Broadcom Inc Earnings - Q3 2026 Analysis & Highlights
Broadcom delivered exceptional Q3 2026 results driven by explosive AI semiconductor growth, with the company guiding for continued acceleration through 2028 as it deepens partnerships with six frontier AI labs and expands its custom accelerator and networking portfolio.
Key Financial Results
Consolidated revenue reached a record $29.6 billion in Q3 2026, up 86% year-over-year.
AI semiconductor revenue grew 221% year-over-year and 54% sequentially to $16.7 billion, representing 56% of total revenue.
Operating income grew 92% year-over-year to a record $20.1 billion, with operating margin at 67.9%, up 240 basis points year-over-year.
Gross margin was 75% of revenue in Q3, down 210 basis points sequentially due to AI semiconductor revenue mix.
Non-GAAP EPS of $3.32 was up 96% year-over-year.
Free cash flow reached a record $13.7 billion, representing 46% of revenue.
XPU shipments were up over 3.5 times year-over-year and represented 73% of AI revenue during the quarter.
AI networking revenue was up over 2.5 times year-over-year.
Business Segment Results
Semiconductor Solutions segment revenue was a record $20.8 billion, up 127% year-over-year, representing 70% of total revenue.
Semiconductor Solutions segment gross margin was approximately 76%, with operating margin of 61%, up 440 basis points year-over-year.
Infrastructure Software revenue was $8.8 billion, up 29% year-over-year, representing 30% of total revenue.
Infrastructure Software gross margin was 94% in the quarter with operating margin of approximately 84%, up 650 basis points year-over-year.
Non-AI semiconductor revenue was $4.2 billion, up 5% year-over-year and flat sequentially, with broadband and server storage up, partially offset by wireless decline.
Broadcom delivered Ironwood TPU v7 in high volume to both Anthropic and Google, and began production shipments of TPU version 8i for Google.
Broadcom shipped Jalapeño, OpenAI's first-generation custom accelerator, which outperforms Grace Blackwell GPU for inference workloads.
Capital Allocation
Broadcom paid stockholders $3.1 billion in cash dividends in Q3 based on a quarterly common stock dividend of $0.65 per share.
The company paid down $5.6 billion of long-term debt in Q3 and an additional $1.5 billion of senior notes upon maturity subsequent to quarter end.
Capital expenditures were $532 million in Q3.
Broadcom 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.
The company closed the first $35 billion tranche in June for Anthropic's 1-gigawatt deployment, which is already underway.
The weighted average coupon rate and years to maturity of gross principal fixed rate debt of $59.6 billion is 4% and 7.4 years, respectively.
Industry Trends and Dynamics
Demand for AI semiconductors is exceptionally strong, with Broadcom's six XPU customers accelerating adoption of custom accelerators.
The vast majority of compute demand for AI workloads originates from a concentrated group of frontier model developers, with their need for compute infrastructure expected to inflate in 2027 and 2028.
Demand for AI networking components, including Ethernet switching and optical interconnects, is growing at similar rates to XPU demand.
Demand for lasers, including EML lasers and CW lasers, is far surpassing supply in the industry.
Broadcom is seeing customers deploy Tomahawk 6 in both 100 gig and 200 gig SerDes configurations across all AI hyperscalers building XPUs.
Competitive Landscape
Broadcom possesses the strongest IP portfolio in semiconductor design, including industry-leading SerDes, chip-to-chip interconnect, leading-edge HBM and SRAM integration, and differentiated advanced packaging.
Broadcom has consistently delivered the fastest time to market for TPUs from product definition to production without the need for resubmits.
When co-developed chips are optimized for particular LLM workloads, they outperform any GPU, as demonstrated by Jalapeño's performance compared to Grace Blackwell Ultra.
Broadcom's custom accelerators can run other frontier models and do so at half the cost of a GPU.
Broadcom is the leader in AI networking, including Ethernet switching for scale up and scale out, PCI Express switching for scaling in, and optical DSPs, EMLs, VCSELs and CW lasers for optical interconnects.
Macroeconomic Environment
The document does not contain specific discussion of macroeconomic factors such as inflation, tariffs, trade, or recession.
Growth Opportunities and Strategies
Broadcom is deepening partnerships with six frontier AI labs: Google, Anthropic, OpenAI, Meta, and two others, each pursuing custom accelerators optimized for their specific workloads.
Google partnership is strengthened by a long-term agreement to develop and supply future generations of TPUs and AI networking, with planning to deliver multi tens of billions of dollars of TPUs annually over the next several years.
Anthropic is expected to deploy 1 gigawatt of Ironwood in 2026, 5 gigawatts of TPU version 8i in 2027, and an incremental 10 gigawatts in 2028, positioning Anthropic as Broadcom's largest XPU customer in 2027 and 2028.
OpenAI's Jalapeño is on track for planned deployment of 1.3 gigawatts in 2027, with line of sight for over 5 gigawatts of Jalapeño and successor generation XPU in 2028, making OpenAI Broadcom's second largest XPU customer.
Broadcom is in development with OpenAI on their third-generation XPU.
Meta partnership will deliver three generations of MTIA accelerators between now and the end of 2027, with line of sight to deploy 3 gigawatts through 2028.
Broadcom announced VMware Private AI Cloud, giving enterprises a secure, cost-effective platform to build and run AI alongside existing applications.
VMware Cloud is 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 opening a new opportunity for Broadcom's infrastructure software business.
Broadcom is building substrate capacity at scale in its Singapore factory to address supply bottlenecks.
Broadcom is more than tripling its EML, CW, and VCSEL factories in indium phosphide in both the US and Singapore.
Financial Guidance and Outlook
For Q4 2026, Broadcom expects consolidated revenue of $34.8 billion, up 93% year-over-year.
Q4 2026 AI semiconductor revenue is expected to be $21.7 billion, up over 236% year-over-year.
Q4 2026 semiconductor revenue is forecast at approximately $26.1 billion, up 136% year-over-year.
Q4 2026 infrastructure software revenue is expected at approximately $8.7 billion, up 24-25% year-over-year.
Q4 2026 consolidated gross margin is expected to be approximately 73%, down from 78% a year ago due to increasing XPU mix.
Q4 2026 operating margin is expected to be approximately 66%, flat from a year ago.
Non-GAAP tax rate for Q4 and fiscal year 2026 is expected to be approximately 16%.
Q4 2026 non-GAAP diluted share count is expected to be approximately 4.94 billion shares, excluding potential share repurchases.
Q4 2026 capital expenditures are expected to be $1.4 billion as the company invests in semiconductor capacity.
Fiscal 2026 AI revenue is expected to be $58 billion for the year, up 186% year-over-year.
Fiscal 2027 AI revenue is expected to double to approximately $115 billion, with demand exceeding this outlook.
Fiscal 2028 AI semiconductor revenue is expected to double again to $230 billion.
Broadcom expects to exceed $30 in earnings per share in fiscal 2028.
Non-AI semiconductor revenue is forecast at approximately $4.3 billion in Q4, up 5% sequentially.
Infrastructure software revenue is forecast to stabilize at approximately $8.7 billion in Q4.
Revenue per gigawatt of XPU deployment is expected to remain in the range of $20 billion to $30 billion, as increasing chip power offsets rising ASPs.
Customer Concentration and Strategic Partnerships
Broadcom's business is concentrated with six frontier model-developing customers, which the company views as manageable given their strategic importance.
Anthropic and OpenAI are expected to become Broadcom's largest and second-largest XPU customers respectively, with Broadcom creating financing vehicles through the XPV platform to help bridge their infrastructure investment gaps.
Anthropic is on track for an IPO, which will change its investment credit profile, while OpenAI's financing structure remains less clear.
Broadcom views its investment in enabling Anthropic and OpenAI as strategically sound, as each gigawatt of compute deployed could achieve approximately $30 billion of annual revenue per gigawatt.
Long-term, Broadcom expects Anthropic and OpenAI to evolve into hyperscalers running their own data centers and offering generative AI APIs as SaaS models to the world.
Supply Chain and Operational Constraints
Land, power, and shell availability are significant constraints on AI data center deployment timing and must be jointly managed with customers.
Substrate availability is a specific supply bottleneck being addressed through Broadcom's Singapore factory expansion.
HBM memory and system memory availability are critical constraints that customers must secure independently.
Supply chain constraints span multiple dimensions including leading-edge silicon production, substrates, memory, and data center infrastructure.
Broadcom has secured supply to meet its fiscal 2027 and 2028 AI revenue guidance.
