Skip to main content

Nvidia robotics chief highlights Korea’s role in physical AI

9 NOVEMBER 2026·2 MIN READ·4 SOURCES

Deepu Talla, NVIDIA’s vice president of Robotics and Edge AI, cited Korea’s long-standing capabilities in heavy industry, automotive and electronics manufacturing. He spoke at an event held Nov. 9 at COEX in Seoul.

Nvidia robotics chief highlights Korea’s role in physical AI

Key takeaways · 4

  • 01

    NVIDIA is working with Hyundai Motor Group to model auto production plants as digital twins using Omniverse and Isaac.

  • 02

    NVIDIA and Doosan Robotics are developing robots to polish automotive body surfaces, using Isaac Lab and the Newton physics simulation engine.

  • 03

    NVIDIA is working with SK hynix and SK Telecom on large-scale AI factories and with LG Electronics on robot training and validation for home-appliance manufacturing.

  • 04

    NVIDIA says AI agents can reduce the time needed to build robot-development simulation environments from several months to within a week.

Why Korea featured in the remarks

Talla cited Korea’s long-standing capabilities in heavy industry, automotive and electronics manufacturing.[2] He said the “golden age” of robotics had begun as AI models and simulation technologies matured.[2] Physical AI refers to AI that can understand real-world physical laws and surroundings and act through robots.[3] That framing connects the event’s manufacturing focus with the challenge of developing robots that operate in physical environments, rather than only processing information digitally. The remarks were delivered at an event held Nov. 9 at COEX in Seoul.[1][2]

Factory projects span simulation and training

NVIDIA is working with Hyundai Motor Group to implement auto production plants as digital twins, using its Omniverse virtual-environment platform and Isaac robotics development platform.[3][2] Separately, NVIDIA and Doosan Robotics are developing robots to precisely polish automotive body surfaces.[3] That work uses Isaac Lab for robotics learning and Newton as a physics simulation engine.[2] The projects point to different uses for simulation: representing production plants digitally and developing robot capabilities for a specific manufacturing task.

More partners, more applications

NVIDIA is working with SK hynix and SK Telecom to build large-scale AI factories.[3] With LG Electronics, it is advancing robot training and validation for home-appliance manufacturing.[3] NVIDIA is also collaborating with South Korean startup RealWorld on technology to evaluate robots’ object-manipulation capabilities.[2] For teams assessing potential applications, the distinction between work under development and demonstrated results matters.

Tools and limits of the announcement

NVIDIA lists Isaac, Omniverse and the Cosmos world model among its physical-AI tools.[3] The company says AI agents can cut the time to build robot-development simulation environments from several months to within a week.[3] Talla contrasted the current moment with car-factory automation over the past 30 to 40 years, which he said relied on programmed robots rather than AI.[2] NVIDIA AI Day Seoul 2026 was expected to draw around 2,000 developers, business representatives and researchers, with 20 partner companies taking part.[1] Some attendees were disappointed by Madison Huang’s sudden absence.[1]

For manufacturing and robotics teams, the announcements offer concrete examples of how simulation, robot learning and AI are being applied to production tasks. The reported collaborations are work in progress, so teams evaluating them should distinguish announced development from proven performance.

Why it matters
Story quiz

Bring this to your team — one session a day, built for your industry.

kju.ai is built for teams: daily sessions, team analytics, and rollout support for your organization.

For teams

How this developed

  1. 9 November 2026

    Nvidia robotics chief highlights Korea’s role in physical AI

Sources

AI fluency, one session a day, built for your work.