OpenAI and Broadcom announce custom-designed AI chip for inference, marking major competitive move away from NVIDIA.

OpenAI and Broadcom have announced the development of Jalapeño, OpenAI's first custom AI chip. Designed by OpenAI and manufactured by Broadcom, the chip reportedly delivers better performance per watt than current state-of-the-art options in early testing—a direct challenge to Nvidia's market dominance. The chip is built for inferencing with OpenAI's own AI models and those across the industry.

"Jalapeño is part of our long-term full-stack infrastructure strategy to make compute more abundant, resulting in AI which is faster, more reliable, more affordable for people and businesses, and can be used to solve more important problems," said OpenAI president Greg Brockman. "By designing more of the stack ourselves, we can serve more intelligence with greater efficiency and keep pushing advanced AI toward broader access."

The custom chip was developed in just nine months and represents the first in a multigeneration computing platform expected to roll out later this year and beyond. This initiative reflects a broader industry trend as AI companies and cloud hyperscalers seek alternatives to Nvidia's high-powered processors.

Despite being one of Nvidia's largest customers, OpenAI faces intense competition from the broader AI industry for access to Nvidia's chips. Developing its own processors accelerates OpenAI's ability to secure the computing power it needs while also enabling better optimization of its systems for its specific requirements.

OpenAI joins a growing list of tech giants developing custom AI chips. Amazon, Google, and Microsoft use or are developing their own processors for AI workloads. Meta similarly designs and deploys custom chips for AI and other applications. Both Amazon and Google have begun offering their chips to third-party customers, and Meta has suggested potential expansion into cloud computing services—moves that intensify competition with Nvidia.

Beyond Nvidia, chipmakers like AMD are competing for market share with their own AI data center offerings, while companies such as Qualcomm and Cerebras are making pushes into the AI chip space.