Broadcom delivered record results driven by explosive growth in AI semiconductor revenue and issued extraordinary multi-year guidance, projecting AI revenue to double in each of the next two fiscal years. The company highlighted strong demand from its six XPU customers and detailed ambitious deployment plans, with a major focus on overcoming supply chain constraints. Q3 FY2026 AI semiconductor revenue was $16.7 billion, up 221% year-over-year, with Q4 guidance implying $21.7 billion.
Buzzberg readFiscal 2028 AI revenue to double to $230BBroadcom delivered record results driven by explosive growth in AI semiconductor revenue and issued extraordinary multi-year guidance, projecting AI revenue to double in each of the next two fiscal years. The company highlighted strong demand from its six XPU customers and detailed ambitious deployment plans, with a major focus on overcoming supply chain constraints. Q3 FY2026 AI semiconductor revenue was $16.7 billion, up 221% year-over-year, with Q4 guidance implying $21.7 billion.Read full analysisCollapse analysis
Broadcom delivered record results driven by explosive growth in AI semiconductor revenue and issued extraordinary multi-year guidance, projecting AI revenue to double in each of the next two fiscal years. The company highlighted strong demand from its six XPU customers and detailed ambitious deployment plans, with a major focus on overcoming supply chain constraints. Q3 FY2026 AI semiconductor revenue was $16.7 billion, up 221% year-over-year, with Q4 guidance implying $21.7 billion.
Management provided first-time long-term guidance: AI semiconductor revenue of ~$115 billion in FY2027 and ~$230 billion in FY2028.
Anthropic is on track to become Broadcom's largest XPU customer by 2027 with planned gigawatt-scale TPU deployments.
OpenAI is ramping its custom 'Jalapeno' accelerator, which Broadcom claims outperforms Nvidia's current GPUs for specific workloads at half the cost.
AI revenue$16.7B
+55% QoQ
SEMICONDUCTOR revenue$20.8B
+39% QoQ
SOFTWARE revenue$8.8B
+22% QoQ
Revenue$29.591B
+33% QoQ
What changed this quarter
01
Guidance
Fiscal 2028 AI revenue to double to $230B
Guidance · revenue to $34.8B
02
AI
OpenAI's Jalapeno outperforms Grace Blackwell for inference
Management described exceptional AI demand, with Q3 AI semiconductor revenue up 221% year-on-year and Q4 guidance of $21.7B, up 236%. They provided multi-year outlooks of approximately $115B in fiscal 2027 and $230B in fiscal 2028, driven by six XPU customers, and highlighted…
03
AI
Entropic to become largest XPU customer in 2027
Management described exceptional AI demand, with Q3 AI semiconductor revenue up 221% year-on-year and Q4 guidance of $21.7B, up 236%. They provided multi-year outlooks of approximately $115B in fiscal 2027 and $230B in fiscal 2028, driven by six XPU customers, and highlighted…
04
Networking
Tomahawk 6 nearly sold out
Anthropic is on track to become Broadcom's largest XPU customer by 2027 with planned gigawatt-scale TPU deployments.
AI, capex & demand read
AI
Platform & monetization
Management described exceptional AI demand, with Q3 AI semiconductor revenue up 221% year-on-year and Q4 guidance of $21.7B, up 236%. They provided multi-year outlooks of approximately $115B in fiscal 2027 and $230B in fiscal 2028, driven by six XPU customers, and highlighted that custom accelerators outperform GPUs for their specific workloads.
Demand
Bookings & conversion
Management said 'Q3 demand was simply hot' and that they are 'just getting started.' They highlighted that demand from XPU customers is exponential and that their demand exceeds their supply-constrained outlook for fiscal 2027.
Capex
Investment and capacity
Management noted Q4 capital expenditures of $1.4B as they invest in semiconductor capacity, including substrate manufacturing in Singapore and expansion of optical component (EML, CW, VCSEL) factories, with capacity planned to more than triple year-on-year.
Tone · Upbeat
Management emphasized record results, rapidly accelerating AI growth, and provided multi-year revenue targets with confidence that they will be met or exceeded.
Bottlenecks
Manufacturing capacitypersistent
Substrate capacity is a bottleneck, prompting Broadcom to build its own substrate factory in Singapore.
Substrate availability constrains Broadcom's ability to expand XPU shipments, and the new Singapore capacity will add supply starting fiscal 2027.
“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.”
“Land power and shell, as I indicated to question with Will, is a big concern. More than a big concern, it dictates specific timing of when This capacity gets deployed and be available.”
“And of course, we all know about memory, HBM memory, and beyond HBM memory, the system memory that goes into AI servers. which we don't supply necessarily but our customers have to secure too.”
Broadcom's custom accelerators are designed to have roughly half the cost of merchant GPUs (like Nvidia's) while delivering comparable or better performance for specific workloads.
Broadcom is building its own advanced substrate capacity in Singapore to address a key supply chain bottleneck, with production starting in fiscal 2027.
“We are going to start deploying our Singapore fat for substrate, by the way, starting fiscal 27. And that would, I guess, address a key part of our supply bottlenecks.”
Broadcom's custom Ethernet switch 'Tomahawk Ultra' is enabling scale-up networking within GPU/XPU clusters for the first time, potentially disrupting proprietary GPU interconnect solutions.
“Tomahawk Ultra is that we are enabling now scaling up within cluster of GPU, XPU in the rack on basically Ethernet for the first time because it won't, Tomahawk Ultra will perform just as well as anything else that exists prior to that bef…”
Google is a key and growing custom accelerator (TPU) customer, with a new long-term agreement for multi-tens of billions of dollars of TPUs annually, indicating deepening reliance on Broadcom's custom silicon and networking.
… 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're 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 Entropic and Google. 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 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. In Q3, We also ship Jalapeno, OpenAI's first generation custom accelerator which outperforms Grace Blackwell GPU for inference workloads. …
ANTHROPIC
Private company
Customers
Anthropic (Entropic) is scaling its custom TPU deployments aggressively, becoming Broadcom's largest XPU customer by 2027, with planned gigawatt-scale deployments indicating massive demand for custom accelerators.
… design, including industry-leading SerDes, chip-to-chip interconnect, leading-edge HPM 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. We found the need for re-smiths. We believe these are very deep modes for any competitor to cross. Moving on to Entropic. Starting with the one gigawatt of iron wood we are deploying in 2026, we expect Entropic to deploy another five gigawatts of TPU V version 8i in 2027. And in 2028, we have clear line of sight to deliver an incremental 10 gigawatts Even as we expect Google to grow for us, Entropic is on track to become our largest XPU customer in 2027 and sustain that in 2028. 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 KBAR. 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. In addition, we are in development with …
OPENAI
Private company
Customers
OpenAI is ramping production of its custom accelerator 'Jalapeno' with Broadcom, and plans to deploy over 5 gigawatts by 2028, making it a top-two customer and signaling a major shift away from merchant GPUs.
… 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 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. In Q3, We also ship Jalapeno, 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 three and a half times year on year and represented 73% of AI revenue during the quarter. Our AI networking revenue was up over two and a half times year on year. And this strong momentum continues into Q4. During the quarter, we expect to accelerate shipments of Ironwood to Entropic. We also expect to ramp up high volume shipments of the TPU version 8i to Google. Shipments of Jalapeno for OpenAI will continue. And for Matter, we expect production shipments of their custom MTIA accelerator optimized for inference and recommendation at scale. In …
Meta is on track to receive three generations of custom MTIA accelerators from Broadcom, representing a significant and growing source of custom silicon demand.
… the Grace Blackwell Ultra in performance, but want latency, throughput, and power. And it's actually comparable. to Verirubin GPUs in running OpenAI workloads. 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. 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 three generations of MTIA accelerators to Meta. Across these three generations, we have line of sight to deploy 3 gigawatts through 28. 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 internet 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're now the leader in leading-edge optical DSPs and are rapidly expanding our …
Broadcom claims to be shipping its next-gen TPU before a competing MediaTek product that was initiated earlier, highlighting its execution speed in custom silicon design.
… volume to both Entropic and Google. 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 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. In Q3, We also ship Jalapeno, 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 three and a half times year on year and represented 73% of AI revenue during the quarter. Our AI networking revenue was up over two and a half times year on year. And this strong momentum continues into Q4. During the quarter, we expect to accelerate shipments of Ironwood to Entropic. We also expect to ramp up high volume shipments of the TPU version 8i to Google. Shipments of Jalapeno for OpenAI will continue. And for …
Broadcom's custom Ethernet switch 'Tomahawk Ultra' is enabling scale-up networking within GPU/XPU clusters for the first time, potentially disrupting proprietary GPU interconnect solutions. — If Ethernet-based scale-up proves viable, it could challenge proprietary, high-bandwidth scale-up networks, opening the door for Broadcom's networking solutions in a key segment where Nvidia's NVLink has been dominant.
… 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 KBAR. 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. In addition, we are in development with OpenAI on their third generation XPU. As OpenAI announced last week, Jalapeno outperforms the Grace Blackwell Ultra in performance, but want latency, throughput, and power. And it's actually comparable. to Verirubin GPUs in running OpenAI workloads. 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. 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 three generations of MTIA accelerators to Meta. Across these three generations, we have line of sight to deploy 3 gigawatts through 28. Our content in AI, as you know, goes …
Broadcom's custom Ethernet switch 'Tomahawk Ultra' is enabling scale-up networking within GPU/XPU clusters for the first time, potentially disrupting proprietary GPU interconnect solutions. — If Ethernet-based scale-up proves viable, it could challenge proprietary, high-bandwidth scale-up networks, opening the door for Broadcom's networking solutions in a key segment where Nvidia's NVLink has been dominant.
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've heard that 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 for Tomahawk 6. But next year, I'm not sure if you can give us an update on that. And it's also good to see the team picked out its next-gen Tomahawk 7 with 200 gigs. Out of your six Frontier model-based XPU customers…
Just to amplify what Charlie is saying with Tomahawk Ultra is that we are enabling now scaling up within cluster of GPU, XPU in the rack on basically Ethernet for the first time because it won't, Tomahawk 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.
Broadcom's custom accelerators are designed to have roughly half the cost of merchant GPUs (like Nvidia's) while delivering comparable or better performance for specific workloads.
… second largest XPU customer. In addition, we are in development with OpenAI on their third generation XPU. As OpenAI announced last week, Jalapeno outperforms the Grace Blackwell Ultra in performance, but want latency, throughput, and power. And it's actually comparable. to Verirubin GPUs in running OpenAI workloads. 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. 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 three generations of MTIA accelerators to Meta. Across these three generations, we have line of sight to deploy 3 gigawatts through 28. 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 internet 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, …
🍪 Cookies
We use cookies to operate this site and for analytics. Declining won't affect your access to content.