OpenAI and Broadcom Unveil First Joint AI Chip 'Jalapeño'

OpenAI and Broadcom today announced the completion of the design for their first co-developed AI accelerator, a chip called 'Jalapeño.' The partnership was first disclosed in October 2025, and now the companies have provided details on the processor.

In a blog post, OpenAI described 'Jalapeño' as its 'first Intelligence Processor: an accelerator architected around OpenAI's vision for the future of LLM inference.' In other words, the chip is designed specifically to run OpenAI's large language models.

The company claims that so far, 'Jalapeño' offers 'performance per watt substantially better than current state-of-the-art' in chip technology. However, the post noted that final performance testing is still underway, and a more detailed technical report on the chip's specifications will be released in the coming months.

The next phase involves deploying 'Jalapeño' in data centers, with initial deployment slated to begin in late 2026. This marks OpenAI's first foray into manufacturing the chips needed for training and operating its ChatGPT models. By controlling more components in the LLM processing pipeline, the company says it can improve efficiency and increase the usability of ChatGPT tools.

OpenAI also highlighted that 'Jalapeño' was co-developed from initial design to manufacturing tape-out in just nine months. The speed, according to the company, reflects deep software-hardware co-development with OpenAI's engineering teams, Broadcom's silicon implementation expertise, and the use of OpenAI models to accelerate parts of the design and optimization process.

'Jalapeño' is the first AI accelerator from the Broadcom-OpenAI partnership, and the two have described their collaboration as a 'multi-generation compute platform' aimed at making advanced AI faster, more reliable, and more accessible to a wider audience.