OpenAI and Broadcom Unveil Jalapeño Chip for Large-Scale LLM Inference

OpenAI and Broadcom have introduced a new chip named Jalapeño, specifically built for large language model inference in data centers. The companies will deploy the chip at scale and describe it as the first generation in an ongoing project to refine the technology over time. This collaboration aims to optimize AI inference performance for models like ChatGPT.