NVIDIA to Hold ‘AI Day’ in Seoul, Share Agentic and Physical AI Use Cases

[Financial News] NVIDIA said on the 2nd that it will hold NVIDIA AI Day Seoul 2026 at COEX in Seoul for two days, from the 9th to the 10th. The event will showcase examples of technology applications by Korean companies and researchers, focusing on agentic artificial intelligence (AI), physical AI and AI infrastructure.
About 2,000 developers, companies, startups and researchers are expected to attend. On the 9th, the program will feature a keynote-style plenary session and presentations across three topic areas. On the 10th, paid hands-on workshops for developers and invited events will be held. Registration is closed, and major sessions will be available to watch on demand after the event through NVIDIA On-Demand.
At the plenary session on the morning of the 9th, NVIDIA leaders will discuss the development of physical AI, agentic AI and AI factories. They are expected to present the technologies and infrastructure strategies needed for robotics, edge AI, enterprise AI agents and building AI factories.
In the afternoon, 15 sessions will be held across three tracks: agentic AI, physical AI and robotics, and AI infrastructure. NAVER Cloud will share its experience with open models using NVIDIA Nemotron and training large-scale models. SK hynix and Samsung Electronics will present ways to accelerate semiconductor process and device simulations and physics analysis using NVIDIA PhysicsNeMo.
In robotics, LG Electronics will present a robot training and validation system using NVIDIA Isaac GR00T, along with examples of its application in factories. Doosan Robotics will discuss methods for simulating complex physical phenomena such as contact and deformation, while Superb AI will cover ways to use synthetic data in robot development.
In the AI infrastructure track, SK Telecom Co., Ltd. (SKT) will present the challenges of building AI factories spanning power grids to semiconductors. NAVER Cloud will share its experience operating clusters made up of thousands of graphics processing units (GPUs) and responding to outages, while Coupang Intelligent Cloud will introduce the training and inference architecture of infrastructure based on NVIDIA GB300 NVL72.
On the 9th, a program called ‘Build-a-Claw’ will run in the COEX lobby, where participants can build AI agents themselves using OpenClaw and NVIDIA NemoClaw. Attendees can also experience demonstrations of always-on agents and physical AI.
On the 10th, the NVIDIA Deep Learning Institute (DLI) will hold hands-on workshops. The workshops will cover practical techniques for deploying and optimizing large-scale AI inference, robotics simulation and training, and developing safe agents based on large language models (LLMs).
During the event, Korean researchers will present examples of AI science research, and an investment-matching event for physical AI startups will also be held. In the first-floor lobby, attendees can consult with NVIDIA experts in AI infrastructure, LLMs and robotics, and take an NVIDIA certification exam at the associate level free of charge.
[email protected] Yoon Hong-jip Reporter