NVIDIA to Unveil Next-Generation Feynman AI Chip at GTC 2026 With First Use of 1-Nanometer-Class Process

Feynman is NVIDIA's planned successor to its Rubin-generation AI computing platform. It is expected to integrate the Rosa CPU, HBM and advanced packaging, with copper interconnects and CPO optical communications supporting a scale-up architecture. If NVIDIA adopts TSMC's A16 process, the platform could generate manufacturing, packaging and thermal-management opportunities for Taiwanese suppliers including TSMC and ASE Technology.

NVIDIA CEO Jensen Huang previewed a demonstration of Feynman during the GTC 2026 keynote scheduled for March 16, 2026. Attention is focused on its anticipated first use of the 1-nanometer-class A16 process and new interconnect designs. Ahead of the presentation, NVIDIA had not disclosed a mass-production timetable, pricing, order values or a list of Taiwanese suppliers.