Samsung will now help NVIDIA make custom CPUs for AI domination

NVIDIA's AI accelerators are a major force behind the development of artificial intelligence. The company wants to sustain this momentum by ensuring that its products cover the entire AI hardware stack.

To that end, the company has inked a partnership with Samsung which will see the Korean giant help NVIDIA make custom non-x86 CPUs and specialized XPUs.

NVIDIA's latest deal with Samsung follows its recent partnership with Intel which enables x86 CPUs to directly connect with NVIDIA's platforms. The company is now expanding its NVLink Fusion ecosystem by tapping Samsung Foundry to help with the design and production of custom CPUs and XPUs.

Samsung is in a unique position to help NVIDIA in this endeavor. The company provides the full range of services, from custom semiconductor design to production, essentially providing NVIDIA with a one-stop solution for its requirements.

It's understandable why NVIDIA wants to push forward with this arrangement. It's countering the rising threat from companies like Google, OpenAI, AWS, Broadcom, and others who are now developing in-house chips to reduce their reliance on NVIDIA's AI hardware.

What NVIDIA is effectively doing is making its hardware unavoidable in the context of AI advancement, ensuring that companies working in this space have to buy hardware from NVIDIA one way or the other.

This works out well for Samsung as well, since the company has made significant investments in its foundry division to provide precisely these services to deep-pocketed customers like NVIDIA.