OpenAI and Broadcom launch first custom AI chip named Jalapeño
OpenAI has unveiled its first hardware component, a dedicated inference chip developed in partnership with Broadcom. The technology is designed to power large language models as part of an effort to build the company's full infrastructure stack.
Amalgamated from The Verge (opens in new tab), TechCrunch (opens in new tab), Android Authority (opens in new tab), CNBC (opens in new tab)
OpenAI has unveiled its first custom intelligence processor, developed in collaboration with Broadcom. The chip, named Jalapeño, is specifically designed for AI inference and is intended to support both current and future large language models.
The announcement follows a partnership between OpenAI and Broadcom that was first disclosed approximately eight months ago. According to reports from The Verge and TechCrunch, the hardware functions as an Application-Specific Integrated Circuit (ASIC). This means the chip is engineered for specific processing tasks rather than general-purpose computing, which allows for more efficient handling of AI workloads.
CNBC and Android Authority report that the development of Jalapeño is part of a broader strategy to build the full stack of infrastructure supporting OpenAI’s models. By producing custom hardware, the company aims to control more layers of the technology powering its systems. This move allows for specialized optimization of the inference process as the company scales its AI capabilities.