AI's Energy Appetite: How Data Centers Strain the Grid

Artificial intelligence runs on electricity, and the amount it needs is climbing fast enough to change how the entire power system is planned. Data centers are now one of the fastest-growing sources of electricity demand in the developed world — pulling tech companies into the energy business, reviving nuclear power, and straining grids that weren't built for this kind of load. Here's what's driving the surge and what it means for the grid.

How Big the Demand Really Is

The clearest numbers come from the IEA. Electricity consumption from data centers is projected to roughly double from about 485 TWh in 2025 to 950 TWh by 2030 — reaching around 3% of global electricity demand, comparable to the entire electricity consumption of Japan today — with consumption from AI-focused facilities specifically set to triple over the same period (IEA, Key Questions on Energy and AI; IEA, Energy and AI). The longer view runs higher: the IEA's base case sees data-center consumption climbing to roughly 1,200 TWh by 2035.

Demand from data centers already jumped 17% in a single year in 2025, well outpacing the roughly 3% growth in overall global electricity demand (IEA). Some forecasts run higher still — Goldman Sachs has estimated data-center electricity demand could rise 160% by 2030 (KAIZEN, citing Goldman Sachs).

The money behind that demand is moving just as fast. Capital expenditure by five large technology companies passed $400 billion in 2025 and is set to rise a further 75% in 2026, driven largely by data-center investment (IEA).

Why It Lands So Hard on the Grid

Two features make this demand especially disruptive. First, it's concentrated. The United States is home to the largest cluster of data centers and is seeing the fastest growth, with these facilities accounting for roughly half of the country's electricity-demand increase between now and 2030. By the end of the decade, the US is on track to use more electricity running data centers than producing aluminium, steel, cement, chemicals, and all its other energy-intensive goods combined (IEA, Energy and AI). Northern Virginia alone hosts a concentration drawing more than 5 GW (Global Electricity).

Second, it's inflexible. AI training and inference want stable, around-the-clock power, and grid connections have become a bottleneck — permitting delays and queue backlogs now constrain projects as much as generation does. AI workloads also cause large, rapid power swings, which makes storage increasingly important; the IEA estimates around 20–25 GW of battery storage may be needed to serve data centers reliably by 2030 (IEA, Key Questions on Energy and AI). There is some slack to exploit: on a typical day the US grid uses only about half the electricity it's capable of producing, so smarter scheduling, demand flexibility, and storage could absorb part of the load (Marketplace). Even so, the IEA expects 15–27 GW of onsite natural gas to be powering data centers by 2030, mostly in the US, as operators look for power they can secure quickly (IEA, Key Questions on Energy and AI).

The Nuclear Pull

The most striking second-order effect is what AI demand is doing to nuclear power. Tech companies want firm, carbon-free, 24/7 electricity, and that has made them a major source of momentum for advanced nuclear.

The pipeline of conditional offtake agreements between data-center operators and small modular reactor (SMR) projects grew from 25 GW at the end of 2024 to 45 GW by early 2026 (IEA). High-profile deals — Microsoft's agreement tied to restarting Three Mile Island, Google's contract for power from Kairos Power's SMRs — signal a structural shift, not a one-off (Global Electricity). The tech sector also accounted for roughly 40% of all corporate renewable power-purchase agreements signed in 2025.

The Efficiency Counterweight

It isn't purely a story of runaway consumption. The energy used per AI task is falling rapidly, with efficiency improving at a pace the IEA describes as unprecedented in energy history (IEA, Key Questions on Energy and AI). The catch is that more people are using AI, and energy-intensive applications like autonomous agents are growing — so total demand keeps rising even as each individual operation gets cheaper to run.

It's also worth keeping the scale in perspective: even at the projected 2030 level, data centers account for only about one-tenth of global electricity-demand growth this decade — less than the growth from industrial motors, air conditioning, or electric vehicles (IEA, Energy and AI). The strain is real but concentrated in specific grids, not uniform across the system.

And AI is increasingly turned back on the energy system itself, improving load forecasting, grid optimization, and the integration of variable renewables, while also accelerating research into core clean-energy technologies like batteries and solar PV. It cuts both ways on security, too: AI enables more sophisticated cyberattacks on utility infrastructure even as it strengthens the tools utilities use to defend against them.

The Infrastructure Underneath

Meeting this demand means building and maintaining a lot of physical plant — new reactors, gas capacity, and the cooling-water systems that large data centers and power stations both depend on. Those cooling intakes, tanks, and pipelines need regular inspection to run reliably, and much of that work happens underwater or in confined spaces where keeping systems online matters most. Remote inspection tools play a quiet but real role here, helping keep the power and cooling infrastructure behind AI running without costly downtime. For the broader picture of which sources will carry this load, see our guide to the energy sources of the future.

Table 3: AI's Energy Footprint — Projections & Implications

Category	Details
Current demand (2025)	Data centers consumed about 485 TWh of electricity — roughly 1.5% of global use1
2030 projection	Roughly doubles to ~950 TWh (~3% of global demand) — comparable to Japan's total electricity use today; rising to ~1,200 TWh by 20351,2
AI-specific growth	Electricity use by AI-focused data centers set to triple by 2030; overall data-center demand growing ~15% a year, more than four times faster than other sectors1,2
Geographic concentration	The US and China account for ~80% of global growth; US data centers drive nearly half of US demand growth, and by 2030 will use more electricity than all US energy-intensive manufacturing combined2
Meeting the demand	Renewables and natural gas lead the supply; ~15–27 GW of onsite gas and ~20–25 GW of battery storage may serve data centers by 2030, with the first SMRs online around 20301,2
Investment	Combined capital spending by the five largest tech firms passed $400 billion in 2025 and is set to rise ~75% in 2026, driven largely by data centers1
In perspective	Despite the surge, data centers account for only about one-tenth of global electricity-demand growth to 2030 — less than industrial motors, air conditioning, or electric vehicles2

Sources: ¹ IEA, Key Questions on Energy and AI (2026) — iea.org. ² IEA, Energy and AI special report (2025) — iea.org.

Frequently Asked Questions

Below, we answer common questions about future energy sources and technologies:

What Is a Small Modular Reactor (SMR)?

This is a new type of nuclear reactor designed to be smaller, safer, and more flexible than traditional nuclear plants. SMRs can be built in factories, transported to sites, and scaled to meet specific energy needs. They provide reliable, carbon-free baseload power for AI data centers, industrial processes, and water desalination.

What Is Battery Storage, and Why Is It important?

Battery storage is a large-scale system that stores electricity for later use. They're essential for balancing a grid with growing solar and wind, releasing stored power when the sun isn't shining or the wind isn't blowing, which makes variable renewables far more dependable.

What Is the Role of AI in Energy and Water Management?

AI will change the way we manage the energy sources of the future. AI optimizes grid operations, forecasts demand, and improves the integration of renewable energy sources. In water management, AI predicts consumption, detects leaks, and automates maintenance, helping to conserve resources and reduce costs. (More in our article on AI's Energy Appetite.)

What Kinds of Energy Will Be Used in the Future?

The future of energy will rely on a mix of sources to meet growing global demand, ensure reliability, and support sustainability goals. A diversified portfolio combines stable baseload power from nuclear and hydroelectric sources with rapidly growing clean energy sources like solar and wind, supported by battery storage for grid flexibility. Natural gas will serve as a bridge fuel during the transition, while geothermal and biomass provide localized solutions tailored to regional needs and industrial applications.

What Is Deep Trekker's Role in the Future of Energy and Water?

Deep Trekker provides ROVs and crawlers that inspect, maintain, and optimize the critical infrastructure behind energy and water systems — improving safety, reducing downtime, and delivering high-quality data across nuclear, hydroelectric, offshore wind, gas, and water applications.

Related Reading

Dig deeper into the forces shaping the future energy landscape:

• The US–China Energy and Clean-Tech Race: Who Powers the Next Decade — how the two largest economies are competing across solar, batteries, critical minerals, and fusion, and what it means for global energy security.
• Water Security and the Promise of Modern Desalination — why fresh water is becoming as strategic as energy, and how desalination is moving mainstream.
• Top 7 Most Promising Energy Sources of the Future — From nuclear and solar to offshore wind and hydro, see the 7 energy sources shaping 2030 — and the inspection work that keeps each one running reliably.

Sources

• IEA, Key Questions on Energy and AI (2026) — Executive Summary: https://www.iea.org/reports/key-questions-on-energy-and-ai/executive-summary
• IEA, data-centre electricity demand surge & SMR offtake pipeline (April 2026): https://www.iea.org/news/data-centre-electricity-use-surged-in-2025-even-with-tightening-bottlenecks-driving-a-scramble-for-solutions
• KAIZEN, Nuclear Energy Trends 2026 (citing Goldman Sachs & IEA): https://kaizen.com/insights/nuclear-energy-trends-2026/
• Global Electricity, Data Centers and AI Energy Consumption: https://www.globalelectricity.org/data-centers-energy-consumption/
• Marketplace, Why the U.S. is losing the AI energy race to China (Feb 2026): https://www.marketplace.org/story/2026/02/05/why-the-us-is-losing-the-ai-energy-race-to-china

Deep Trekker Resources for Further Exploration

Deep Trekker Nuclear Energy Solutions: Discover how Deep Trekker ROVs enhance safety and efficiency in nuclear power plant inspections.

Deep Trekker Hydroelectric Solutions: Learn how ROVs are transforming inspections and maintenance for hydroelectric dams.

Deep Trekker Offshore Wind, Oil, & Gas Solutions: Explore how ROVs provide efficient and safe inspection tools for offshore energy infrastructure.

Deep Trekker Clean Water Solutions: See how ROVs ensure the integrity and cleanliness of municipal water tanks and treatment facilities.

Deep Trekker Wastewater and Stormwater Solutions: Understand how pipe crawlers streamline inspections of critical sewer and stormwater networks.

Deep Trekker Underwater Vehicles (ROVs): Explore our range of robust, portable ROVs designed for diverse underwater missions.

Deep Trekker Pipe Crawlers: Learn about our battery-operated systems for efficient and safe pipeline inspections.

Deep Trekker Utility Crawlers: Discover our versatile crawlers for tank cleaning and inspection without draining or divers.

Deep Trekker BRIDGE Technology: Dive into the innovative software platform powering our ROVs with autonomous navigation and advanced data capture.

Case Study: Expanding Offshore Inspections with Deep Trekker ROVs: Read a real-world example of how Deep Trekker ROVs are transforming underwater inspections.

Photogrammetry and 3D Modeling: Learn how Deep Trekker ROVs enable detailed 3D modeling for comprehensive infrastructure assessment.

Request a Quote: Ready to explore Deep Trekker solutions for your operations? Contact us for a customized quote.