OpenAI's Jalapeño chip targets Nvidia
openai revealed 'jalapeño', its first custom inference chip built with broadcom. the move signals a shift to reduce reliance on nvidia gpus for running models.
openai has debuted jalapeño, its first custom inference chip co-developed with broadcom to optimize the performance-per-watt of its model inference. early tests indicate significant efficiency gains over state-of-the-art alternatives, specifically for running real-time coding models. this vertical integration reduces openai's dependence on nvidia gpus, aligning with strategies employed by google and amazon.