OpenAI and Broadcom on Wednesday revealed Jalapeño, their first co-designed custom chip, marking OpenAI’s debut entry into AI silicon. The chip is an application-specific integrated circuit (ASIC) built for inference, the process of serving AI models to users in ChatGPT and other applications, according to CNBC.

Nine Months From Design to Sample

OpenAI President Greg Brockman told CNBC’s David Faber that the chip was designed from end to end in nine months, with OpenAI’s own AI models accelerating the process. “The degree to which our models have been able to accelerate it was very surprising to us,” Brockman said on Squawk on the Street. OpenAI also designed large parts of the computer system where Jalapeño will be deployed.

The companies are calling the chip an “Intelligence Processor” and describe it as the first “AI accelerator” in a platform they are building “to make advanced AI faster, more reliable, and more accessible to more people,” per the press release cited by CNBC. A physical sample was delivered to OpenAI on Wednesday.

Deployment Timeline

Initial deployment of Jalapeño chips is targeted by the end of 2026, with a ramp from there. Broadcom CEO Hock Tan told CNBC the schedule starts with “small prototype development” in late 2026. “We will start seeing it really ramp up in ‘27 and really going full tilt in first half ‘28,” Tan said.

The partnership, which went public in October 2025 after 18 months of joint work, aims to build enough compute capacity to require 10 gigawatts of power. ASICs like Jalapeño are less flexible than Nvidia’s GPUs but less expensive and designed for specific AI tasks, CNBC reported.

Demand Outstripping Supply

Brockman said OpenAI “cannot get compute fast enough.” Tan backed that assessment, describing demand from Broadcom’s six customers as “simply insatiable.” “It’s just much more than we can address,” Tan said. “And this is not just ‘26, not ‘27, we’re seeing that same and even elevated demand in ‘28 as well.”

OpenAI has been one of the largest buyers of Nvidia GPUs since launching the generative AI boom in 2022. To diversify its supply, the company has signed deals with Amazon Web Services for Trainium chips, Advanced Micro Devices, and AI chipmaker Cerebras, which held its IPO in May, CNBC reported.

The Vertical Integration Bet

Jalapeño represents a step in OpenAI’s plan to “build the full stack behind its models and products.” “By designing more of the stack ourselves, we can serve more intelligence with greater efficiency and keep pushing advanced AI toward broader access,” Brockman said in the press release.

For companies running autonomous agents at scale, the inference layer is the bottleneck. Training a model happens once; serving it to millions of concurrent agent sessions happens continuously. OpenAI’s move to control its own inference silicon is a bet that whoever owns the chip layer owns the cost structure of the entire agent deployment pipeline.