Jakub Porzycki Nurphoto Getty Images
OpenAI and Broadcom On Wednesday they unveiled their custom chip, called Jalapeño, marking the ChatGPT maker’s first foray into artificial intelligence silicon.
The chips will be manufactured by Broadcom and used by OpenAI for inference, a computerized process for rendering its AI models to users in ChatGPT and other applications.
OpenAI President Greg Brockman told CNBC’s David Faber on Wednesday that the chips were made end-to-end in nine months with the help of the company’s AI models.
“The rate at which our models were able to accelerate really surprised us,” said Brockman.
Broadcom has been one of the biggest beneficiaries of the manufacturing AI boom by helping hyperscalers and frontier labs build their own AI chips. The chipmaker’s shares are up 10% year-to-date in 2026 and nearly sevenfold by the end of 2022.
The manufacturer’s shares rose on Wednesday following the announcement.
Jalapeño is a big step in OpenAI’s plan to “build the full stack behind its models and products,” according to a press release.
“By building a wider stack, we can use more intelligence more effectively and continue to push advanced AI into wider reach,” Brockman said in a statement announcing Jalapeño.
Since OpenAI started to generate AI boom in 2022, the company has been one of the biggest consumers. Nvidia’s expensive graphics processing units, an essential piece of infrastructure for building AI models and running large workloads. But OpenAI is facing such an explosion in demand that it needs more advanced silicon sources.
Earlier this year, OpenAI struck a deal with it Amazon Web Services include the use of the company’s Trainium AI chips. OpenAI has also signed deals with rival Nvidia Advanced Micro Devices and AI chipmaker Cerebras, holding its initial public offering in May.
In October, after 18 months of working together, OpenAI and Broadcom went public with plans to develop and ship chips designed by OpenAI starting later this year, eventually aiming to build enough to require 10 gigawatts of power.
The chip with Broadcom is an ASIC, which industry experts say is less flexible than Nvidia’s GPU, but also less expensive and can be built for specific AI tasks. OpenAI said it designed the chip in nine months, and built the major components of the computer system in which it will be used.
The companies call the chip an “Intelligence Processor” and describe it as the first “AI accelerator” in the space they are building to “make advanced AI faster, more reliable, and accessible to more people.”
A physical sample of the new chip will be delivered to OpenAI on Wednesday. The companies said they are aiming for the first shipments of Jalapeño chips by the end of 2026, “expandable over the coming years.”
Broadcom CEO Hock Tan told CNBC on Wednesday that there will be “small prototype development” in late 2026 and it will go from there.
“We will start to see it grow well in ’27 and be full tilt in the first half of ’28,” Tan said.
WATCH: Nvidia GPUs in Google TPUs: Breaking down all AI chips
