AMD, South Korea Launch AI Research Center to Challenge GPU-Centric Model

South Korea's Ministry of Science and ICT (MSIT) signed a memorandum of understanding with AMD on July 23, 2026, establishing an AI Research Center of Excellence to build heterogeneous computing infrastructure that moves beyond GPU-only architectures.

The agreement was formalized during a meeting between MSIT Deputy Prime Minister Bae Kyung-hoon and AMD Chair and CEO Lisa Su at the AMD Advancing AI 2026 event in San Francisco. The partnership integrates AMD's CPUs and GPUs with South Korea's domestically produced Neural Processing Units (NPUs) to create a more open and diversified AI computing ecosystem.

A Shift From Single-Vendor Stacks

The collaboration addresses demand for heterogeneous architectures capable of handling inference workloads for generative and agentic AI more efficiently. Rather than depending on a single vendor's GPU-centric stack, the initiative combines multiple computing resources into unified infrastructure.

"The global AI competition is shifting toward heterogeneous computing that combines various computing resources such as CPUs, GPUs, and NPUs, and an open ecosystem," Bae said at the signing.

The AI Research Center of Excellence will serve as a hub for technical cooperation, software development, and joint research among South Korean universities, research institutions, and domestic companies. AMD will provide high-performance computing resources, AI software, and technical experts to support AI-based scientific research centered on the National AI Science Research Center (NAIS).

Education and Next-Generation Research

AMD plans to support education and research programs at major South Korean universities and graduate schools, with joint research initiatives targeting next-generation AI fields including physical AI. The company will also pursue collaboration with memory, CXL, and server companies to create a comprehensive computing environment.

"South Korea is a core strategic market for AMD with world-class competitiveness in semiconductors, AI, advanced research, and physical AI," Su said.

Strategic Implications

For South Korea, the partnership strengthens its domestic AI semiconductor ecosystem by integrating local NPU technology with global leaders. The country aims to reduce dependence on single-vendor AI stacks while fostering homegrown chip capabilities.

For AMD, the agreement solidifies its footprint in a market with deep semiconductor expertise and manufacturing capacity. By promoting open, heterogeneous infrastructure, the company positions itself as an alternative to the dominant GPU-centric market structure.

The global AI semiconductor market has recognized limitations of GPU-only architectures as inference demands grow. Heterogeneous systems that distribute workloads across CPUs, GPUs, and specialized NPUs offer improved efficiency and flexibility for diverse AI workloads.

The MOU marks one of several international partnerships AMD announced at Advancing AI 2026, as chipmakers race to secure footholds in key markets while governments seek to build sovereign AI capabilities.