Cars and factories evolving from isolated systems into a single interconnected ecosystem
On the final day of October, in the ancient Korean city of Gyeongju, Hyundai Motor Group and NVIDIA formalized a $3 billion partnership that reaches beyond a corporate agreement into something more consequential: the deliberate construction of an AI-powered industrial ecosystem. By weaving together 50,000 Blackwell GPUs, digital twin environments, and embedded vehicle intelligence, the two companies are not merely upgrading their tools — they are redesigning the relationship between human ingenuity, machine learning, and the physical world. South Korea's government stands alongside them, recognizing that the nation's future competitiveness may depend on mastering what officials call 'physical AI' — the art of making artificial intelligence work not in the abstract, but in factories, on roads, and inside the vehicles that carry human lives.
- A $3 billion AI factory anchored by 50,000 NVIDIA Blackwell GPUs is rising in Korea, representing one of the most ambitious industrial AI deployments any automaker has ever attempted.
- The pressure is existential: automakers worldwide are racing to embed AI into every layer of production and mobility, and falling behind now risks irrelevance in a multitrillion-dollar industry being fundamentally rewritten.
- Hyundai and NVIDIA are not simply sharing software — they are co-developing the foundational infrastructure for autonomous vehicles, humanoid factory robots, and over-the-air vehicle intelligence that evolves long after a car leaves the showroom.
- Korea's government has signed on as an active partner, committing to AI application centers and data centers that will train the next generation of engineers, treating this collaboration as a national industrial strategy rather than a private venture.
- The initiative is converging toward a single interconnected ecosystem — digital twins of factories, simulated driving environments, and AI-powered vehicles — where the virtual and physical worlds inform each other in real time.
On the last day of October, Hyundai Motor Group and NVIDIA announced a partnership in Gyeongju, South Korea, that goes well beyond a typical technology agreement. Together, they are building an AI factory — a vast computational infrastructure powered by 50,000 NVIDIA Blackwell GPUs — designed to train, validate, and deploy artificial intelligence across vehicle production and autonomous driving. The $3 billion investment is backed by South Korea's Ministry of Science and ICT, which views the project as central to the nation's ambitions in physical AI: the application of machine intelligence to real-world manufacturing and mobility.
The partnership rests on three technological pillars. NVIDIA's DGX systems will handle the heavy computational work of training large AI models. Omniverse and Cosmos platforms will generate digital twins — virtual replicas of factory floors and driving environments — where engineers can test and refine systems before anything touches the physical world. And NVIDIA DRIVE AGX Thor, a safety-certified compute platform, will serve as the embedded AI brain inside Hyundai vehicles, enabling real-time autonomous driving decisions and intelligent in-vehicle experiences.
Hyundai's factories will be transformed through these digital twins, unifying manufacturing data to optimize production, integrate humanoid and robotic systems, and anticipate equipment failures before they occur. On the vehicle side, Hyundai will develop proprietary large language models using NVIDIA's open tools, powering personalized assistants, adaptive comfort systems, and infotainment — all updatable over the air, so a car's intelligence can grow throughout its lifetime.
Korea's government sees the collaboration as a public-private model worth replicating. Deputy Prime Minister Bae Kyung-hoon described it as a 'Win-Win' that pairs Korea's manufacturing depth with NVIDIA's infrastructure. New facilities — including a Hyundai Physical AI Application Center and an NVIDIA AI Technology Center — will also serve as training grounds for Korea's emerging generation of AI engineers.
Hyundai's Executive Chair Euisun Chung called the moment pivotal for both the company and the country. NVIDIA's Jensen Huang framed the stakes even more broadly, noting that AI will reshape transportation — from design and manufacturing to robotics and autonomy — across a multitrillion-dollar sector. What is being built in Korea is not simply a factory or a research hub. It is the infrastructure through which the next decade of automotive innovation will flow.
On the last day of October, in the South Korean city of Gyeongju, Hyundai Motor Group and NVIDIA formalized a partnership that signals a fundamental shift in how the world's automakers will build cars and factories. The two companies announced they are constructing an AI factory—a sprawling computational infrastructure powered by 50,000 NVIDIA Blackwell GPUs—designed to train, test, and deploy artificial intelligence across every phase of vehicle production and autonomous driving. The investment totals approximately $3 billion, backed by South Korea's Ministry of Science and ICT, which sees the project as essential to the nation's ambitions in what officials call "physical AI," the application of AI to real-world manufacturing and mobility systems.
This is not a minor technology upgrade. Hyundai and NVIDIA are moving beyond simply adopting each other's software platforms. They are now co-developing the foundational AI technologies that will power the next generation of cars and factories. The partnership rests on three pillars of NVIDIA infrastructure. First, the DGX platform and related systems will handle the raw computational work of training massive AI models. Second, NVIDIA Omniverse and Cosmos—running on RTX PRO servers—will create digital twins, virtual replicas of manufacturing floors and driving environments where engineers can test and refine systems before they touch the physical world. Third, NVIDIA DRIVE AGX Thor, a safety-certified compute platform, will serve as the "AI brain" embedded in vehicles themselves, handling real-time decisions for autonomous driving and in-vehicle intelligence.
The scale of this undertaking becomes clearer when you consider what Hyundai plans to build. Using Omniverse Enterprise, the company will construct digital twins of its factories—virtual environments that unify all manufacturing data and allow engineers to optimize production, test robots before they're deployed on actual assembly lines, and predict maintenance needs before equipment fails. These simulations will accelerate the integration of humanoid and robotic systems into production, moving toward what Hyundai describes as fully autonomous, software-defined factories. On the vehicle side, Hyundai will use NVIDIA's Nemotron open models and NeMo software tools to develop proprietary large language models that power everything from personalized digital assistants to intelligent infotainment systems and adaptive comfort controls. These systems will be updated over the air, meaning a car's capabilities can evolve throughout its lifetime.
For autonomous driving specifically, Hyundai is building digital twins of regional driving environments and conditions, running sophisticated simulations to accelerate its development pipeline. The DRIVE AGX Thor platform will provide the compute power for advanced driver-assistance systems, next-generation safety features, and the immersive in-vehicle AI experiences that distinguish modern vehicles from their predecessors. All of this—the factory robots, the autonomous systems, the in-vehicle AI—will operate as a single interconnected ecosystem rather than isolated technologies.
The Korean government sees this partnership as a model for how public and private sectors can collaborate on critical infrastructure. Bae Kyung-hoon, South Korea's Deputy Prime Minister and Minister of Science and ICT, emphasized that physical AI advancement is essential for Korea to lead globally, and that combining Korea's manufacturing expertise with NVIDIA's infrastructure creates a "Win-Win model" that can accelerate AI transformation across industries. The collaboration will establish both a Hyundai Physical AI Application Center and an NVIDIA AI Technology Center, along with physical AI data centers in the region. These facilities will also serve as training grounds for Korea's next generation of AI engineers and technicians.
Euisun Chung, Executive Chair of Hyundai Motor Group, framed the partnership as a pivotal moment for the company and the nation. "Together, we are not only building advanced technologies but also laying the foundation for a robust AI ecosystem in Korea," he said. NVIDIA's Jensen Huang echoed the scale of ambition, noting that AI will revolutionize every industry, and that in transportation alone—from vehicle design and manufacturing to robotics and autonomous driving—the transformation will reshape a multitrillion-dollar sector. What Hyundai and NVIDIA are constructing in Korea is not just a factory or a research center. It is the infrastructure through which the next decade of automotive innovation will flow, and a template for how other nations and companies might build their own physical AI capabilities.
Notable Quotes
For Korea to leap forward as a leading nation in AI, the advancement of physical AI is essential. By combining Korea's rich manufacturing data with NVIDIA's cutting-edge AI infrastructure, we expect to build a Win-Win model through collaboration with domestic companies.— Bae Kyung-hoon, South Korea's Deputy Prime Minister and Minister of Science and ICT
Together, we are not only building advanced technologies but also laying the foundation for a robust AI ecosystem in Korea—one that fosters innovation, nurtures talent, and positions us at the forefront of global AI leadership.— Euisun Chung, Executive Chair of Hyundai Motor Group
The Hearth Conversation Another angle on the story
Why does a car company need 50,000 GPUs? That seems like an enormous amount of computing power.
Because training AI models for autonomous driving, manufacturing optimization, and in-vehicle intelligence requires processing vast amounts of data—millions of hours of driving scenarios, sensor inputs, factory operations. You can't do that on a laptop. The scale of computation needed to validate that a self-driving system is safe across infinite edge cases demands industrial-grade infrastructure.
But why is South Korea investing $3 billion in this? What's at stake for the government?
Korea sees physical AI as a strategic advantage. They have world-class manufacturing expertise and a strong industrial base, but they're competing with China and the U.S. for leadership in AI. By anchoring NVIDIA's infrastructure and Hyundai's innovation in Korea, they're building an ecosystem that keeps talent, investment, and intellectual property within the country.
The announcement mentions digital twins. What does that actually mean for a factory?
A digital twin is a virtual replica of a physical factory—every machine, every process, every workflow. Engineers can run simulations, test new robot configurations, predict when equipment will fail, all without stopping production. It's like having a perfect sandbox where you can experiment before touching the real thing.
And the cars themselves—what changes for a driver?
Over time, quite a bit. The DRIVE AGX Thor platform gives vehicles the compute power for real-time autonomous driving decisions, but also for AI assistants that learn your preferences, infotainment systems that adapt to you, safety systems that anticipate hazards. And because these can be updated over the air, the car gets smarter throughout its life, not just when you buy it.
Is this partnership exclusive to Hyundai, or could other automakers use this infrastructure?
The announcement focuses on Hyundai's use, but the Korean government's involvement suggests they're building a national physical AI cluster—infrastructure that could eventually serve other domestic companies. That's part of the ecosystem-building strategy.