NVIDIA Partners with Samsung Foundry for Custom AI Silicon Production, Signaling a Significant Milestone for the Korean Tech Leader

NVIDIA has officially welcomed Samsung Foundry as a strategic partner in the development of custom CPUs and XPUs, utilizing the latter’s extensive expertise in design and manufacturing.

Samsung Foundry’s Role in Custom CPUs & XPUs within NVIDIA’s NVLink Ecosystem

A major highlight from the recent OCP Global Summit was NVIDIA’s announcement of Samsung Foundry joining its list of collaborators, a move that signifies a pivotal advancement for both NVIDIA and Samsung. In a blog post detailing this partnership, NVIDIA confirmed that Samsung Foundry has been integrated into the NVLink Fusion ecosystem, where it will leverage its capabilities to design custom silicon and provide essential manufacturing support for diverse customers. Though specific details about the partnership’s scope remain undisclosed, it is evident that Samsung Foundry has firmly allied itself with NVIDIA’s initiatives.

Intel and Samsung Foundry are joining the NVLink Fusion ecosystem that includes custom silicon designers, CPU and IP partners, so that AI factories can scale up quickly to handle demanding workloads for model training and agentic AI inference.

Samsung Foundry has partnered with NVIDIA to meet growing demand for custom CPUs and custom XPUs, offering design-to-manufacturing experience for custom silicon.

This collaboration extends beyond mere manufacturing; Samsung Foundry will deliver comprehensive ‘end-to-end’ support for custom silicon projects. This involves a complete process that includes silicon design, verification, integration, and tape-out. The aim is to meet the escalating demand for specialized CPUs and XePUs, facilitating NVIDIA’s access to custom silicon developed by Samsung Foundry without directly competing with NVIDIA’s own products.

NVIDIA Blackwell chips — Image Credits: NVIDIA

The strengthening of ties between Samsung and NVIDIA is notable, especially following a recent meeting between NVIDIA CEO Jensen Huang and Samsung Chairman Lee Jae-yong. This meeting played a crucial role in securing approval for the supply of HBM3E. The newly formed NVLink Fusion partnership suggests that companies, including OpenAI, may turn to Samsung Foundry for manufacturing needs, thereby ensuring that their chips can take advantage of NVIDIA’s NVLink ecosystem. Although this alliance could potentially open avenues for Intel to engage with Samsung for its data-center-focused x86 CPUs, such a collaboration seems improbable at this point.

The incorporation of Samsung Foundry into NVIDIA’s NVLink ecosystem underscores the Korean tech giant’s preparedness to address the surging demand for custom silicon from major technology players. This readiness is further bolstered by advanced manufacturing processes, such as the SF2 (2nm), which have captured significant industry interest.

Source & Images