Samsung and SK are pouring trillions of won into regional areas. The center of gravity of South Korea’s advanced industries is shifting beyond the capital region.
Samsung on the 29th announced plans to invest a total of 2,655 trillion won in Korea. SK hynix also unveiled 2,100 trillion won on the same day. Combined, the two companies’ domestic investment totals 4,755 trillion won. The figures were announced at the “National Report Meeting on the Three Major Mega Projects for a Great Leap Forward in Korea” held at the presidential office that day. The plan is based on making semiconductors, physical AI, and AI data centers the new three pillars of Korea’s industrial strategy.
The center of investment is moving to the regions
The notable point is where the money is going. Samsung will spend 2,030 trillion won on the semiconductor cluster in the capital region centered on Pyeongtaek and Yongin. The remaining 625 trillion won will be spread across Honam, Chungcheong, and Yeongnam.
Honam will receive the largest share. A total of 425 trillion won will go there, with 400 trillion won allocated to semiconductors alone. The core plan is to build a new semiconductor plant in Gwangju. Samsung Electronics Chairman Lee Jae-yong directly named Gwangju as a candidate site. He said, “We are planning Gwangju as a candidate site, where support for incentives such as securing power, water, labor, and infrastructure is expected.”
AI semiconductors and data centers are industries that consume massive amounts of electricity. They require huge power supplies, water, and large tracts of land. The capital region is already close to saturation. In contrast, Honam, Chungcheong, and Yeongnam still have available land, and renewable energy and water supplies are relatively abundant there. In other words, industrial location conditions and the policy goal of balanced national development have aligned.
Chairman Lee’s remarks also conveyed urgency. He said, “Following Giheung, Hwaseong, and Pyeongtaek, the investment schedule for the Yongin national industrial complex has accelerated significantly, and the time to prepare a new complex has also been brought forward.” This is interpreted as meaning that demand is rising so quickly that the next production base must be prepared earlier than planned.
What matters more than the numbers is what will be made
Samsung plans to expand its production capabilities for key AI memory, including HBM, mainly in the Chungcheong region. (Photo = Samsung Electronics)
Looking only at the scale of the investment can obscure the essence. What technology the money goes into matters more.
Chungcheong has been allocated 140 trillion won. The core item is high-bandwidth memory, or HBM. HBM is memory that stacks semiconductor chips vertically to enable much faster data transfer. It is an indispensable component for AI training and inference. Samsung plans to concentrate these HBM plants in Cheonan and Onyang. The strategy is to cluster production of a core component for the AI era in one region.
In Haenam, a different kind of investment will take place. Samsung SDS will lead the establishment of an infrastructure for sovereign AI. Sovereign AI refers to an independent AI ecosystem that a country develops and controls on its own rather than relying on foreign big tech. The data center to be built in Solaseado, Haenam, will serve as that hub. It will support the AI transformation of finance, defense, and public services. The idea is to keep AI sovereignty within Korea.
Yeongnam will take on robotics, batteries, and shipbuilding. In Gumi, a mass-production line for humanoid robots that resemble humans, along with a smartphone production base, will be established. In Ulsan, next-generation solid-state batteries will be developed, and in Geoje, Samsung Heavy Industries’ high-value-added shipbuilding base will be set up. It is a plan to spread a value chain running from semiconductors and AI to robotics, batteries, and shipbuilding across the country.
A distinctive feature of this plan is that it plants different industrial identities in each region. By breaking up and reallocating investments that were previously concentrated in one place, the intention is to rebuild regional industrial ecosystems.
Between blueprint and reality, power and speed will decide
There was not only applause. The figure of 4,755 trillion won itself raised skepticism. It far exceeds several times the national budget for a single year.
The presidential office was also aware of this. Kim Yong-beom, policy chief, predicted ahead of the announcement, “Because the scale and figures are so large, people will say, ‘Is this real?’” The opposition is questioning the appropriateness of the site selection in Honam and its feasibility.
The most realistic hurdle is power and water. Semiconductor plants and data centers consume as much electricity as an entire city. For new complexes to be built, power generation facilities, transmission networks, and water supply systems must all be in place. Workforce and living conditions are also essential. Building the factories alone is not enough.
The government has promised full support. President Lee Jae-myung bowed 90 degrees to Samsung Electronics Chairman Lee Jae-yong and SK Group Chairman Chey Tae-won, saying, “I would like to call you national heroes and heroes of the people.” He added that he would personally take responsibility for streamlined one-stop administrative procedures and said dedicated staff would be assigned to the presidential office. He also said power and water would be prioritized for the regions under the Special Semiconductor Act.
The market responded with optimism. On the day of the announcement, the Kosdaq rose by nearly 8 percent. In Gwangju, there were also signs of a property market mood expecting a “Pyeongtaek effect.”
In the end, success comes down to two things: whether the promised infrastructure can be installed on time, and whether Korea can keep pace with the speed of global AI competition. Chairman Lee said that day, “As the president said, it is a race for speed.” More than the size of the blueprint, the execution ability to turn that blueprint into reality is likely to determine future judgment.