Which is the more astonishing sight: three billionaires handing out fried chicken in a Seoul diner, or the fact that the same night marked one of the largest AI chip deals in South Korean history?
On Thursday evening, Nvidia CEO Jensen Huang, Samsung Electronics chairman Lee Jae-yong, and Hyundai Motor Group executive chair Chung Eui-sun walked into Kkanbu Chicken, a boisterous fried chicken-and-beer joint in central Seoul. In a country where chaebol leaders rarely appear in such casual public settings, the sight of the trio combined net worth of about $195 billion linking arms for a beer toast drew hundreds of onlookers. Huang, in his signature black leather jacket, passed baskets of chicken and cheese sticks to diners before ringing the restaurant’s “golden bell,” a sign to pay for everyone’s meal. Samsung’s Lee eventually picked up the $1,800 tab and Chung paid for a second round, but the gesture became a viral “Jensanity” moment on Korean social media.
Beneath the conviviality lay strategic intent. Huang was in South Korea ahead of the APEC summit in Gyeongju, where he announced Nvidia would supply more than 260,000 of its most advanced GPUs to South Korean companies and the government. The chips will be deployed into AI-driven manufacturing, robotics, and autonomous vehicles, among others, and to national AI infrastructure. The presidential office said the collaboration covers Samsung, Hyundai, Naver Cloud, and SK Group, with each building AI factories or data centers powered by tens of thousands of GPUs.
Samsung’s new AI mega-factory will combine more than 50,000 Nvidia GPUs with the Omniverse platform, enabling real-time production analysis and optimization in semiconductor, mobile device, and robotics manufacturing. The two firms are also collaborating on HBM4, a next-generation high-bandwidth memory essential for future AI workloads. Hyundai, meanwhile, will leverage 50,000 Blackwell GPUs to accelerate autonomous mobility, smart factories, and robotics with the support of AI research centers in South Korea.
SK Group is building Asia’s first enterprise-led manufacturing AI cloud using Nvidia’s simulation and digital twin platforms to serve public institutions and startups. Naver Cloud is focusing on “Physical AI”- systems where AI operates directly within such industrial environments as shipbuilding, energy, and biotechnology. This integration of AI with physical systems marks a shift toward the deployment of industry-specific AI models in the real world.
This deployment of GPUs at scale is remarkable from the engineering perspective. Nvidia’s Blackwell architecture brings a quantum leap in energy efficiency and computational throughput, thereby allowing complex AI model training and inference with lower operational costs. The NVLink interconnect allows high-speed communication across chips, something that will be crucial for large-scale model parallelism in autonomous driving simulations and robotics control. In Hyundai’s autonomous vehicle projects, this computational power will support sensor fusion, real-time decision-making, and high-fidelity simulation environments-all of which are critical for ensuring safety before deployment on roads.
The push for sovereign AI is South Korea’s path toward reducing dependence on foreign cloud providers and creating domestically controlled AI models. More than 50,000 GPUs will fuel the National AI Computing Center, while additional capacity will be distributed across companies like Kakao and NHN Cloud. It will be this infrastructure that enables “digital twin” factories, or virtual replicas of physical production lines, allowing predictive maintenance, process optimization, and rapid prototyping without disturbing actual operations.
The timing of the announcement was telling. Only the day before, U.S. President Donald Trump and Chinese leader Xi Jinping had met to diffuse trade tensions, a development which could impact future policy on AI chip exports. Nvidia, which once possessed 95% of China’s AI chip market, now has zero share because of U.S. export controls. Huang conceded the geopolitics: “We’d like to see American technology be the global standard.”
For South Korea, already a powerhouse in semiconductors, the deal positions the nation as a regional AI hub with capabilities in vertical integration: from chip design and manufacturing to AI model deployment in industries as diverse as automotive, telecommunications, and heavy manufacturing. The public chimaek spectacle may have been lighthearted, but it underlined a deeper reality-the fusion of social optics with high-stakes engineering and global trade strategy.
