Table of Contents
Han Dong-hoon, Busan Buk-gu Gap: A Record of the 100 Days Before and After the Election (Mar. 26-Jul. 3, 2026)
Kim Kyung-jin
Table of Contents and 13 sections
From March 26 to July 3, 2026, this record follows the spring after expulsion, the Busan Buk-gu Gap by-election, victory as an independent, and the first bill submitted in the National Assembly.
Table of Contents
Artificial Intelligence and Medicine
Kim Kyung-jin, Attorney at Law
AI in clinical care, hospitals, education, and research
AI in medical imaging, risk prediction, treatment planning, hospital operations, education, and research, with patient safety, privacy, and accountability.
[AI Library] Chapter 7. Electricity, Carbon, and Global Warming
Ten Questions AI Poses to Humanity
Chapter 7. Electricity, Carbon, and Global Warming
Kim Kyung-jin
"The empire of civilization is built on kilowatts." - Winston Churchill
On the empty plains of Abilene, Texas, a massive building glows bright even at midnight. It is the first data center of OpenAI's Stargate Project. This single building consumes 1.2 gigawatts of power, enough for 750,000 households. And this is only the beginning. OpenAI announced it would invest $500 billion over the next four years to build a total of 10 gigawatts of AI infrastructure. That is roughly equal to the entire electricity consumption of the Netherlands.
1. Artificial Intelligence Runs on Electricity
Just as humans eat food to sustain life, artificial intelligence feeds on electricity to compute, write, and reason. But AI's appetite is beyond imagination. A single Google search requires 0.3 watt-hours of power. Ask ChatGPT the same question, and it consumes 2.9 watt-hours. Nearly a tenfold difference.
This is exactly why OpenAI's Sam Altman asked users to stop saying "thank you" to the AI. Every time a user types "thanks," the AI burns another round of electricity generating a response. Across a service used by hundreds of millions of people every day, these small differences add up to astronomical power consumption.
As AI journalist Karen Hao put it sharply, "The AI empire is entirely the same as a real-world empire that greedily consumes resources." We have been so absorbed in the convenience of the digital world that we forgot the physical cost behind it. The underside of a Tesla self-driving car is packed solid with batteries. Behind AI's convenience hides a similarly enormous consumption of electricity.
2. The Staggering Power Consumption of AI Data Centers
The latest report from the International Energy Agency (IEA) presents shocking figures. Global data center electricity consumption is projected to more than double, from 415 terawatt-hours in 2024 to 945 terawatt-hours by 2030. That is roughly equal to Japan's entire current electricity usage.
What is even more striking is that AI is the primary driver of this increase. Power demand from AI-optimized data centers is expected to quadruple by 2030. MIT researchers reported that North American data center power requirements nearly doubled in a single year, from 2,688 megawatts at the end of 2022 to 5,341 megawatts at the end of 2023.
The scale of the Stargate Project illustrates this trend starkly. Running two million AI chips requires 12.2 billion kilowatt-hours of electricity per year. At average U.S. electricity rates, the annual power bill alone comes to $1.2 billion. But actual power consumption would be even higher. Beyond the chips themselves, servers, cooling systems, UPS units, and communications equipment can push total power consumption to twice what the chips alone require.
The cooling problem is especially severe. Rack power density in AI data centers is surging from the traditional 10-15 kilowatts to 40-250 kilowatts. That is far beyond what conventional air cooling can handle. Liquid cooling technology has become essential, and even that consumes massive amounts of power. Cooling accounts for roughly 40% of a data center's total electricity consumption.
3. America Is Burning Coal Again
There is a painful irony here: AI, a technology of the future, is reviving coal, an energy source of the past. The United States is facing a serious power shortage right now. According to the North American Electric Reliability Corporation (NERC), more than half the states in the country face a growing risk of power supply shortfalls over the next decade.
In Georgia, the surge in industrial power demand from AI data center construction forced planners to revise their ten-year power needs estimate upward by a factor of 17. In northern Virginia, supplying electricity to new data centers would require several additional large nuclear power plants.
Facing this urgent situation, the United States is choosing power supply over environmental concerns. In Kansas, Nebraska, Wisconsin, and South Carolina, coal-fired power plants that were scheduled for retirement are being kept running. This collides head-on with the generational challenge of addressing climate change.
Goldman Sachs estimated that roughly $50 billion (about 69 trillion Korean won) in investment is needed between 2022 and 2030 to meet the growth in U.S. AI-related power demand. The problem is that renewable energy plants cannot be built fast enough to keep up. Solar, wind, and other renewables still cost more than fossil fuels to generate, and their output fluctuates with the weather.
4. By 2027, Power Shortages Could Shut Down AI
Gartner issued a startling prediction: by 2027, 40% of existing AI data centers will face power availability problems. In other words, there may not be enough electricity for AI to function properly.
At the heart of the problem is an aging transmission grid. The grid is the highway for electricity, carrying power from generating plants to our homes, offices, and data centers. More than 2 terawatts of renewable energy projects in the United States are currently waiting for grid connection permits. Historically, only 15 out of every 100 renewable energy projects have been approved for grid connection.
The connection costs are staggering. At PJM, the power management company for the mid-Atlantic region of the United States, a solar power plant must pay $100,000 per megawatt, roughly 130 million Korean won, to connect to the grid. Building new transmission lines typically takes more than ten years in total: several years for planning, several more for government and community permits, and several more for actual construction.
In this environment, companies are choosing data center locations based on power availability first. They are leaving traditional tech hubs like Silicon Valley and Seattle and looking as far as the cornfields of the American Midwest. This is rapidly changing land use patterns in rural areas and having a major impact on the lives of local residents.
5. Accelerating Climate Change and Soaring Carbon Emissions
The spread of AI is driving a sharp increase in carbon emissions. The IEA estimates that the global data center industry emitted approximately 180 million tons of greenhouse gases per year as of 2024. According to a Morgan Stanley report, greenhouse gas emissions from data centers are projected to reach 2.5 billion tons per year by 2030.
The reality at major corporations is even more sobering. Google alone emitted 14.3 million tons of greenhouse gases in 2023, a 13% increase from the previous year. Compared to 2019, four years earlier, that is a rise of nearly 50%. Microsoft also reported that its carbon emissions increased by 29.1% since 2020.
What makes this worse is that the figures companies officially report are likely far lower than the actual numbers. According to an investigation by The Guardian, greenhouse gases emitted from data centers owned by Google, Microsoft, Meta, and Apple between 2020 and 2022 are estimated to be 7.62 times higher than the officially reported figures.
Against this backdrop, corporate carbon neutrality pledges are becoming hollow promises. In July, Google gave up its title as "the first company to achieve carbon neutrality." Advances in AI technology and data center expansion have kept pushing carbon emissions higher. With the entire world targeting carbon neutrality by 2050, the rise in greenhouse gas emissions driven by AI is on a direct collision course with climate action.
6. Cooling Water and the Depletion of Water Resources
AI's thirst for water is as serious as its hunger for power. On average, a 100-megawatt data center consumes 2 million liters of water per day. That is equivalent to the water used by 6,500 households. Globally, data centers consume 560 billion liters of water per year, enough to fill 224,000 Olympic swimming pools.
Google used 6.1 billion gallons (roughly 23.1 billion liters) of water in 2023 alone. That is about as much as a small city consumes in a year. What makes this worse is that 80 percent of the water data centers use evaporates and never returns to its original source.
According to a Bloomberg analysis, nearly two-thirds of the new AI data centers built or under development in the United States since 2022 are located in areas already suffering from water shortages. More than 160 new AI data centers are going up in water-scarce regions, putting even greater pressure on local water supplies.
In Arizona, farmers are abandoning their fields and households are losing tap water service, yet data centers keep drawing massive volumes of water. In The Dalles, Oregon, a Google data center accounts for 25 percent of the city's total water use, and its consumption nearly tripled between 2017 and 2022.
The water shortage problem is especially severe in desert regions. Arid countries like Saudi Arabia and the UAE are courting more data centers to capitalize on the AI boom, but doing so only deepens the strain on water resources that are already scarce.
7. The Price of Unsustainable Growth
An OpenAI employee's remark, 'We are running out of land and power,' captures the severity of the situation in a single sentence. With the AI sector projected to consume 85 to 134 terawatt-hours by 2027, roughly equal to the entire electricity consumption of the Netherlands, we need to ask ourselves seriously whether it is acceptable to keep developing AI at this pace.
The biggest problem is that AI development relies almost entirely on market logic. Companies race to build bigger and more powerful AI models to beat their competitors, and environmental and social costs get ignored along the way. The benefits of AI development concentrate in a handful of giant corporations, while the damage from environmental destruction and energy shortages falls on all of humanity. The structure is fundamentally unfair.
As Alex de Vries, a data scientist at the Dutch central bank, has pointed out, 'AI is energy-intensive, so deploying it for all sorts of tasks where it isn't actually needed is a pointless waste.' We have become so addicted to AI's convenience that we can no longer tell the difference between where it is truly needed and where it is not.
AI development needs regulation and guidelines. Using AI where it is genuinely needed matters more than spreading it indiscriminately. AI companies must be held to stronger environmental accountability, with stricter environmental standards and carbon emission rules. And far more investment should go into developing technologies that improve energy efficiency.
There are things individuals can do as well. Before using an AI service, pause and ask whether you really need it; cut back on unnecessary AI use. Choose AI companies that take environmental responsibility seriously. Speak up and demand stricter environmental standards from governments and corporations.
We stand at a crossroads. We must decide whether to destroy the environment and pass the burden on to future generations for the sake of AI's convenience, or to develop AI in a sustainable way. Artificial intelligence is a powerful technology that can reshape humanity's future. But its power is precisely what demands caution in how we use it.
The choices we make now will determine the fate of future generations. We can enjoy AI's convenience, but we must understand exactly what it costs and steer its development in a sustainable direction. That is a shared responsibility. Finding the balance between technological progress and environmental protection, between economic growth and social fairness: it is difficult, but it is a challenge we cannot afford to leave unsolved. If we want to hand future generations a healthy planet and a viable society, the time to start changing is now.
Kim Kyung-jin
Attorney · Former Member of the National Assembly · AI Policy Researcher
© 2026 Kim Kyung-jin. All rights reserved.