Power Consumption to Reach 30 TWh by 2038 with AI Proliferation
Power Grid Saturation in the Capital; Communication and Talent Shortages in the Regions
Overseas: Leveraging Pricing, Regulation, and Incentives
A 'Push-and-Pull' Strategy Needed to Make Regional Relocation Profitable for Businesses

An AI data center simulation using Samsung SDI's high-energy prismatic battery is exhibited at 'InterBattery 2026' held at COEX in Gangnam-gu, Seoul. 2026.3.11 Photo by Kang Jinhyung

An AI data center simulation using Samsung SDI's high-energy prismatic battery is exhibited at 'InterBattery 2026' held at COEX in Gangnam-gu, Seoul. 2026.3.11 Photo by Kang Jinhyung

View original image

As the era of artificial intelligence (AI) unfolds, data centers are emerging as key national infrastructure. However, South Korea faces a pronounced structural challenge—the concentration in the greater Seoul area. While the Seoul metropolitan area boasts optimal operating conditions such as robust communication networks and abundant talent, its power grid is already saturated. In contrast, non-metropolitan regions have ample power and land but lack sufficient communication networks and residential amenities, making it difficult to attract businesses. With competition over AI data center construction intensifying, there is growing consensus that the government must pursue both regulatory measures in the capital and infrastructure enhancement in the regions.


The Seoul Metropolitan Area Is Best for Operation; The Regions Offer the Best Construction Sites

According to a recent report by the Korea Electric Power Corporation Management Research Institute titled "Status and Implications of Overseas Decentralization Policies for Data Centers," published on July 28, the advantages and disadvantages of domestic data center locations differ sharply between the capital and the regions.


The metropolitan area has a high concentration of cloud and platform companies as well as skilled professionals, and is equipped with strong communication infrastructure like internet exchange (IX) points, making it the most advantageous environment for operating data centers.


However, conditions for building new data centers are worsening. As the power system reaches its limits, access to new power connections is being restricted, with further setbacks including project delays or cancellations due to community opposition over concerns about electromagnetic waves and landscape disruption.


By contrast, outside the metropolitan area, securing power, water, and land is relatively straightforward and land prices are lower. Conditions for utilizing renewable energy and community acceptance are also better than in the capital region.


Nonetheless, communication networks and living conditions remain lacking in the regions. In areas with few internet exchange points, dedicated long-distance networks must be used, resulting in higher communication costs. Additionally, securing IT experts necessary for AI data center operations is not easy.


As a result, analysis suggests that operating environments are optimal in the metropolitan area but construction is difficult, while in the non-metropolitan regions, construction is facilitated but competitive operation is challenging. This structural divide is becoming entrenched.

Power in the Regions, Infrastructure in the Capital: "AI Data Center Dilemma Calls for a 'Push and Pull' Solution" View original image

In the AI Era... Power Demand Set to Quadruple

The spread of AI data centers is expected to further exacerbate this dilemma.


Domestic data center power consumption is projected to rise from 8.2 terawatt-hours (TWh) in 2025 to 30 TWh in 2038—nearly a fourfold increase. The government is also pushing to build more than 8.4 gigawatts (GW) of AI data centers by 2029 and at least 15 GW by 2035.


If data centers continue to concentrate in the greater Seoul area, the resulting burden on the power grid and the cost of system reinforcement will inevitably grow. Meanwhile, the regions, despite having sufficient power and land, may continue to face difficulties in attracting investment due to a lack of communication networks and skilled labor.

Power in the Regions, Infrastructure in the Capital: "AI Data Center Dilemma Calls for a 'Push and Pull' Solution" View original image

"Regulation in the Capital, Support for the Regions"

Residents of a local government in Gyeonggi are holding a protest against the construction of a data center. Photo by The Asia Business Daily

Residents of a local government in Gyeonggi are holding a protest against the construction of a data center. Photo by The Asia Business Daily

View original image

The report suggests implementing a "Push & Pull" strategy, simultaneously pursuing regulation in the metropolitan area and support for the regions.


In the capital region, it proposes raising the location cost for companies by having them bear the actual expenses for network reinforcement and community compensation, while in the regions, the government should proactively enhance communication infrastructure and residential environments.


It also emphasizes the need to connect advantages in the regions—such as affordable land, abundant power, and renewable energy—with direct cost savings for companies, thereby encouraging voluntary relocation.


Other major countries are pursuing similar policies. While regional areas receive infrastructure, tax, and financial support, areas with saturated power grids are subject to system reinforcement costs, self-generation obligations, or have local power conditions reflected in electricity rates. This approach is meant to promote the decentralization of data centers.


However, the report notes that special electricity rates exclusively for data centers should be approached with caution, as other sectors—such as small businesses, SMEs, railways, and key manufacturing industries—might make similar requests for preferential treatment.



Hyunwoo Jung, Senior Researcher at the Korea Electric Power Corporation Management Research Institute, stated: "Policies must be designed so that they function as regulations (Push) in the metropolitan area and as incentives (Pull) in the regions. An asymmetric structure, where the cost burden rises for staying in the capital and the benefits grow for relocating to the regions, is necessary to change companies' location decisions."

KRW 550 Trillion AI Data Center Investment... Decentralization Strategy Put to the Test
Last month, the government announced at the "National Briefing on Korea’s Three Major Mega Projects for a Leap Forward" that it plans to invest a total of KRW 550 trillion by 2029 to build AI data centers (AIDC) with a total capacity of 8.4 GW. SK Telecom (5 GW), GS (2.4 GW), and Naver (1 GW) will be responsible for constructing hyperscale data centers, while the government will support power and land acquisition and licensing procedures.

Private sector investment is also gaining traction.

Naver has secured a total investment and infrastructure worth USD 10 billion (approximately KRW 14.7 trillion) from Nvidia and Brookfield to pursue its global AI factory business, and it plans to expand its data center to 200 MW by 2028 with "Kak Sejong" as its hub.

SK Telecom is also working with Nvidia to build AI data centers with a capacity of up to 2 GW and is discussing AI data center investment cooperation at the gigawatt scale with Anthropic. The company also plans to expand its AI computing infrastructure nationwide using the Ulsan AI data center that it is building with Amazon Web Services (AWS).

Hyundai Motor Group is investing about KRW 9 trillion in Saemangeum, North Jeolla Province, to create the "Saemangeum AI Valley," which combines an AI data center, a robotics manufacturing and component cluster, a water electrolysis plant, and solar power generation. Of this, KRW 5.8 trillion is being invested solely in the AI data center.

As the government and private sector ramp up large-scale investment, AI data center location strategy is emerging as a core issue determining national competitiveness. With energy resources concentrated in the regions and AI demand and digital infrastructure clustered in the metropolitan area, policy design to ease over-concentration in the capital and enhance regional competitiveness has become a new imperative in the AI era.


This content was produced with the assistance of AI translation services.

© The Asia Business Daily. All rights reserved. Unauthorized AI training and use prohibited.

Today’s Briefing