Hock Tan guides Broadcom's AI revenue to $21.7 billion after a $16.7 billion quarter

---

---

Broadcom’s custom AI silicon brought in $16.7 billion in its fiscal third quarter.

CEO Hock Tan told investors the run is not slowing.

Semiconductor solutions climbed 127% to $20.8 billion

Broadcom’s revenue for the quarter ended August 2 was $29.6 billion, up 86% from a year earlier, the company said in an earnings release.

The semiconductor solutions segment grew 127% to $20.8 billion. Infrastructure software grew 29% to $8.75 billion.

AI semiconductor sales were $16.7 billion, a 221% increase from the same period in 2025 and up 54% from the prior quarter, Tan said. He called demand for the company’s custom accelerators and networking gear “very strong.”

Broadcom generated $13.7 billion in free cash flow, or about 46% of revenue. Non-GAAP operating margin was 66%.

Tan said in the earnings release he expects AI semiconductor revenue to climb to $21.7 billion in the fourth quarter. That would be a 236% rise from the same period a year ago.

Hock Tan’s statement in Broadcom’s third quarter fiscal 2026 earnings release, published September 2, 2026.

CFO Amie Thuener said the company sees fourth-quarter revenue guidance of about $34.8 billion, up 93% from a year ago. The company foresees that its non-GAAP operating margin will remain at 66%.

Google’s custom silicon deal runs through 2031

The board declared a quarterly dividend of $0.65 per share. Broadcom’s fiscal fourth quarter ends November 1, and the company cautioned that actual results may vary materially from the outlook.

That demand is from a short list of named clients who are building their own chips to reduce their dependence on Nvidia GPUs.

In April, Broadcom signed a long-term agreement with Alphabet’s Google to supply custom AI chips through 2031, prolonging a TPU partnership that has been in place for about 10 years, Cryptopolitan reported.

The same stretch landed a deal to supply Anthropic with 3.5 gigawatts of Google-powered computing capacity starting in 2027. That commitment could approach 5 gigawatts over time.

In June, ChatGPT’s maker unveiled “Jalapeño,” an inference chip it designed and built with Broadcom. It’s the first part of a larger hardware platform, both companies said.

OpenAI president Greg Brockman framed the move as a way to “serve more intelligence with greater efficiency.” The two companies set a target of eventually powering about 10 gigawatts of capacity.

Broadcom’s play is centered on the ASIC, an application-specific chip that’s less flexible than a general-purpose GPU but cheaper and tuned for specific AI jobs. That tradeoff has found a market in inference, the work of running trained models for end users.

Nvidia’s lead in inference is not quite as absolute as it is in training. None of Broadcom’s customers is ditching Nvidia.

OpenAI’s Sam Altman has publicly said Nvidia makes “the best AI chips in the world.” Anthropic runs Claude on AWS Trainium, Google TPUs, and Nvidia GPUs to distribute its risk.

The smartest crypto minds already read our newsletter. Want in? Join them.