Jensen Huang is making a strategic pivot that extends far beyond Nvidia's dominant data center business. This week, the CEO met with leaders from several major Japanese industrial companies to discuss implementing physical AI in their factories, while simultaneously launching joint AI research centers with top South Korean universities. The moves signal that Huang sees the next trillion-dollar opportunity not just in training AI models, but in getting those models to control robots and machines in the real world.

What Is Physical AI, and Why Does Huang Think It's the Next Frontier?

Physical AI refers to artificial intelligence systems that operate in the real world, controlling robots, manufacturing equipment, and autonomous vehicles rather than just processing text or images on servers. Huang has been vocal about this vision, stating that "the next frontier of AI is in the physical world, and this is a once-in-a-generation opportunity for Japan". Unlike data center AI, which runs on massive server farms, physical AI requires specialized computing platforms that can run at the edge, closer to where robots and machines operate.

The timing matters. While Nvidia's data center business is generating $75 billion in quarterly revenue and growing 92 percent year over year, the edge computing segment (which includes robotics, automotive, and personal computers) is still small at $6.4 billion in revenue, though it is growing at a healthy 29 percent annually. For investors focused on near-term stock performance, data centers remain the main story. But Huang is clearly thinking decades ahead.

Which Japanese Companies Just Joined Nvidia's Physical AI Push?

Huang met with leaders from some of Japan's most influential industrial manufacturers to discuss implementing physical AI in their factories. The companies meeting with Nvidia's CEO include:

• Toyota: The world's largest automaker, which has long invested in robotics and manufacturing automation.
• Fanuc: A global leader in industrial robots and factory automation systems.
• Kawasaki Heavy Industries: A major robotics and automation player already using Nvidia's technology.
• Kioxia: A semiconductor manufacturer that produces memory chips for data centers and edge devices.
• Fujitsu Limited: A diversified technology company with significant manufacturing operations.

Three robotics and automation players, Kawasaki, Fanuc, and Yaskawa, are already using Nvidia's technology in their operations. By formalizing these relationships, Huang is essentially locking in long-term partnerships with companies that control some of the world's most advanced factories.

How Is Nvidia Building the Technology Stack for Physical AI?

Nvidia is not simply selling chips to robotics companies. Instead, the company is assembling a complete software and hardware ecosystem designed specifically for physical AI applications. This full-stack approach includes three key components:

• DGX Computing Systems: High-performance computers using Nvidia's Blackwell and Vera Rubin chips, which provide the raw computing power needed to train and run complex AI models.
• Jetson Robotics Platform: Specialized hardware and software designed to run AI models on robots and edge devices, allowing machines to make decisions in real time without constantly communicating with distant data centers.
• Cosmos Simulation Software: A tool that lets engineers simulate the physical world digitally, allowing robot developers to train and test their systems in virtual environments before deploying them in factories.

This approach mirrors Nvidia's strategy in data centers, where the company provides not just chips but an entire ecosystem of software, tools, and partnerships. By controlling multiple layers of the stack, Nvidia makes it harder for competitors like Advanced Micro Devices and Broadcom to chip away at its market dominance.

What Are Nvidia's Plans for AI Research in South Korea?

Beyond Japan, Huang is also investing heavily in AI research infrastructure in South Korea. Nvidia announced two major research partnerships with leading South Korean universities.

The first is a joint AI research center with Seoul National University, called the Nvidia AI Technology Center (NVAITC). The center will focus on foundation models, accelerated computing, physical AI, and scientific AI. Researchers will also work with Nvidia's open-source AI models, including Nemotron and Cosmos, the same simulation software being used for robotics development.

The second partnership involves the Korea Advanced Institute of Science and Technology (KAIST), where Nvidia is establishing the Nvidia-KAIST AI Joint Research Lab within Seoul's Kim Jae-cheol AI Graduate School. This lab will conduct AI research tailored to Korean and domestic industries, with a focus on training the next generation of AI specialists.

Both initiatives signal that Huang views South Korea as a critical hub for AI innovation, particularly in manufacturing and robotics. South Korea is home to major industrial conglomerates like Samsung and Hyundai, which have significant robotics and autonomous vehicle programs.

How to Understand Nvidia's Long-Term Strategy in Physical AI

• Full-Stack Ecosystem: Nvidia is not competing on chips alone; it is building an integrated platform that includes hardware, software, and simulation tools, making it difficult for competitors to displace the company once customers adopt its entire system.
• Geographic Expansion: By establishing research centers in South Korea and meeting with Japanese manufacturers, Huang is securing relationships with some of the world's most advanced industrial economies, locking in demand for years to come.
• Long-Term Valuation Opportunity: The market is not currently pricing in the potential of physical AI, meaning investors who believe in Huang's vision may see significant upside as the edge computing segment grows from $6.4 billion to potentially much larger scale over the next decade.

Why Should Investors Care About Physical AI If Data Centers Are Still Growing?

The short answer is that Nvidia's data center business is already priced into the stock. Nvidia is trading at 23 times forward earnings, with analysts projecting around 44 percent annualized earnings growth over the next few years. While that sounds impressive, the market is not currently assigning any premium valuation to the long-term potential of physical AI, which could represent a massive growth opportunity over the next two to three decades.

Huang is essentially positioning Nvidia to capture the next major computing transition, just as the company did with data centers and AI training. By building relationships with industrial giants in Japan and establishing research centers in South Korea, he is creating a moat that will be difficult for competitors to cross. The company's tailored computing solutions for specific industries, combined with its global relationships with enterprises and AI researchers, give it a significant competitive advantage.

The $1 trillion in confirmed demand for Nvidia's Blackwell and Rubin chips through 2027 provides a strong financial foundation for these longer-term bets. But the real story is that Huang is already thinking about what comes after that trillion-dollar opportunity.