India, Aug. 18 -- NVIDIA is accelerating the development of its next-generation "Feynman" GPU architecture by utilizing TSMC's cutting-edge A16 (1.6nm-class) process node.

This move marks a major shift in NVIDIA's strategy, dethroning Apple as the primary launch customer for TSMC's most advanced lithography. NVIDIA is actively using its own Blackwell GPUs to run AI-driven simulations to bootstrap, verify, and speed up the design cycle for the Feynman silicon.

Technical pillars of Feynman architecture

TSMC A16 Process: A 1.6nm-class node utilizing NanoFlex Gate-All-Around (GAA) nanosheet transistors.

Backside Power Delivery: Employs TSMC's "Super Power Rail" technology. This separates power and signal routing to drastically cut down re...