The bottleneck choking artificial intelligence has quietly shifted from computer chips to electricity. While NVIDIA solved the GPU shortage that once threatened AI's expansion, no amount of computing power matters if data centers cannot find enough power to run. Global data center electricity consumption reached approximately 415 terawatt-hours (TWh) in 2024, but the International Energy Agency (IEA) projects this could skyrocket to over 1,700 TWh by 2035 in an aggressive growth scenario, representing 4.4 percent of all global electricity production.

The scale of this challenge is staggering. A single training run for a large language model consumes around 50 gigawatt-hours (GWh), equivalent to the annual electricity use of 40,000 American households. Former Google CEO Eric Schmidt testified before U.S. authorities that data centers will need 29 gigawatts (GW) of new power capacity by 2027 and a further 67 GW by 2030, more than six times New York City's total peak electricity demand.

Why Is AI's Energy Hunger Growing So Rapidly?

Artificial intelligence servers are the fastest-growing component of data center electricity demand. AI workloads are expected to grow by 30 percent annually and will account for nearly half of the net growth in data center consumption between 2024 and 2030. As AI models expand from text to images and video, their energy requirements grow proportionally. A single ChatGPT query consumes roughly 10 times the energy of a Google search.

The problem is not just about raw electricity demand. It is about the timeline mismatch between when hyperscalers need power and when the grid can deliver it. The chip shortage that made NVIDIA worth $4 trillion was an 18 to 24 month manufacturing problem. By contrast, new nuclear plants take 10 to 15 years from approval to operation, and new transmission lines require 8 to 12 years to permit and build. These timelines do not compress, regardless of how much capital gets thrown at them.

How Are Tech Giants Securing Power for AI Data Centers?

• Nuclear Partnerships: Microsoft signed a 20-year deal to restart the Three Mile Island nuclear plant, offline since 2019, specifically to power its AI ambitions. Amazon paid $650 million for a data center campus directly co-located with the Susquehanna nuclear station in Pennsylvania. Google announced agreements with Kairos Power for small modular reactors, and Meta issued a request for proposals seeking up to 4 gigawatts of new nuclear capacity.
• Renewable Energy Commitments: The largest technology companies, Google, Microsoft, Amazon, and Meta, accounted for 43 percent of all global power purchase agreements in 2024. Microsoft committed to 10.5 GW of renewable energy by 2030, and Amazon is targeting 100 percent renewable energy by 2025.
• Strategic Power Acquisition: Companies like Bitzero Holdings secured over 1 gigawatt of low-cost power capacity across strategic sites in Norway, Finland, and the United States before the AI boom accelerated. Bitzero's Norwegian facility operates at 3 to 4 cents per kilowatt-hour, compared to the U.S. average of 12 cents, giving secured operators a dramatic cost advantage.

The reality, however, is more complex than corporate green pledges suggest. Renewable energy currently covers only around 27 percent of global data center electricity. Natural gas accounted for 40 percent in 2024, and coal for 15 percent. Through 2030, fossil energy is still expected to cover more than 40 percent of new demand growth, meaning the AI boom will drive continued reliance on carbon-intensive power sources despite hyperscalers' renewable commitments.

What Other Infrastructure Challenges Are Data Centers Facing?

Electricity is only part of the problem. Cooling systems account for between 20 and 40 percent of a data center's electricity consumption, and the share is even higher in AI facilities. Liquid cooling, including direct chip cooling and immersion in coolant, is becoming the standard for heavy AI workloads and can reduce emissions by 21 percent compared to air cooling.

Water consumption presents another rarely discussed concern. By 2030, data centers' combined water use could reach 9,300 billion liters, enough to cover the basic drinking water needs of Earth's 8.1 billion inhabitants for 1.6 years. As AI moves from text to image and video, models grow larger and their environmental footprint expands accordingly.

"As AI moves from text to image and video, models grow larger, and their energy footprint grows with them," noted Vijay Gadepally, senior researcher at MIT Lincoln Laboratory. "The next two to three years will likely bring increased emissions, and nuclear power may be necessary to achieve the scale of clean energy the AI sector requires."

Vijay Gadepally, Senior Researcher at MIT Lincoln Laboratory

What Do Industry Projections Reveal About Future Power Demand?

Estimates of future power demand vary considerably, reflecting uncertainty about AI adoption rates and efficiency improvements. Goldman Sachs projects global data center power demand will surge up to 165 percent by 2030 compared to 2023 levels. Deloitte estimates that AI data centers' power needs in the U.S. could grow thirtyfold by 2035. In the United States specifically, Lawrence Berkeley National Laboratory estimated that American data centers will account for between 6.7 and 12 percent of the country's total electricity consumption by 2028, up from 4.4 percent in 2023.

Industry forecasts now put AI data center capital expenditure at roughly $5.2 trillion between now and 2030. For commodity markets, the message is clear: energy demand from data centers is a structural and growing driver, and fossil fuels will play a significant role well into the next decade, regardless of companies' green ambitions.

The companies that secured power capacity before the AI surge accelerated are now sitting on assets that cannot be easily replicated. Norway has effectively closed the door on new entrants, capping new operators at just 5 megawatts of initial allocation, and Finland and Sweden are tightening their capacity allocations as well. This scarcity is reshaping the competitive landscape of AI infrastructure, where access to cheap, reliable power has become as critical as access to the latest GPU chips.