Rumours suggest Nvidia's next-next gen Fenyman AI GPUs will use Intel Foundry tech for some of the components

Nvidia and Intel shocked the tech world last year with a partnership announcement, in which the two giants said they would collaborate on new x86 SOCs with integrated RTX GPU chiplets. The friendship may well be going deeper than that, though, as rumours suggest Nvidia's next-next gen AI GPUs will use the odd Intel Foundry-produced component.

Industry sources speaking to Digitimes have indicated that the I/O die on Nvidia Fenyman GPUs will "partially utilise" either Intel's 18A or 14A process (via Notebookcheck). The GPU die itself will still be TSMC manufactured, although Intel's also said to be lined up to provide 25% of the advanced packaging.

Nvidia's Fenyman GPU architecture was nominally revealed at GTC last year, although details have been scant since then—other than a posited 2028 release date.

Perhaps. Given the state of the tech supply chain right now I'd say all long-term release dates are up for debate—particularly as Rubin, the architecture before Fenyman, is only just beginning to get its time in the sun.

Anyway, it wouldn't be a huge surprise if Nvidia's next, next gen hardware ended up using Intel Foundry tech somewhere in the chain, especially as the two companies now appear to be best buds.

As are some other famous tech providers. Intel's most recent financials looked particularly strong, which is something of a turn around given the previous few years of strife. CEO Lip-Bu Tan has highlighted the importance of foundry customers and partnerships, and getting more tech built on its 18A and 14A processes is seen as a crucial goal for the company.

And with Nvidia cruising from strength to financial strength, it's quite the horse to hitch your wagon to. While an I/O die is not the most important part of a AI GPU, it'd still represent another step forward for team blue if these rumours prove to be true.