Nvidia expands South Korea partnerships to advance physical AI

The shift to physical AI

Nvidia is currently focusing on the transition from agentic AI to what it calls physical AI. While agentic AI refers to systems that can make judgments and act autonomously in digital spaces, physical AI is designed to perform tasks in the real world. This evolution is a key part of the current robotics landscape.

Deepu Talla, a Vice President at Nvidia, noted that the golden age of robotics has started because AI models and simulation technologies reached a level of maturity last year. He highlighted that South Korea is a primary location for this growth due to its long history in heavy industry and electronics manufacturing.

The goal is to move beyond the automation methods used over the last 30 to 40 years. In the past, robots in factories were programmed to follow specific, rigid instructions. Physical AI allows robots to handle environments that are constantly changing, which is nearly impossible to do with traditional programming.

Simulating the factory floor

A major part of this transition involves using virtual worlds to train robots before they ever touch a physical object. Nvidia is using its Omniverse platform to create virtual production lines where industrial robots can practice handling car bodies. This allows companies to test workflows in a risk-free digital environment.

For specific tasks, Nvidia is utilizing its Isaac robotics development platform and the Isaac Lab learning platform. These tools allow for high-fidelity physics simulations, such as the Newton engine. These technologies help robots learn complex movements by simulating the physics of the real world with high precision.

One specific example involves a partnership with Doosan to develop robots that can polish automotive door surfaces. The simulation allows the robotic arm to analyze surface roughness and the specific contact pressure of the polishing pad. By practicing in a digital twin, the robot learns the nuances of the material before being deployed on a real assembly line.

Industrial applications in Korea

Nvidia's expansion in Korea includes several major industrial players across different sectors. The company is working with Hyundai Motor Group on automotive manufacturing processes and with LG Electronics on robot learning and verification. These partnerships aim to integrate AI into the core of Korean manufacturing strengths.

In the semiconductor space, Nvidia is working with SK hynix and SK Telecom on digital twin technology. This involves creating virtual versions of large-scale equipment where AI agents can analyze the causes of anomalies. This helps maintain complex factories by predicting and identifying issues before they cause significant downtime.

Nvidia is also engaging with the startup ecosystem, such as the company RealWorld. Together, they are working on technology to evaluate how well robots can manipulate objects. These diverse partnerships show that Nvidia is targeting every level of the industrial chain, from large corporations to specialized startups.

Accelerating development speeds

One of the most significant impacts of these new technologies is the speed at which companies can develop and deploy new robotics. Talla explained that construction work for simulation environments, which used to take multiple experts several months, can now be finished by a single engineer in about one week.

This acceleration is made possible by AI agents that can significantly speed up the robotics development lifecycle. By providing the core technologies, Nvidia acts as a platform company. They do not manufacture the physical robots themselves but provide the underlying software and simulation tools that others use to build them.

This approach allows for a more modular way of building industrial automation. Instead of starting from scratch, companies can use Nvidia's established platforms to quickly iterate on specific tasks. This reduces the barrier to entry for companies looking to adopt sophisticated AI in their manufacturing plants.

The role of Korea in AI

Nvidia views South Korea as a critical hub for the next decade of physical AI development. The company pointed to the country's established capabilities in electronics and automotive manufacturing as a foundation for this growth. The goal is to create a collaborative environment where these industries can leverage AI to stay competitive.

Nvidia Korea's leadership is encouraging a broad exchange of knowledge between domestic companies, research institutes, and developers. They believe that the rapid development of AI requires a connected ecosystem where different entities can share insights on agentic and physical AI. This collaborative approach is intended to foster a more robust AI infrastructure in the region.

As the technology matures, the focus will likely shift toward more complex, autonomous tasks in real-world settings. By establishing these partnerships now, Nvidia and its Korean partners are aiming to set the standard for how AI interacts with the physical world. The next decade will determine how these simulations translate into widespread industrial adoption.

Nvidia robotics chief calls Korea key to physical AI, highlights partnerships with…