Nvidia's Feynman AI Platform Accelerates Development, Pressuring TSMC
Nvidia is pushing forward with the development of its next-generation Feynman AI platform, accelerating research and supply chain resources in the process.
The company's Vera Rubin platform has entered mass production and ramp-up, but Nvidia is shifting focus to Feynman, which promises significant architectural evolution over Rubin.
Feynman will feature 3D chiplets, TSMC’s System on Integrated Chips (SoIC) technology, high-bandwidth memory (HBM), and co-packaged optics (CPO).
TSMC is racing to expand its advanced packaging capacity in response to the accelerated Feynman roadmap, with plans to increase monthly SoIC capacity to around 50,000 wafers by year-end 2027.