Speaker #2: Thank you. Our next question that will come from the line of Vijay Rakesh with Mizuho. Your line is open.

Speaker #7: Thank you.

Speaker #6: Thank you, Jim. There's a hell of a question. And you're right in all counts. It's a multi-dimensional issue. That to get to get a to get an AI data center deployed, yeah, we now have almost sensitive about especially at scale and the scale we are now delivering at scale to our customers.

Hock Tan: Thank you, Jim. That's a hell of a question, and you're right in all counts. It's a multidimensional issue to get an AI data center deployed. We now are all more sensitive about, especially at scale, and we are now delivering at scale to our customers XPUs, AI data center built upon XPU computing accelerators. Land, power, and shell, as I indicated to a question with Will, is a big concern. It's more than a big concern, it dictates specific timing of when this capacity gets deployed and be available. Very much in the mix that we have to think about with our customers, and it's a joint collaborative effort, not just one direction. It's, you said it correctly, leading-edge silicon. For one, to get the chip produced and availability of those chips and the time it comes in.

Speaker #6: Yeah. Hi. Thanks, Hock and Amie. just a quick question. thanks for giving the, visibility on fiscal 27 and fiscal 28 AI revenues. just a quick question.

Speaker #6: As you go through, through subsequent generations of XPU, can you talk to how your dollar per gigawatt should, improve, you know, into the suc subsequent XPU generations?

Speaker #6: XPU's data AI data center is built upon XPU computing accelerators. Land, power, and shell, as I indicated in response to Will's question, is a big concern.

Speaker #6: And also, I saw your capex went up. Just wondering if you're adding capacity on the EML, CW, indium phosphides, right? Thanks.

Speaker #6: It's more than a big concern. It dictates specific timing—of when this capacity gets deployed and becomes available. But it's very much in the mix that we have to think about with our customers.

Speaker #4: Well, well, Charlie, you wanted to take the capex issue?

Speaker #5: Sure.

Speaker #4: Particularly?

Speaker #5: We, so on the capex side, as, we've been sharing with you, Hock and I, for several quarters, we continue to invest in our factories.

Speaker #5: Substrate is what Hock talked about, which was actually going in production shortly. but also, our EML, CW, and Vixel factories, our indium phosphide factories, both in the US as well as in Singapore, were actually more than tripling them year on year.

Speaker #6: And it's a—it's a joint collaborative effort, not just one-directional. It's—you said it correctly—leading-edge silicon. Because, for one, to produce, to get the chip produced, and the availability of those chips and the time it comes in.

Speaker #5: So we've already expanded the capacity for this year. And we're increasing it significantly for the next two years. And that's part of what you actually are seeing in capex.

Speaker #6: Then, even deeper than that, we talk about substrates, which is a specific issue, which is leading us to build our own substrate capacity at scale in our factory in Singapore.

Hock Tan: Even deeper than that, we talk about substrates, which is a specific issue, which is leading us to build our own substrate capacity at scale in our factory in Singapore, jointly with one of our partners. Of course, we all know about memory, HBM memory. Beyond HBM memory, the system memory that goes into AI servers, which we don't supply necessarily, but our customers have to secure too. All these are multidimensional problems coming from various sources. Depending on different times, each might become a bottleneck. It's a constant interesting challenge as we work this through, which is why, in some ways, I'm so glad we have only six customers to deal with.

Speaker #4: Yeah. And that's this one we I indicated in my remarks too, which is we I know demand for lasers. Whether it's, you know, EML lasers, CW lasers, is far surpassing supply out there in the industry.

Speaker #6: Jointly with one of our partners. And of course, we all know about memory, HBM memory, and beyond HBM memory, the system memory that goes into AI servers.

Speaker #4: So we are doing a part to it to double down, actually, on capacity and with with a fairly substantial share of this market to to us in our interest to enable this ecosystem of growth.

Speaker #6: Which we don't supply necessarily, but our customers have to secure, too. So all these are—it's not a multi-dimensional problem. They're coming from various sources.

Speaker #6: And then, depending on different times, each might become a bottleneck. And so it's a constantly interesting challenge as we work this through—which is why, in some ways, I'm so glad we have only six customers to deal with.

Speaker #4: On your earlier question, a gigawatt dollars per gigawatt, it's very interesting what you're saying. because keep in mind something that is very interesting too, which is you're right.

Speaker #4: We've every generation of XPU or GPU, you know, the, the performance increases. And theref and therefore, and the s-silicon goes further leading edge. More expensive to produce.

Speaker #1: Thank you. Our next question will come from the line of Vijay Rakesh with Mizuho. Your line is open.

Operator: Thank you. Our next question, that will come from the line of Vijay Rakesh with Mizuho. Your line is open.

Speaker #5: Yeah, hi. Thanks, Hock and Amy. Just a quick question—thanks for giving the visibility on fiscal '27 and fiscal '28 AI revenues. Just a quick question.

Vijay Rakesh: Yeah. Hi, and thanks, Hock and Amie. Just a quick question. Thanks for giving the visibility on fiscal 2027 and fiscal 2028 AI revenues. Just a quick question. As you go through subsequent generations of XPU, can you talk to how your dollar per gigawatt should improve into the subsequent XPU generations? Also, I saw your CapEx went up. Just wondering if you are adding capacity on the EML CW indium phosphide side. Thanks.

Speaker #4: And so ASPs go up per XPU and per GPU. But keep in mind, as they become higher performance, their power increase. Per chip, per XPU or GPU, which means they're less of it of a more a new advanced XPU in one gigawatt.

Speaker #5: As you go through subsequent generations of XPU, can you talk to how your dollar per gigawatt should improve into the subsequent XPU generations?

Speaker #5: And also, I saw your capex went up. Just wondering if you're adding capacity on the EML/CW indium phosphide site. Thanks.

Speaker #4: So what we are seeing per, per, per gigawatt is in the range of less than 30 billion dollars 20 to 30 billion dollars per gigawatt.

Hock Tan: Well, Charlie, you want to take the CapEx issue?

Speaker #6: Well, well. Charlie, you wanted to take the capex issue?

Speaker #2: Sure.

Charlie Kawwas: Sure. On the CapEx side, as we have been sharing with you, Hock and I, for several quarters, we continue to invest in our factories. Substrate is what Hock talked about, which was actually going in production shortly. Also our EML CW and VCSEL factories, our indium phosphide factories, both in the US as well as in Singapore, we are actually more than tripling them year on year. So we have already expanded the capacity for this year, and we are increasing it significantly for the next two years, and that is part of what you actually are seeing in CapEx.

Speaker #6: So on the CapEx side, as we've been sharing with you, Hock and I, for several quarters, we continue to invest in our factories.

Speaker #4: And we expect that to be very sustaining in that level because the power of each chip goes up. So even as increase the price of the chip, the, the content in dollars is relatively stable.

Speaker #6: Substrate is what Hock talked about, which was actually going into production shortly. But also, our EML, CW, and VCSEL factories—our indium phosphide factories, both in the US as well as in Singapore—were actually more than tripling them year on year.

Speaker #4: Just except the fact that there's going to be a lot more gigawatts out there. But the dollars per gigawatt will remain in the 20 to 30 billion dollar level.

Speaker #6: So, we've already expanded the capacity for this year, and we're increasing it significantly for the next two years. That's part of what you're actually seeing in CapEx.

Speaker #4: Of our content, but we have seen and we expect to see an acceleration in the number of gigawatts just to address the exponential compute demand required by our customers.

Speaker #6: Yeah. And that's this one I indicated in my remarks, too, which is, I know demand for lasers, whether it's EML lasers, CW lasers, is far surpassing supply out there in the industry.

Hock Tan: Yeah. This one I indicated in my remarks too, which is demand for lasers, whether it is EML lasers, CW lasers, is far surpassing supply out there in the industry. So we are doing our part to double down actually on capacity, and we have a fairly substantial share of this market. So to us, in our interest, we enable this ecosystem of growth. On your earlier question of dollars per gigawatt, it is very interesting what you are saying. Because keep in mind something that is very interesting too, which is, you are right. With every generation of XPU or GPU, the performance increases. Therefore, the silicon goes for the leading edge, more expensive to produce. So ASPs go up per XPU and per GPU.

Speaker #6: Got it. Thanks. Very interesting, Hock and, Charlie. Thank you.

Speaker #2: Thank you. That is all the time we have today for Q&A. I would now like to turn the call back over to Ji Yoo for any closing remarks.

Speaker #6: So we are doing a part two to double down, actually, on capacity and with a fairly substantial share of this market to us. It is in our interest to enable this ecosystem of growth.

Speaker #3: Thank you, Sherri. This quarter, Broadcom will be presenting at the Goldman Sachs Cominocopia and Technology Conference on Tuesday, September 8th. Broadcom currently plans to report its earnings for the fourth quarter and fiscal year 2026 after close of market on Wednesday, December 9th, 2026.

Speaker #6: On your earlier question, a gigawatt dollars per gigawatt—it's very interesting what you're saying because, keep in mind, something that is very interesting too, which is you're right.

Speaker #3: A public webcast of Broadcom's earnings conference call will follow at 2:00 PM Pacific. That will conclude our earnings call today. Thank you all for joining.

Speaker #6: With every generation of XPU or GPU, you know the performance increases, and therefore the silicon goes further—leading edge, more expensive to produce.

Speaker #3: Sherri, you may end the call.

Speaker #6: And so ASPs go up per XPU and per GPU. But keep in mind, as they become higher performance, their power increases per chip, per XPU or GPU, which means there's less of a more advanced XPU in one gigawatt.

Hock Tan: But keep in mind, as they become higher performance, their power increase per chip, per XPU or GPU, which means there are less of a more advanced XPU in 1 gigawatt. So what we are seeing per gigawatt is in the range of less than $30 billion, $20 to $30 billion per gigawatt. And we expect that to be very sustaining in that level because the power of each chip goes up. So even as increase the price of the chip, the content in dollars is relatively stable. Just accept the fact that there is going to be a lot more gigawatts out there. But the dollars per gigawatt will remain in the $20 to $30 billion level of our content. But we have seen and we expect to see an acceleration in the number of gigawatts just to address the exponential compute demand required by our customers.

Speaker #6: So what we are seeing per per per gigawatt is in the range of less than 30 billion dollars 20 to 30 billion dollars per gigawatt.

Speaker #6: And we expect that to be very sustaining at that level because the power of each chip goes up. So even as we increase the price of the chip, the content in dollars is relatively stable.

Speaker #6: Just accept the fact that there's going to be a lot more gigawatts out there. But the dollars per gigawatt will remain in the $20 to $30 billion level.

Speaker #6: ...of our content. But we have seen, and we expect to see, an acceleration in the number of gigawatts just to address the exponential compute demand required by our customers.

Vijay Rakesh: Got it. Thanks. Very interesting, Hock and Charlie. Thank you.

Speaker #5: Got it. Thanks. Very interesting, Hock and Charlie. Thank you.

Speaker #1: Thank you. That is all the time we have today for Q&A. I would now like to turn the call back over to Ji Yoo for any closing remarks.

Operator: Thank you. That is all the time we have today for Q&A. I would now like to turn the call back over to Ji Yoo for any closing remarks.

Speaker #3: Thank you, Sherri. This quarter, Broadcom will be presenting at the Goldman Sachs Communacopia and Technology Conference on Tuesday, September 8th. Broadcom currently plans to report its earnings for the fourth quarter and fiscal year 2026 after the close of market on Wednesday, December 9th, 2026.

Ji Yoo: Thank you, Cherie. This quarter, Broadcom will be presenting at the Goldman Sachs Communacopia and Technology Conference on Tuesday, 8 September. Broadcom currently plans to report its earnings for Q4 and fiscal year 2026 after close of market on Wednesday, 9 December 2026. A public webcast of Broadcom's earnings conference call will follow at 2:00 PM Pacific. That will conclude our earnings call today. Thank you all for joining. Cherie, you may end the call.

Speaker #3: A public webcast of Broadcom's earnings conference call will follow at 2:00 p.m. Pacific. That will conclude our earnings call today. Thank you all for joining.

Speaker #3: Sherri, you may end the call.

Operator: Thank you all for participating. This concludes today's program. You may now disconnect.