Why AI Data Centers Use So Much Electricity?

Key Takeaways

• AI workloads require far more computing power than traditional web services.
• Specialized AI chips and GPUs consume significantly more electricity than standard servers.
• Cooling systems can account for a substantial share of a data center’s energy use.
• Global data center electricity consumption is expected to more than double by 2030.
• AI is now the biggest factor driving growth in data center power demand.
• Energy infrastructure is becoming one of the biggest challenges facing the AI industry.

What Is an AI Data Center?

An AI data center is a facility filled with servers designed specifically to train and run artificial intelligence models.

These facilities contain thousands of specialized processors, primarily GPUs, that perform trillions of calculations every second.

Companies such as Microsoft, Google, Amazon, Meta, and OpenAI rely on these centers to power services like chatbots, image generators, recommendation systems, search engines, and AI assistants.

While traditional data centers mainly store data, host websites, and run business applications, AI data centers focus on intensive mathematical computations that require far more electricity.

How Much Electricity Do Data Centers Use Globally?

In 2024, data centers in the United States used around 183 terawatt-hours of electricity roughly 4 to 5 percent of everything the entire country used that year. For comparison, that is about the same as the annual electricity consumption of Pakistan, a country of 240 million people.

Globally, data centers used approximately 415 terawatt-hours in 2024, according to the International Energy Agency. By the end of this decade, that figure is expected to grow to around 945 terawatt-hours roughly equal to Japan’s entire electricity use today.

This growth is happening faster than almost any other industry on Earth. The International Energy Agency expects data center electricity use to grow at around 15 percent per year between 2024 and 2030. That is more than four times the growth rate of every other sector combined.

How Is an AI Data Center Different from a Regular One?

For most of the internet’s history, data centers were basically large warehouses full of standard computers. They stored files, handled emails, ran websites, and processed business software. Their electricity use grew steadily, but nothing alarming.

AI changed everything. Training and running large AI models requires completely different hardware. Standard computers use chips called CPUs, designed to handle a wide variety of tasks. AI needs GPUs chips originally built to render video game graphics.

Video game graphics require millions of simple calculations happening simultaneously, and that same ability is exactly what AI needs to learn from data.

The problem is that GPUs use far more electricity than standard chips. A typical server CPU uses around 150 to 250 watts. An AI GPU uses two to four times that amount. And AI data centers are packed with GPU racks from floor to ceiling.

Why Do AI Chips Use So Much Power?

NVIDIA’s H100 chip, the workhorse of the AI boom, draws up to 700 watts per unit roughly the same as a household iron running at full heat, every second of every day, without a break. NVIDIA’s newer B200 chip pushes that to around 1,000 watts per chip.

To understand what that means in practice, consider a single server cabinet. A cabinet holding eight H100 chips draws around 5,600 watts just for the processors. Add in the networking hardware, storage, and memory inside the same cabinet, and the total climbs to 10,000 or 11,000 watts. A full row of these cabinets draws 70,000 to 120,000 watts.

NVIDIA’s latest rack-scale system, the GB200 NVL72, packs 72 Blackwell chips into a single rack and requires 120,000 to 140,000 watts so much that air cooling cannot handle the density, requiring built-in liquid cooling.

NVIDIA’s upcoming Rubin Ultra NVL576 system, expected in 2027, is projected to draw up to 600,000 watts from a single rack: enough electricity to power a large apartment building.

These are not experimental prototypes. They are the commercial products going into data centers being built right now.

Most people only talk about the electricity AI uses when it is running in data centers. But there is a bigger part of the story that usually gets ignored. It is the energy needed to build the chips in the first place.

Take a powerful chip like NVIDIA’s H100. Making just one of these can use around 700 kilowatt-hours of electricity during manufacturing. Most of this work happens in huge factories run by TSMC in Taiwan, the company that produces most of the world’s advanced AI chips.

In 2023, TSMC used about 23 terawatt-hours of electricity to run its factories. That is about the same as the yearly electricity use of a medium sized European country. And this was before the recent AI boom fully increased demand.

Water use is another serious issue. Chip production needs extremely pure water in very large amounts. In 2022, TSMC used around 170 million tonnes of water. This is especially important because Taiwan has faced major droughts in recent years, which puts pressure on both industry and local water supplies.

This hidden side of AI is often called embodied energy. It means all the resources used to make something before it is ever turned on.

Since chip manufacturing happens in different countries and is counted separately from data centers, it usually does not appear in AI energy reports. But it is very real and it adds a large part to the true environmental cost of AI systems.

What Uses More Energy: Training AI or Running It?

Training is how an AI model gets built. Developers feed massive amounts of text into a system and adjust hundreds of billions of internal settings until the model learns to generate language properly. It takes an enormous amount of computing power.

Training GPT-3 used an estimated 1,287 megawatt-hours. That is roughly the same as the yearly electricity use of 120 American homes. Training GPT-4 used somewhere between 27,000 and 50,000 megawatt-hours which is enough to power a small town for an entire year.

Inference is what happens every time you actually use an AI. Every ChatGPT message you send, every AI image you generate and every AI-powered search result is inference. It runs continuously 24 hours a day.

Here is what most people miss. Training only happens once. Inference never stops and that is where the real electricity goes. Research cited by the US Department of Energy estimates that 80 to 90 percent of all AI computing is inference not training.

OpenAI’s CEO has stated that the average ChatGPT request uses about 0.34 watt-hours. That is significantly more than a standard Google search. ChatGPT handles an estimated 2.5 billion queries every single day. At 0.34 watt-hours per query that adds up to more than 310 gigawatt-hours per year from just one product at one company.

How Much Energy Does Cooling an AI Data Center Take?

All those chips generate enormous amounts of heat, and that heat has to go somewhere. Cooling is the single biggest non-computing energy cost in any AI data center in older or less efficient facilities, it can account for more than 30 percent of total electricity use.

Data center engineers measure efficiency with Power Usage Effectiveness, or PUE. A score of 1.0 means every watt goes directly to the computers. The global average is about 1.58, meaning nearly 40 cents of every electricity dollar goes to something other than computing.

Traditional air cooling is rapidly becoming inadequate. When rack densities sat at 5,000 to 10,000 watts, air was fine. At 50,000 to 120,000 watts per rack, air simply cannot remove heat fast enough.

Immersion cooling submerging entire servers in non-conductive liquid can reduce cooling energy costs by up to 90 percent. Immersion-cooled capacity equaled air cooling capacity in 2025 and is expected to double it by the end of 2026.

How Much Water Do AI Data Centers Consume?

Electricity is not the only resource AI data centers consume at scale. For every kilowatt-hour of energy used, a data center typically requires around 2 liters of water for cooling. Large facilities can consume up to 5 million gallons of water every single day.

A June 2026 United Nations University report found that global data centers consumed an estimated 448 terawatt-hours of electricity in 2025 more than Saudi Arabia uses in a year and that the total water footprint of AI data centers is projected to reach 9.3 trillion liters by 2030.

About two-thirds of data centers built since 2022 are located in areas already experiencing water stress, according to a Bloomberg investigation. A University of Houston study found that Texas data centers alone will use up to 399 billion gallons of water annually by 2030.

What Else Besides GPUs Drives the Energy Bill?

One of the most common misconceptions about AI energy use is that GPUs account for most of it. In reality, GPUs represent only about 40 percent of total power during peak operation. Research from Epoch AI breaks down where the rest goes:

For every watt an AI chip uses, roughly 2.4 watts must enter the building from the power grid. Data centers also cannot afford blackouts: a power cut during a multi-week AI training run can corrupt the entire process, costing millions of dollars.

Every facility therefore runs backup diesel generators and multiple redundant power delivery systems which consume electricity even on standby.

How Does Energy Use Scale Across a Full AI Facility?

Individual chip power figures become truly alarming when multiplied across a modern AI facility. A 50,000-GPU training cluster the kind used to train frontier AI models draws approximately 35 megawatts of power, comparable to a small city’s electricity demand.

Meta’s planned Hyperion project in Louisiana will need at least 5 gigawatts to run three times the electricity consumption of the entire city of New Orleans. Microsoft’s Fairwater data center currently in development will use more electricity than the city of Los Angeles.

In Northern Virginia, which holds the densest concentration of data centers on Earth, installed capacity has grown to roughly 4,000 megawatts, and new facilities face power grid connection wait times of five to ten years.

How are AI Data Centers Straining Local Power Grids?

In Ireland, data centers already consume more than 21 percent of the country’s total metered electricity more than all urban households combined. Ireland’s national grid operator has paused approvals for new data centers around Dublin until 2028.

In the United States, the PJM electricity market saw data centers add an estimated $9.3 billion in costs to the 2025–2026 capacity market, translating to average monthly bill increases of around $18 for residents in western Maryland and $16 in Ohio.

A Carnegie Mellon University study estimates that data centers and cryptocurrency mining combined could raise average US electricity bills by 8 percent by 2030, potentially exceeding 25 percent in northern Virginia.

What Energy Sources Power AI Data Centers?

Big Tech companies have collectively become the largest corporate buyers of renewable energy on Earth, signing 43 percent of all clean energy purchase agreements globally in 2024. But renewable energy has a fundamental limitation: solar panels do not generate electricity at night, and wind turbines only work when the wind blows. Data centers need power 24 hours a day, seven days a week.

This gap has pushed tech companies toward nuclear power. In September 2024, Microsoft signed a 20-year agreement with Constellation Energy to restart Unit 1 of the Three Mile Island nuclear plant in Pennsylvania.

Google has signed a deal with Kairos Power to build multiple small modular reactors. Amazon has committed to backing 5 gigawatts of new small modular reactor projects.

Meta has issued a formal request for 1 to 4 gigawatts of new nuclear generation. Oracle has announced plans for a data center powered by three small modular reactors.

Are Big Tech Climate Promises Real?

Tech giants like Google, Microsoft, and Amazon have promised to run on 100% clean energy or reach net-zero emissions by 2030. While those goals are real, the situation is more complicated than it sounds.

Most companies do not power their data centers with clean energy every hour of every day. Instead, they often buy renewable energy credits from solar or wind projects in other locations and count that energy toward their climate goals. This means their data centers may still use electricity from gas or coal at certain times.

The numbers show how difficult the challenge has become. Microsoft’s emissions increased by about 30% between 2020 and 2023, while Google’s emissions rose by about 48% during the same period. Both companies say the rapid growth of AI and the expansion of data centers are major reasons for the increase.

Google is taking a stricter approach called 24/7 clean energy matching. The goal is to use clean electricity at the same time and in the same location where it is consumed. By 2024, Google reported reaching about 64% of that goal. Most other major tech companies have not yet adopted this standard.

The climate commitments are real, but progress is proving more difficult than expected. As AI continues to grow, meeting those targets will become an even bigger challenge.

Are AI Data Centers Becoming More Efficient?

Real progress is happening. Google reported in 2025 that between May 2024 and May 2025, it reduced the median energy use per Gemini prompt by a factor of 33, and cut the associated carbon footprint by a factor of 44.

Smaller, more efficient model designs are also emerging: a new approach called Mixture of Experts activates only a fraction of a model’s total processing power for any given question. Research from Epoch AI estimates that leading AI hardware has become roughly 40 percent more energy-efficient each year.

DeepSeek Made AI More Efficient — So Why Is Energy Still Rising?

In 2025, DeepSeek released an AI model that matched top U.S. models but reportedly cost only about $6 million to train. Similar models from companies like OpenAI can cost hundreds of millions. So people naturally asked: if AI is becoming cheaper and more efficient, why isn’t electricity use going down?

The answer is simple: efficiency usually leads to more usage, not less.

When something becomes cheaper, people use it more. Companies don’t slow down spending. They scale up. They add more AI tools, build more products, and run more systems. So total AI usage keeps growing.

The International Energy Agency (IEA) says that even with more efficient AI, data center electricity use is still expected to rise sharply, reaching around 950 terawatt-hours by 2030. Cheaper AI doesn’t reduce energy use. It increases demand for it.

This is known as the Jevons paradox. It means that when technology becomes more efficient, total consumption can actually go up instead of down.

A Comparison you Probably Have Not Seen