Colette Kress, chief financial officer of NVIDIA, has hinted once again at a possible collaboration between Intel Foundry Services and NVIDIA for the manufacture of next-generation chips.
NVIDIA describes Intel as a possible “Third” foundry partner, Samsung as a current partner, and TSMC as a fantastic partner.
NVIDIA’s CFO was asked whether the company would think about approaching Intel as a Foundry partner for the manufacturing of their next-generation chips at the most recent UBS Global Technology Conference. Based on their response, the CFO responded to this in a very good manner.
A generation ago, Samsung was the company responsible for creating NVIDIA’s gaming GPUs, whereas TSMC now produces the majority of the company’s data center GPUs for AI/HPC and gaming chips. The Ampere GPU family from NVIDIA, which powers the GeForce RTX 30 “Gaming” graphics cards, was developed at the Samsung Fabs. But, as we’ve covered in-depth in our earlier coverage, Samsung wants to supervise a much more significant part of “nourishing” NVIDIA’s data center earnings in the next years, since the company has now established a strong partnership with Team Green.
In order to maintain its momentum in the AI market share, it is anticipated that TSMC will continue its important cooperation with NVIDIA for the manufacturing of Hopper H200 and Blackwell B100 GPUs. Samsung will remain accessible in the event that more orders are needed. In addition, NVIDIA hopes to have a diverse network of foundry partners and is willing to work with a third (in reference to Intel). What NVIDIA said was as follows:
There are, in my opinion, several excellent foundry partners. TSMC has performed well. You are aware that we still use Samsung nowadays. Would we be up for a third? Yes. A third one would be wonderful. And to do that, it requires research about their service-related interests. Remember that there are others who may go to the United States. For us, TSMC in the US may also be a possibility. Again, in terms of the distinct area, but not necessarily dissimilar. Nothing that would prevent us from perhaps establishing a second foundry.
NVIDIA CFO Colette Kress addresses the UBS Global Technology Conference
Given that the Korean behemoth has improved its operations and has been working with Team Green consistently over the last two years to establish itself as a dependable supplier, Samsung Foundry has structured its services in a manner that has forced Team Green to raise the amount of orders it places with the company.
With its wide range of equipment suitable for semiconductor, memory, and packaging stages, Samsung has an advantage over other foundry competitors. This not only gives NVIDIA a “hybrid partner,” but Samsung may also benefit if it can offer competitive pricing and delivery times.
Intel’s name has been brought up by industry sources as a foundry partner. Even Jensen Huang, the CEO of NVIDIA, made reference to future processors that will use Intel’s Foundries. Given that Intel wants to work with additional IFS partners in the future, a partnership with NVIDIA would undoubtedly be significant.
For NVIDIA and other chip behemoths like AMD and even Intel, who depend on TSMC for the manufacture of certain intellectual property in the chips, having US-based Foundry partners will be advantageous as well. However, the CEO of NVIDIA thinks that the US is still decades away from becoming supply chain independent and that the first significant step toward accomplishing this objective would be attracting TSMC and other significant fabs to the US.
