OpenAI and broadcom On Wednesday, the ChatGPT maker announced its debut custom chip called Jalapeño, marking its first foray into artificial intelligence silicon.
The chip will be manufactured by Broadcom and used by OpenAI for inference, the compute-intensive process that provides AI models to users in ChatGPT and other applications.
OpenAI President Greg Brockman told CNBC’s David Faber on Wednesday that the chip was designed end-to-end in nine months with the help of the company’s AI models.
“It was very surprising to us how much our model was able to accelerate that,” Brockman said.
Broadcom has been one of the biggest beneficiaries of the generative AI boom by helping hyperscalers and Frontier Labs create their own custom chips for AI. The company’s stock is up 10% so far in 2026 and has increased nearly seven times since the end of 2022.
The company’s stock price rose on Wednesday following the announcement.
Brockman told CNBC that OpenAI “cannot deliver computing fast enough,” a sentiment echoed by Broadcom CEO Hock Tan, who said the demand for computing from the company’s six customers is “absolutely insatiable.”
“It’s more than we can handle,” he said. “This isn’t just 26 or 27, we’re seeing similar and even higher demand in 2028.”
Jalapeño is an important step in OpenAI’s plan to “build a full stack behind our models and products,” according to a press release.
“By designing more of our stack in-house, we can deliver more intelligence more efficiently and continue to push advanced AI toward broader access,” Brockman said in a release announcing Jalapeño.
Since OpenAI sparked the generative AI boom in 2022, the company has been one of the largest buyers. Nvidia’s Expensive graphics processing units are critical infrastructure for building AI models and running large-scale workloads. But OpenAI is experiencing exploding demand, requiring other sources of advanced silicon.
Earlier this year, OpenAI signed agreements with: Amazon Web services that include the use of the company’s Trainium AI chip. OpenAI also signs deal with Nvidia rival advanced micro device with AI chip manufacturer Cerebrum, The company held an initial public offering in May.
After 18 months of collaboration, OpenAI and Broadcom in October announced plans to develop and deploy racks of OpenAI-designed chips starting later this year, with the goal of eventually building enough racks to power 10 gigawatts of power.
Broadcom’s chips are ASICs that industry experts say are less flexible than Nvidia’s GPUs, but are also cheaper and can be designed for specific AI tasks. OpenAI said it designed the chip in nine months and also created most of the computer systems in which it will be used.
The companies call 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.”
Physical samples of the new chip are expected to be delivered to OpenAI on Wednesday. The companies said they aim to have initial introductions of jalapeño chips by the end of 2026 and “will expand over the next few years.”
Tan told CNBC on Wednesday that there will be “small-scale prototype development” in late 2026, followed by scaling up from there.
“It will really ramp up in 2027, and it will pick up steam in the first half of 2028,” Tan said.
