ai
OpenAI and Broadcom unveil Jalapeño chip for ChatGPT inference, with 2026 deployment planned
OpenAI and Broadcom introduced the Jalapeño chip Wednesday as the company's first custom AI silicon. The chips will be manufactured by Broadcom and deployed for inference workloads in ChatGPT and other services starting late 2026.
forbes.comAudio version
Tap play to generate a narrated version.
Developing·Limited corroboration so far. This page will refresh as more sources emerge.
OpenAI and Broadcom unveiled a custom chip called Jalapeño on Wednesday. The chip marks OpenAI's first entry into artificial intelligence silicon and will be manufactured by Broadcom for inference tasks that serve models to users in ChatGPT and other applications. Greg Brockman, OpenAI's president, said the move allows the company to design more of its own stack.
"By designing more of the stack ourselves, we can serve more intelligence with greater efficiency and keep pushing advanced AI toward broader access," he stated. Broadcom shares rose about 2 percent after the announcement, according to CNBC. The companies aim to build enough chips to require 10 gigawatts of power and plan initial deployment by the end of 2026.
OpenAI designed the Jalapeño chip in nine months and also developed large parts of the surrounding computer system. A physical sample will be delivered to OpenAI on Wednesday. The partnership follows an October announcement of plans to develop and deploy racks of OpenAI-designed chips.
OpenAI has previously signed deals to use AI chips from Amazon Web Services, Advanced Micro Devices and Cerebras, which held its initial public offering in May. The company has been one of the largest buyers of Nvidia GPUs since 2022.
