Show notes
OpenAI unveiled Jalapeño, its first custom inference chip — co-developed with Broadcom, taped out in nine months with AI assistance, and named after a mid-tier pepper. AMD doubled its server CPU addressable market to $120 billion, but Nvidia sold $41 billion in inference chips in a single quarter — more than AMD and Broadcom combined. And Micron collected $22 billion in upfront deposits from customers desperate to lock in memory supply, with $100 billion in take-or-pay contracts designed to break a thirty-year boom-bust cycle.
Website: youtu.be

