The Rise of the Electric State in the Age of AI: Where Are the Voices of Affected Communities?
By Hyelyn Kim, Executive Director, ARC Center
There is little doubt that semiconductors dominated South Korea in the first half of 2026. Riding the boom in AI chips, SK hynix and Samsung Electronics posted record-breaking earnings, reversing the gloomy mood surrounding the Korean economy almost overnight. Their presence in the stock market was equally overwhelming. Together, the two companies accounted for more than half of the total market capitalization of the KOSPI, and some forecasts suggested that they could generate nearly 70 percent of the combined net profits of all KOSPI-listed companies this year.
Debate soon began over how the gains from the boom should be shared. SK hynix’s large performance bonuses triggered a backlash from the Samsung Electronics labor union, and a labor dispute that had escalated to the brink of a general strike was narrowly resolved through government mediation. Around the same time, the discussion expanded beyond the internal distribution of corporate profits to the broader question of social redistribution.
Kim Yong-beom, Chief of Staff for Policy at the Presidential Office, argued that “the gains of the AI infrastructure era are not the result of any single company, but have emerged from an industrial foundation built collectively by the Korean people over the past half-century.” He proposed the idea of a “National Dividend,” under which part of the additional tax revenue generated by the boom would be used to support youth entrepreneurship, rural and fishing communities, old-age security, and education for the AI transition.
However, the emerging discussion over the social distribution of profits changed direction abruptly when the government announced its Three Mega Projects for Korea’s Great Leap Forward. By presenting semiconductors, AI data centers, and physical AI as the country’s new engines of growth, the government effectively declared that what South Korea needed was not redistribution, but large-scale investment. The government described the “AI revolution” as the country’s last opportunity to overcome the twin crises of low growth and population decline. It pledged full-scale state support for corporate investment and industrial development.
The total planned investment amounts to KRW 1,600 trillion—more than twice the size of South Korea’s annual national budget. The government’s language in describing these enormous investment projects was strikingly combative. The Ministry of Trade, Industry and Energy described the international order as being reorganized around the principles of “survival of the fittest” and “every nation for itself.” It framed AI competition as a “war” on which the country’s future depends. Military expressions such as a “war of speed” and a “battle for strategic hubs” were attached to the semiconductor strategy, while the government proposed an “all-out support” system that would mobilize the full capacity of the state. The AI industry was no longer presented as an ordinary sector in which individual companies compete in the market. It was redefined as a battlefield on which the country must prevail for the sake of its national future.
At the center of this war is the idea of securing an early lead. Under the logic that victory depends on planting the flag first and securing production capacity ahead of competitors, speed becomes the most important source of competitiveness. Regulations and permitting procedures that delay projects are treated as obstacles to be removed. Yet no matter how much administrative procedures are streamlined, the physical time required to construct power plants and transmission networks and secure water supplies cannot simply be compressed. Ultimately, the scale and speed of the government’s planned expansion of the AI industry depend on electricity infrastructure.
The government does not hide this fact. Its announcement devoted a separate section to a “vision for an electric state leading the AI era.” Under the government’s plan, AI data centers with a combined capacity of 18.4 GW and a semiconductor cluster of approximately 6.3 GW in southwestern South Korea are expected to be developed by 2035. Together, they would create more than 24 GW of new electricity demand—equivalent to roughly one-quarter of the country’s current peak power demand. To meet this demand, the government plans to expand renewable energy while mobilizing nuclear power, small modular reactors, and fossil fuel generation. It also intends to expand the transmission grid and introduce a dedicated electricity tariff for AI data centers.
Electricity is no longer merely infrastructure that supports industry. In the age of AI, it is increasingly treated as a strategic resource that determines national competitiveness and must be directly secured and managed by the state. Electrification is, of course, an essential tool for responding to the climate crisis. Yet in the AI era, electrification is being reconfigured not only as a carbon-neutrality policy, but also as a national strategy for industrial competitiveness. The problem is that, in this process, the massive expansion of electricity demand is no longer treated as a subject for debate. It is increasingly accepted as a fixed and unquestionable assumption.
Countries around the world have already begun to treat AI infrastructure not merely as a matter of private investment, but as a public policy issue that must take energy use, environmental impacts, and community acceptance into account. Singapore halted approvals for new data centers for approximately three years and now allows only a limited number of projects that meet energy-efficiency and sustainability standards. In Ireland, after electricity demand from data centers exceeded 20 percent of national electricity consumption, the government introduced a new grid-connection policy requiring data centers to provide their own generation capacity and procure renewable energy. In the United States, New York State introduced the country’s first one-year moratorium on the construction of large data centers with a capacity of more than 50 MW.
Until now, the energy transition has focused primarily on replacing fossil fuels with electricity. Now that both climate action and AI industrial competitiveness have become national priorities, the transition toward an electric state has gained stronger political justification and momentum than ever before. The emerging question is no longer simply which fossil fuels should be replaced by electricity. It is how much expanding electricity demand society can and should bear.
Technology may become more efficient, and the share of renewable energy may continue to grow. But if total electricity demand grows even faster, much of the energy saved through efficiency improvements may be offset. The International Energy Agency projects that global electricity consumption by data centers will more than double by 2030. While renewable energy is expected to meet around half of the additional demand, more than 40 percent is projected to be supplied by natural gas and coal. This suggests that the expansion of the AI industry may make an increase in fossil fuel generation unavoidable, at least for a certain period.
Producing, transmitting, and storing more electricity requires additional power plants, transmission networks, and storage facilities. It also requires more minerals, land, water, and fuel. Inevitably, this creates new environmental and social problems in mining regions and in communities where power plants, transmission lines, and related infrastructure are located. Yet current AI industrial policy calculates the amount of electricity to be supplied in great detail while saying little about the resources required to produce that electricity or the communities that will be expected to bear the resulting harm.
An electric state requires more than simply producing more electricity. It requires social agreement on how far resources should be mobilized to generate that electricity, and which communities and ecosystems will be asked to pay the price.
This article was originally published in Pressian on July 16, 2026.