Summary
Dwarkesh Patel spent two days touring Intel’s fabs and then interviewed its leadership to understand what it takes to manufacture the world’s most advanced chips and how hardware will shape AI’s future. The conversation opens with the current shock to the semiconductor industry from AI demand, which one executive describes as 40-50% above maximum capacity.
In the first interview, Chris Auth, who leads development of Intel’s manufacturing processes, explains how a node like 14A moves from development to production. He covers the roughly 2,000 steps in the factory, how defects are found through inline metrology and end-of-line testing, how engineers use cross-sections to trace them to a stage of the flow, and why yield improves month over month. The discussion also revisits Intel’s delayed 10 nanometer node, which slipped from a 2016 target to 2018, and what the company learned about limiting innovation per node.
The video also examines concrete production problems such as epitaxy nodules, which can short transistors, and shows how teams work in parallel to isolate and fix their sources. It is a detailed look at the physical bottlenecks, including tools from suppliers like ASML, that constrain the supply of AI compute.
📺 Source: Dwarkesh Patel · Published October 22, 2026
🏷️ Format: Interview
