OpenAI’s First Custom AI Chip Bets on 50% Cheaper Inference and Less Nvidia (opens in new tab)

OpenAI and Broadcom unveiled Jalapeño on June 24, the company’s first custom AI inference chip, built on TSMC’s 3nm process and targeting 50% lower cost per token than Nvidia GPUs. The chip reached tape-out in nine months and is set for initial deployment by end of 2026.