OpenAI and Broadcom announce chip designed for LLM inference at scale (techsentiments.com)