Bet on AI and Independence: NVIDIA and Samsung to Start Custom CPU and XPU Production
According to exclusive information from the TechRadar Pro portal, NVIDIA and Samsung have begun joint development of non-standard processors (CPU) and heterogeneous computing units (XPU) not based on the x86 architecture. This move is aimed at strengthening NVIDIA's position in the face of growing competition from players like OpenAI, Google, AWS, Meta, Broadcom, and others who are actively developing their own chips and AI infrastructure.
What's Known
According to insider information, Samsung Foundry will be the manufacturing partner, providing advanced process technology (possibly 3nm GAA), and NVIDIA will develop custom CPUs and XPUs focused on AI workloads, cloud computing, and autonomous systems.
The architecture they create will not be based on x86, indicating possible use of ARM or RISC-V.
Why It Matters
The project aims to reduce reliance on third-party CPU suppliers like Intel and AMD. NVIDIA already dominates the GPU and AI accelerator segment, but the lack of its own CPUs limits its flexibility in building comprehensive platforms. Joint development with Samsung will allow the company to offer complete computing solutions optimized for its software stacks, including CUDA, TensorRT, and AI Enterprise.
The cooperation of these two giants could change the balance of power in the data center and AI infrastructure market. As NVIDIA's largest customers become its potential competitors, creating its own CPUs and XPUs is a logical step towards strengthening technological independence and expanding the ecosystem.
Source: TechRadar Pro
