Nvidia contemplates Intel as potential manufacturing partner for GPU and AI chips

In a nutshell: Nvidia currently dominates the AI accelerators and GPU chip market. The company relies heavily on manufacturing prowess to sustain its sales momentum, to the extent that it may even seek assistance from Intel.

Nvidia GPUs held more than 70 percent of the data center processor market share in the latest quarter, with the majority of chips being manufactured by the Taiwanese giant TSMC. However, the landscape might change in the not-so-distant future, as the Santa Clara-based corporation is once again hinting at additional partnership opportunities with unexpected “third parties” in the burgeoning chip foundry business.

Speaking at the recent UBS Global Technology Conference, Nvidia’s Chief Financial Officer Colette Kress shared optimistic remarks about the Intel Foundry Services initiative. Kress noted that the company already collaborates with “great” foundry partners for its manufacturing needs, specifically mentioning TSMC and Samsung.

Expressing a desire to work with a third foundry partner, Kress pointed out that Intel’s foundry business recently entered the global top 10 foundry list, thanks to a significant quarterly revenue increase (34.1 percent). While TSMC remains Nvidia’s primary manufacturing partner, Samsung was primarily involved in the development of Ampere GPUs, powering the previous generation of GeForce graphics cards and HPC accelerators.

The Korean giant has collaborated with Nvidia for the past two years to enhance its position as a chip supplier, offering advanced manufacturing processes and integrated services for various component types. The company now aspires to secure a larger share of Nvidia orders for current and next-gen GPU and HPC AI chips.

Nvidia is predominantly focused on its partnership with TSMC. Kress emphasized how the additional manufacturing capabilities being built by the Taiwanese company in the US will undoubtedly be an option to consider. Nevertheless, this doesn’t preclude Nvidia from potentially adding another foundry, such as Intel’s.

The x86 giant has been frequently mentioned as Nvidia’s potential next foundry partner in the US. A collaboration with Nvidia could significantly support Intel in transforming its Foundry Services into a fully-fledged business unit.

Nvidia CEO Jensen Huang previously suggested Intel as a potential manufacturing partner earlier this year, but progress might be slower than anticipated. Huang remarked that the US is still 20 years away from achieving supply chain independence in the chip business.