AI Data Centers to Consume 1,000 TWh by 2030

Global electricity use by data centers could double by 2030, reaching roughly 1,000 terawatt-hours a year, as the AI boom adds unprecedented pressure to power grids, the International Energy Agency warned in its annual report.

Artificial intelligence accounts for about three-quarters of this growth, the IEA said.

Within four years, AI data centers alone will consume an extra 350 TWh of electricity every year, more than the entire annual electricity consumption of Great Britain, according to the Paris-based agency.

Data center capacity has climbed tenfold in the last 24 months to 17 GW for AI-specific facilities, with a further 120 GW of combined data center capacity expected by the end of the decade under the IEA's central scenario.

The IEA based its forecasts on satellite imagery tracking 93 large "AI factories" around the world and data showing annual global shipments of AI servers have climbed eighteen-fold since 2020.

The agency described the data center buildout as unprecedented in speed and scale, comparing it to the early buildout of global telecommunications infrastructure.

The world's existing stock of data centers, including those with AI-optimized servers, now stands at 120 GW, with 80% of this capacity in China, the US, and the EU.

Energy Mix by Region: US Turns to Gas, China to Coal

The agency found stark regional differences in how data center developers plan to power their expansion. The US is expected to rely on gas to provide about 60% of the electricity for its data centers by 2030.

Developers had plans for roughly 180 GW of new gas-fired power plants linked to US data centers as of August this year, almost as much new gas capacity as the total now installed across the European Union.

In China, about half of the electricity needed for AI data centers will come from coal power plants by the end of the decade, with the rest from renewable energy, nuclear reactors, and a very small amount from gas.

This dual reliance on fossil fuels in the world's two largest economies means that the AI boom is having a measurable impact on global carbon trajectories, despite pledges by major tech companies to run on 100% renewable energy.

The appetite for gas power among US data center developers is more than 16 times higher than in Canada, where plans for 11 GW of new gas generation exist to power AI servers.

In the rest of the world, AI developers have put forward plans for new gas plants with a combined capacity of just 4 GW, the agency said.

The European Union is taking a different path. The IEA expects EU data centers to get about 70% of their electricity from renewables by 2030, followed by nuclear power, gas generation, and a very small amount of coal.

This difference reflects both stricter EU climate regulations and the availability of established renewable energy infrastructure in member states.

Emissions on Track to Rise 70%

As a result of fossil fuel reliance in the US and China, emissions associated with the data center sector are on track to rise 70% by 2030, the IEA said.

Data center developers have emerged as the world's biggest corporate buyers of renewable energy, but the report showed that the appetite for fossil fuels is growing even faster, particularly in the two largest economies.

"I don't see any credible pathway for the energy world to return to the way it was before," IEA Executive Director Fatih Birol said of the broader energy landscape shaped by the Middle East war and surging AI demand.

The agency described the combined energy and geopolitical crisis as the greatest energy market disruption in history.

Before the Iran war disrupted shipping, an average of 20 million barrels of crude oil and oil products a day transited the Strait of Hormuz in 2025, about 25% of global seaborne oil trade.

The strait also carried nearly 20% of global LNG trade.

Grid Capacity Under Strain

Peak electricity demand is projected to rise 40 to 50 percent by 2035, the IEA said.

Under its Stated Policies Scenario, grids would need to be expanded by 26 million kilometers of new lines by 2035, while about 16 million kilometers of existing lines need replacement or refurbishment.

Supplies of copper, cables, and transformers are already tight, and the agency warned that supply chain bottlenecks could slow the buildout of both grid infrastructure and the data centers themselves.

Grid spending is projected to rise 75% over the next decade under existing policies, which the IEA called "one of the biggest energy investment stories of the next decade."

Renewable power generation is expected to grow 85% more in the next decade than in the previous one, with solar photovoltaics becoming the world's largest source of electricity from the late 2030s.

Electricity would account for nearly 30% of global final energy consumption by 2035, up from 23% today.

Without significant upgrades, new data center projects face longer wait times to connect, and grids could become less reliable over time. In several regions, the combined capacity of data centers waiting to connect already exceeds current peak demand.

The IEA warned that tight supply chains, local opposition, and limited grid capacity could combine to slow construction of new data centers substantially.

"The pressures of the current crisis must not obscure the choices that will shape energy security, sustainability and prosperity for the decades to come," Birol said, urging governments to plan grid upgrades alongside AI infrastructure investments.

What This Means for AI Development

The IEA report underscores a tension at the heart of the AI industry: the same technology that promises to solve complex problems is itself a major driver of energy demand and carbon emissions.

For AI companies racing to build larger models and deploy more capable systems, access to reliable and affordable power is becoming a strategic bottleneck.

The agency's data suggest that data center capacity has grown tenfold in two years, a rate that far outpaces the grid infrastructure needed to support it.

Tech companies including Google, Microsoft, Amazon, and Anthropic have made public commitments to run on carbon-free energy, and data center developers have become the largest corporate buyers of renewable energy globally.

But the IEA's report shows that in practice, the scale of the buildout is pushing developers toward whatever power is available and affordable, often gas or coal.

The gap between corporate renewable pledges and the actual energy mix powering AI infrastructure is likely to become a growing point of scrutiny for investors and regulators alike.

Common questions

How much electricity will data centers use by 2030?

Data centers could consume roughly 1,000 terawatt-hours of electricity a year by 2030, double today's levels, according to the IEA.

What is driving data center electricity demand?

AI accounts for about three-quarters of the growth in data center electricity demand, with the IEA tracking 93 large AI factories under construction around the world.

How will data center growth affect carbon emissions?

The IEA projects carbon emissions from data centers will rise 70% by 2030, driven by US reliance on gas and Chinese reliance on coal to power the expansion.

Sources & further reading

  1. Data center electricity use could double by 2030, IEA says ↗
  2. Datacentres to create 70% more carbon emissions by 2030, IEA report warns ↗
How this story was made

Written by AI Lookout with AI assistance and checked against the linked sources. Company performance claims are attributed to the company. Analysis reflects AiLookout’s interpretation; we have not independently tested the products discussed. Cover photography is illustrative and does not depict the specific announcement or product.

Our editorial standards
Back to all stories