SEOUL — The South Korean government is launching a massive strategic offensive to secure a "super-gap" in Physical AI, a field where artificial intelligence intersects with the physical world through robotics and autonomous systems. The Ministry of Science and ICT (MSIT) announced on March 9, 2026, that it will invest 170 billion won this year to develop indigenous core technologies, including lightweight Vision-Language-Action (VLA) models and digital twin simulations.

Bridging the Virtual and Physical Worlds
The initiative is built on a two-track strategy: securing universal core technologies and accelerating regional AI Transformation (AX). At the heart of this project is the development of "World Models"—AI systems that understand physical laws and environmental contexts to enable robots to make autonomous decisions.

A total of 34 billion won will be allocated through 2027 to create high-fidelity virtual environments. These digital twins will generate synthetic data to train AI in a risk-free space, bridging the "sim-to-real" gap. Industry giants and rising tech firms are already moving; consortia led by NC AI (including Samsung SDS) and LG Electronics (partnering with Maum AI) are currently undergoing technical evaluations to lead these projects.

On-Device Autonomy and Hardware Optimization
A key differentiator in Korea's strategy is the focus on On-Device VLA models. The goal is to enable robots to process natural language commands and perform complex physical tasks without a persistent internet connection. To support this, the government will invest 38.8 billion won over four years into software environments that optimize AI semiconductors for physical movement.

"Our goal this year is to implement Physical AI within limited computing infrastructures and integrate AI into manufacturing sites using real-time field data," an MSIT official stated.

Regional Empowerment: From Labs to Smart Factories
The government is also pouring 143.4 billion won into regional hubs like Jeonbuk and Gyeongnam to apply these technologies to heavy industries:

Jeonbuk Province: Focusing on "Software Defined Factory" (SDF) technology and AI Meta-factories to standardize global manufacturing protocols.
Gyeongnam Province: Developing specialized AI models for high-precision tasks like welding and painting, aiming to replicate the expertise of veteran human laborers.
By securing synthetic data generation and virtual training technologies, South Korea aims to drastically reduce the cost of AI training while creating high-value-added services in logistics and manufacturing. This bold roadmap signals the nation’s intent to transition from a hardware powerhouse to a leader in the cognitive-physical AI era.