Key facts
- South Korea plans to invest over 1 quadrillion won (US$650 billion) by 2035 to build AI data centers.
- An initial 550 trillion won investment will create 8.4 GW of AI data center capacity by 2029.
- The government aims to designate the AI sector as a national strategic industry.
- The plan includes tax breaks and the formation of a cluster ecosystem to support data center equipment and solutions.
- South Korea aims to develop a general-purpose foundation model for physical AI within three years.
South Korea is set to invest over 1 quadrillion won (US$650 billion) by 2035 to construct artificial intelligence (AI) data centers, aiming to bolster its position in the global AI landscape. Science Minister Bae Kyung-hoon announced the plan, emphasizing the critical role of ample data in the ongoing AI race.
The initiative includes an initial investment of 550 trillion won to build 8.4 gigawatts (GW) of AI data center capacity by 2029. The ministry intends to gradually expand this infrastructure by an additional 10 GW until 2035. To promote balanced regional development, the new infrastructure will be concentrated in areas outside the capital region.
Minister Bae highlighted that global investment in data centers is projected to reach US$5.5 trillion (8.4 quadrillion Korean won) over the next five years, citing Nvidia's chief Jensen Huang's description of the current AI boom as the "largest infrastructure build up in human history." The ministry also plans to support the localization of equipment and solutions for these data centers through tax incentives and the establishment of a cluster ecosystem and large-scale test laboratories.
This data center development is part of a broader national strategy. President Lee Jae Myung has previously announced an 800 trillion-won investment plan for semiconductor and AI competitiveness, and SK Group has committed 2,100 trillion won to build an AI memory production belt. Following the data infrastructure build-out, the science ministry aims to develop a general-purpose foundation model for physical AI within the next three years, utilizing world models to understand real-world dynamics. The ultimate goal is to boost productivity in key industries by 20% and create a significant competitive edge through Physical AI.