Nvidia announces Feynman GPU architecture for 2028 release

Nvidia has announced its future GPU architecture, named Feynman, which is set to launch in 2028. The architecture is named after physicist Richard Feynman, who won a Nobel Prize and was known for his contributions to science and communication. Nvidia’s CEO, Jensen Huang, mentioned Feynman during his recent keynote at the GTC event. However, he did not provide details about this new architecture. Feynman will succeed the upcoming Rubin architecture, which is expected to focus on AI and is scheduled for release in 2026. Currently, Nvidia offers the Blackwell architecture, used in both AI chips and the new RTX 50-series gaming GPUs. Rubin and Feynman will follow Blackwell, but Rubin gaming GPUs likely won't be available until early 2027. It remains uncertain if Feynman will also be used for gaming, though it is likely, given Nvidia's recent trends. Nvidia's Blackwell GPUs are currently made using TSMC’s N4 process. Rubin is anticipated to utilize TSMC's advanced N3 node. By 2028, TSMC's N2 node should be fully operational. Nvidia is also considering Intel's foundry services for future manufacturing, including their 14A and 18A processes. While expectations are high for Rubin, Feynman's future remains uncertain. It is clear that Nvidia plans to promote Feynman in the coming years.