OpenAI has unveiled Jalapeño, its first custom AI chip, developed with Broadcom and designed to power ChatGPT and the company's next generation of large language models. According to TechCrunch, it is an "inference" chip — the kind used to run AI models rather than train them — and OpenAI built it specifically to reduce its reliance on Nvidia's GPUs.
Nvidia has dominated the AI chip market for years, and several reports frame Jalapeño as a sign that the era of total dependence may be ending. The Indian Express reports that the chip is part of OpenAI's push to build the computing infrastructure it needs for future models.
The move is notable because of the companies' financial ties. According to AOL/Yahoo Finance, Nvidia had once pledged up to $100 billion to OpenAI, and the new chip was designed "from scratch" after a deal with Nvidia fizzled — a development that has prompted some to ask whether Nvidia investors should be worried.
OpenAI is not acting alone. TechCrunch notes the company joins Google, Apple, and SpaceX in a growing list of firms building their own chips, turning up competitive heat on Nvidia. Outlets including The Tech Buzz and Yahoo Finance describe OpenAI as pulling ahead on custom silicon, with some framing Jalapeño as a signal of the end of Nvidia's AI "monopoly."
Why it matters: if the biggest AI companies can design their own chips, it could loosen Nvidia's grip on one of the most valuable markets in technology and reshape who controls the hardware behind modern AI.