OpenAI and Broadcom unveil LLM-optimized inference chip
OpenAI
OpenAI and Broadcom have introduced Jalapeño, a new custom AI accelerator chip designed to optimize Large Language Model inference performance.Summary
OpenAI and Broadcom have unveiled Jalapeño, a specialized AI processor designed specifically for Large Language Model (LLM) inference. Developed in just nine months, the chip features a unique architecture that balances compute, memory, and networking to improve performance-per-watt efficiency over existing hardware. As the first entry in a multi-generation partnership, Jalapeño is set for deployment in gigawatt-scale data centers starting in 2026, aiming to make advanced AI more affordable, reliable, and accessible.
(Source：OpenAI)