Nvidia Feynman GPUs to use Intel Foundry for some components

Unlike 18A, which was largely written off by external consumers, Intel Foundry’s 14A node is shaping up to be promising, with insiders calling it ‘the real deal’. Some of its prospective clients include Apple and puzzlingly, AMD. Even Nvidia is listed at one, but its scope will be a lot more limited. A new DigiTimes report tells us more about Nvidia’s potential collaboration with Intel Foundry.

Nvidia’s 2028-bound Feynman GPUs will supposedly use Intel 18A or 14A for their I/O dies. The primary GPU die, however, will continue to rely on TSMC’s cutting-edge node, likely a custom N3 or N2 variant. In addition to the IOD, Nvidia also plans to incorporate Intel’s EMIB (Embedded Multi-die Interconnect Bridge) packaging into some chips. However, the bulk of Feynman silicon will continue to rely on TSMC’s CoWoS (Chip-on-Wafer-on-Substrate), with an estimated 75-25 split between Intel and TSMC.

While earlier rumours stated some entry-level Nvidia products, such as an N1X successor, could be fabricated on Intel 14A, the DigiTimes report makes no such observation. Instead, it reiterates earlier leaks, which insisted future Apple M-series silicon could be based on 14A. Their Pro/Max/Ultra variants, on the other hand, will stick with TSMC, as will the A-series silicon for smartphones.