Intel: A Potential New Partner for NVIDIA Chip Production

According to a report by DigiTimes, Colette Kress, the Chief Financial Officer of NVIDIA, was recently queried at the UBS Global Technology Conference about the possibility of considering Intel as a foundry partner for their next-generation chips. In response, Colette Kress acknowledged the presence of many formidable foundries in the market. TSMC is a notable example, and Samsung also ranks highly. As for NVIDIA’s contemplation of a third foundry, the answer is affirmative.

Currently, most of NVIDIA’s GPUs are manufactured by TSMC, encompassing a vast array of products used for Artificial Intelligence (AI) and High-Performance Computing (HPC) in data centers, as well as the new generation Ada Lovelace architecture GPUs aimed at gamers. Samsung, on the other hand, mainly produces GPUs based on the previous generation’s Ampere architecture. In the short term, the production capacities of TSMC and Samsung are sufficient to meet NVIDIA’s needs.

Beyond the usual concerns of cost and capacity, NVIDIA must also consider other factors. Hence, the impressive technical capabilities of Intel’s foundry services (IFS) have naturally caught NVIDIA’s attention. As early as May of this year, during a media interaction session at Computex 2023, NVIDIA’s founder and CEO Jensen Huang had expressed efforts towards diversifying chip manufacturing. He mentioned that they had received test chips manufactured by Intel’s foundry services, and the results appeared promising.

At that time, Huang did not reveal the specifics of these chips nor the particular manufacturing process used. However, NVIDIA’s actions suggest that the plan to incorporate a third foundry has been under consideration for some time.