Korea’s $1.3T AI mega‑projects target chips, physical AI and data centers

Korea is staking out an aggressive position in the AI race by fusing its strengths in semiconductors and heavy industry into a national AI industrial strategy. The three mega‑projects outlined by President Lee—chips, physical AI and hyperscale AI data centers—aren’t just budget lines; they’re an attempt to lock in Korea as a manufacturing and deployment hub for the compute infrastructure AGI will require. Building 8.4 GW of AI data center capacity with SK, GS and Naver and compressing fab timelines by up to a decade is a clear bid to compete with the US, China and Taiwan on the “steel and silicon” of AI.

The emphasis on “physical AI,” with thousands of robots deployed annually into factories, logistics and defense, is particularly interesting. It pushes AI beyond cloud APIs into embodied systems and aligns with global trends toward agentic and robotics‑centric AI. If Korea really does field a top‑tier physical AI base model in three years, it could become the go‑to platform for manufacturers worldwide looking to retrofit plants with intelligent automation.

For frontier labs and cloud providers, this creates both a market and a competitor. Korean champions backed by this policy could become preferred partners for global companies wanting politically diversified supply chains and AI infrastructure outside US‑China fault lines.