On July 23, 2026, at the AMD Advancing AI event in San Francisco, South Korea's Ministry of Science and ICT signed a memorandum of understanding with AMD to establish an AI Center of Excellence and build an open heterogeneous computing infrastructure. The deal combines AMD's CPUs and GPUs with neural processing units (NPUs) from South Korean companies to boost computing efficiency and cut costs as demand for generative AI inference grows.
From GPU to Heterogeneous Computing
The partnership signals a shift in AI computing architecture. As agentic AI and large-scale inference workloads increase, the industry is moving from single-GPU setups to heterogeneous systems that mix CPUs, GPUs, and NPUs on demand. This approach can improve performance while reducing power consumption and deployment expenses.
Bae Kyung-hoon, South Korea's Deputy Prime Minister and Minister of Science and ICT, said at the signing: "Global AI competition is shifting toward heterogeneous computing and open ecosystems. Through our partnership with AMD, we will enhance our domestic AI semiconductor competitiveness and establish a new cooperation model to address changes in global AI computing infrastructure."
R&D Implementation and Talent Development
Beyond hardware, AMD will take part in South Korea's national-level AI scientific research initiatives. The company will supply high-performance CPUs, GPUs, and AI software to the country's National Scientific AI Research Center and send technical experts to advance collaborative work in the AI for Science & Research field.
AMD also plans to work with several South Korean universities and research institutions. It will offer students and researchers development environments and training resources built on the AMD AI platform and conduct joint research on next-generation technologies such as physical AI.
Lisa Su, AMD's CEO, said: "South Korea possesses global competitiveness in semiconductors, AI, frontier research, and physical AI, and is a core strategic market for AMD. This collaboration is not only about hardware support, but also reflects AMD's long-term commitment to contributing to South Korea's AI innovation and economic growth through technological cooperation, talent cultivation, and close collaborative R&D."
Nvidia had already forged partnerships with Korean institutions including KAIST and Seoul National University. The arrival of a second major AI chip vendor deepens South Korea's role as a hub for AI infrastructure, backed by its mature semiconductor supply chain and government policy support.
Why this matters for Science & Research professionals
The new center and computing infrastructure will give researchers direct access to heterogeneous systems that combine AMD processors with specialized Korean NPUs. This can accelerate AI for Science projects-from climate modeling to materials discovery-by offering more efficient, cost-sensitive hardware configurations. The talent development programs and collaborative research opportunities with AMD's technical experts also create paths for scientists and research engineers to build skills in emerging areas like physical AI, without leaving their home institutions.
Your membership also unlocks:
