At the AI Summit in San Francisco, South Korean President Jae Myung Lee, NVIDIA CEO Jensen Huang, and executives from NAVER, SK Group, and Hyundai Motor Group announced a multi-billion-dollar, full-stack infrastructure pact. The initiative scales regional compute power, builds national AI factories on the NVIDIA Vera Rubin architecture, and establishes joint research facilities with KAIST and Seoul National University.
Scaling the Silicon Backbone: From GAK Sejong to SK Group’s $500-Billion Partnership
NAVER, Brookfield, and NVIDIA are expanding NAVER’s NVIDIA DSX AI factory at the GAK Sejong data center to a massive 200-megawatt footprint. This buildout will harness roughly 100,000 GPUs running on the NVIDIA Vera Rubin platform, more than tripling the initial capacity announced earlier in June.
Simultaneously, SK Group locked in a comprehensive partnership exceeding $500 billion to construct next-generation AI infrastructure. This deployment relies directly on NVIDIA Vera Rubin hardware coupled with SK hynix HBM4 memory stacks. During the summit’s stage events at San Francisco’s Midway venue, SK Group Chairman Chey Tae-won noted the insatiable velocity of enterprise compute demand, pointing out that hardware allocations scale upward continuously to meet modern LLM parameter scaling requirements.
Institutional Engineering: Bridging Academic Rigor and Agentic AI
To counter this, NVIDIA and the Korea Advanced Institute of Science and Technology (KAIST) inaugurated a joint research lab at the KAIST Kim Jaechul Graduate School of AI in Seoul. This marks the first joint AI lab between a Korean university and any global technology company, utilizing NVIDIA Nemotron open models and AI Cloud compute tiers to build localized agentic AI systems optimized for the Korean language.
Further cementing the academic pipeline, NVIDIA and KAIST’s Department of Mechanical Engineering are establishing an NVIDIA AI Technology Center (NVAITC) focusing on physical AI applications. Seoul National University (SNU) is rolling out a parallel NVAITC initiative. SNU’s center will span foundation models, accelerated computing, and AI for science, leveraging both NVIDIA Nemotron and NVIDIA Cosmos open world foundation models for advanced simulation workloads.
Physical AI and Autonomous Deployment
Beyond data centers, industrial robotics and automotive platforms are moving to accelerated hardware. Hyundai Motor Group outlined a physical AI roadmap anchored by a dedicated robot reference platform built alongside NVIDIA. This initiative builds directly on an existing deployment utilizing 50,000 NVIDIA Blackwell GPUs.
Hyundai is also integrating the NVIDIA DRIVE Hyperion platform into its vehicle architectures. This stack combines NVIDIA DRIVE AGX in-vehicle compute modules with the safety-certified NVIDIA DriveOS operating system, standardizing the software-defined vehicle pipeline from development clusters to production fleets.
The Geopolitical and Macro-Market Calculus
During their bilateral meeting at the Fairmont San Francisco, President Lee and Jensen Huang framed these agreements as a foundational pillar of South Korea’s strategy to anchor the global AI supply chain. OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei, speaking alongside regional tech leaders at the summit, emphasized that modern foundation model development remains bottlenecked by specialized memory and hardware supply chains anchored in Korean engineering.
By locking down domestic silicon manufacturing, hyperscale data center construction, and localized foundation model training, South Korea is bypassing traditional cloud dependency models. The nation is establishing a self-sustaining ecosystem designed to weather the next decade of hyper-scale compute expansion.
