OpenAI and Broadcom Unveil Jalapeño, First Custom AI Inference Chip Designed in Nine Months
About This Episode
OpenAI and Broadcom on June 24 unveiled Jalapeño, OpenAI's first custom-built AI inference chip, designed end-to-end in nine months with assistance from OpenAI's own AI models. The chip cuts inference costs roughly 50% versus current GPUs, is fabricated on TSMC's 3nm process, and will deploy at Microsoft data centers by year end.
Our Take
OpenAI's Jalapeño chip signals the AI industry's shift from renting Nvidia's GPUs to designing its own silicon, reshaping who controls the cost and means of intelligence production
Continue Reading on Unscarcity
AI Compute Clusters 2026: The $800B GPU Arms Race
Direct match — maps the GPU arms race among tech giants building custom silicon, the exact competitive dynamic OpenAI's Jalapeño chip enters
Altman's UBC: Why Give Computing Power Instead of Cash?
Complementary — if inference costs drop 50% per chip generation, compute-as-infrastructure becomes viable, supporting Altman's vision of universal compute access