OpenAI and Broadcom Launch Jalapeño Chip for LLM Inference at Scale

OpenAI and Broadcom have announced Jalapeño, a new chip designed specifically for large language model inference in data centers. The ASIC was developed from scratch over nine months based on insights from OpenAI's researchers and informed by the company's product roadmap.

Broadcom says Jalapeño is more specialized for current LLM needs than existing inference hardware. OpenAI claims early testing shows the chip delivers substantially better performance per watt than current solutions, though detailed technical findings will be published in the coming months.