Introduction
Artificial intelligence has moved far beyond answering simple questions and now runs complex tasks non-stop inside huge data centers, making it one of the world’s fastest-growing users of electricity—already bigger than Bitcoin mining—and raising urgent questions about power demand, water use, costs for ordinary people, and how carefully this growth should be managed.
The Basic Numbers
In 2024, data centers worldwide used about 415 terawatt-hours (TWh) of electricity. That is roughly 1.5% of all the electricity used on Earth. By 2025 the figure rose to around 485 TWh.
Bitcoin mining uses far less. Most careful estimates put it between 140 and 200 TWh a year. Data centers already use about three times more power than Bitcoin, and AI is the main reason the number keeps climbing.
| Energy User | Estimated Annual Electricity Use | Key Trend |
|---|---|---|
| Global Data Centers, 2024 | About 415 TWh | Rapidly increasing |
| Global Data Centers, 2025 | About 485 TWh | Strong growth |
| Bitcoin Mining | About 140–200 TWh | More stable |
What Will Happen by 2030?
Experts at the International Energy Agency expect data centers to use nearly twice as much electricity by 2030—around 945 to 950 TWh. That would be about 3% of the world’s total electricity.
The part used only for AI could triple in the same time. Growth will be strongest in the United States and China.
AI Server and Data Center Energy Demand
- A single modern AI server rack can need as much power as dozens of homes.
- The biggest new AI centers need as much electricity as a small city.
- Bitcoin’s power use is more stable.
- Better mining machines and market limits keep its growth slower.
The Problems: Power, Water, and Local Costs
Power and Pollution
Bitcoin mining has improved. Roughly half or more of its electricity now comes from clean sources (renewables and nuclear). AI data centers often rely more on the local power grid, which can still use a lot of coal or gas. This means higher carbon emissions in many places.
Water
Data centers need huge amounts of water to cool the machines. When you add the water used at power stations, AI’s water use in 2025 was already estimated in the hundreds of billions of liters. This number is expected to rise further.
In dry areas the extra demand can hurt local water supplies.
Cost to Ordinary People
Big data centers can push up electricity prices for nearby homes and businesses. When power companies build new lines or plants mainly for these centers, the costs often fall on regular customers.
Local communities also face noise, extra traffic, and sometimes dirtier air if gas turbines are used.
| Issue | Potential Impact |
|---|---|
| Electricity Demand | Higher demand for power generation and transmission infrastructure |
| Carbon Emissions | Greater emissions where electricity comes from coal or gas |
| Water Use | Additional pressure on local water supplies, especially in dry areas |
| Electricity Costs | Potentially higher costs for nearby households and businesses |
| Local Impacts | Noise, traffic, and potentially increased air pollution |
Rules and Court Cases
Governments and courts are beginning to examine the environmental costs of large AI data centers more closely.
United Kingdom Data Centre Case
In the UK, groups such as Foxglove and the Global Action Plan challenged the government’s approval of a hyperscale data center in Buckinghamshire.
They argued that electricity demand, carbon emissions, and water use had not been properly assessed. The government later admitted a serious error, and the developer agreed that environmental protections must be made legally binding.
United States and the xAI Clean Air Act Case
In the United States, the NAACP sued xAI in 2026 under the Clean Air Act.
The case focused on dozens of gas turbines used to power an AI data center in Mississippi without the required clean-air permits. The plaintiffs claimed the turbines caused harmful pollution in nearby communities.
The Department of Justice stepped in and asked the court to dismiss the case, arguing the project was important for national security and AI innovation.
A Growing Legal and Environmental Tension
These cases highlight a growing tension: how quickly AI data centers should be built and how carefully their effects on the environment and local people must be controlled.
What Needs To Happen
AI’s electricity use is real and growing fast. It has already passed Bitcoin and is set to keep rising quickly until at least 2030.
The good news is that solutions exist:
- More efficient chips and software
- Better cooling that uses less water
- Building data centers next to clean power sources (solar, wind, nuclear, or storage)
- Clear rules that make companies report energy and water use
- Fair pricing so ordinary households do not pay extra for big tech projects
Managing AI Growth Responsibly
AI can bring big benefits, but only if its growth is managed carefully.
Policymakers, companies, and communities need to work together so that new technology does not come at the cost of clean air, water, or higher bills for ordinary people.
The shift from Bitcoin’s steady power use to AI’s rapid rise is a clear warning—and a chance to get the rules right before the problem grows even larger.
Conclusion
AI’s electricity use has already overtaken Bitcoin mining and is set to grow much faster, with data centers expected to nearly double their power demand by 2030 while the AI share alone could triple.
This rapid rise brings real risks of higher carbon emissions, heavy water use, and rising costs for ordinary households.
Yet practical solutions such as more efficient technology, cleaner power, better cooling, and fairer rules already exist.
Governments, companies, and communities must act together now to guide AI growth in a way that protects the environment and people’s daily lives instead of letting the problem become far larger.
