Artificial intelligence has set off one of the largest infrastructure programs in the history of technology. Workloads that once ran on small clusters now demand thousands of chips, warehouse-scale data centres, and grid connections rated in the hundreds of megawatts. The International Energy Agency (IEA) estimates that global data centres consumed roughly 485 terawatt-hours (TWh) of electricity in 2025, and it expects that figure to reach about 950 TWh by 2030. It is also why AI announcements increasingly read like energy news, with power deals and grid constraints now sharing headlines with chip launches.

This roundup compiles the principal statistics on electricity use, construction, chips, spending, and power supply. All figures are drawn from the sources cited at the end of the article.

How Much Electricity Do Data Centres Use?

The IEA estimates that global data centres consumed about 485 TWh in 2025, equivalent to roughly 1.5% to 2% of world electricity use and more than the annual consumption of many major economies.

Under the agency's central forecast, demand reaches about 950 TWh by 2030, or around 3% of global electricity demand — an almost two-fold increase in five years. Growth does not stop there: the IEA base case climbs to roughly 1,200 TWh by 2035. Measured instead as the generation required to supply data centres, including system losses, the figure could reach 1,300 TWh — close to three times the 2025 level.

The forecasts carry substantial uncertainty. Small changes in AI efficiency, model demand, or chip technology multiply into large differences when applied across millions of servers.

AI Is the Main Driver of Growth

Not every data-centre megawatt serves AI. Traditional facilities host websites, cloud storage, databases, streaming, enterprise software, social networks, search, and financial systems.

AI is the fastest-growing workload. The IEA expects electricity use by accelerated servers — the hardware mainly used for AI — to grow about 30% a year through 2030, versus only about 9% a year for conventional servers. Accelerated servers could account for nearly half of the net increase in global data-centre electricity demand over that period, with cooling and supporting equipment adding another large share. Taken together, electricity use by AI-focused data centres is expected to triple between 2025 and 2030.

AI Could Be Half of Data-Centre Workloads by 2030

JLL estimates that AI accounted for about 25% of data-centre workloads in 2025, a share that could reach 50% by 2030.

The character of the demand is shifting as well. The first phase of the boom was dominated by training large models; increasingly, power and compute are being consumed by inference — running AI models for end users. Every chatbot prompt, AI-powered search result, generated video, and software agent requires inference computing. As hundreds of millions of people and businesses use AI each day, inference could eventually require more total computing capacity than training.

U.S. Data-Centre Power Demand Could Quadruple

The United States sits at the centre of the building boom. U.S. data centres used about 177 to 192 TWh of electricity in 2024, and some current projections suggest the figure could climb as high as 793 TWh by 2030 — more than four times the 2024 level in six years.

The IEA expects data centres to account for almost half of U.S. electricity-demand growth through 2030. By the end of the decade, U.S. data centres could use more electricity than the country's combined production of aluminium, steel, cement, chemicals, and other energy-intensive goods.

Almost 100 GW of New Capacity, at a Cost of $3 Trillion

Electricity demand translates directly into construction. JLL expects almost 100 gigawatts (GW) of new data-centre capacity worldwide between 2026 and 2030, which would roughly double global capacity. Delivering it could require about $3 trillion of investment, with the global data-centre market growing about 14% a year through 2030.

For scale: a single 1 GW AI campus can need the output of a large conventional power station.

North America Has 66 GW Under Construction

North America leads the boom. JLL reported about 66 GW of capacity under construction in mid-2026, with around 95% of it already committed to tenants — the main reason vacancy remains near 1% despite record building.

North American absorption reached 25 GW in the first half of 2026, twice the level of a year earlier and five times the level of two years earlier. Hyperscalers made up about 59% of tenant demand, neocloud providers 11%, and pure-play AI companies 7%.

Atlanta Overtakes Northern Virginia in Construction

CBRE, focusing on major North American wholesale markets, reported 7.48 GW under construction across primary markets in the first half of 2026, up 24.8% from a year earlier.

Atlanta overtook Northern Virginia for capacity under construction, with 2.88 GW against 2.42 GW. Northern Virginia remains the largest existing market, however, with about 4.50 GW of operating capacity and a vacancy rate of 0.2%. Roughly 80% of the capacity under construction in primary markets was already preleased.

Europe Is Expanding Beyond Traditional Hubs

Europe is also growing quickly. The traditional FLAP-D markets — Frankfurt, London, Amsterdam, Paris, and Dublin — had about 3.8 GW of live capacity by the first half of 2026, with another 1.4 GW under construction and about 2 GW planned.

Grid limits are pushing large AI projects toward regions with more available power, and growth is spreading to Scandinavia, Spain, Portugal, Italy, and Eastern Europe rather than staying confined to the historic hubs.

AI Servers, GPUs, and Custom Chips

TrendForce estimates that global AI server shipments will reach about 2.8 million units in 2026, growth of 30% to 31% from a year earlier. AI servers could generate more than 74% of total server-industry value; they cost far more than standard machines and can carry multiple high-end chips, advanced networking, high-bandwidth memory, and liquid cooling.

GPUs are expected to make up 69.7% of AI server shipments in 2026. Servers built on custom chips — application-specific integrated circuits, or ASICs — are expected to make up 27.8% and could approach 40% by 2030. Examples include Google TPUs, Amazon Trainium, Microsoft Maia, and Meta MTIA. Microsoft said in October 2026 that it could deploy more than one million Maia 300 chips in 2027; a Maia 300 rack holds 96 chips and draws about 167 kilowatts. Reuters reported that Microsoft's chip push is aimed in part at saving costs — one motivation behind the industry's shift toward custom silicon.

Nvidia Blackwell and the Shift to Inference

Nvidia remains the principal supplier of high-end AI chips. TrendForce expects Blackwell products to make up about 71% of Nvidia's high-end GPU shipments in 2026, up from about 61%.

The five largest North American cloud companies are expected to account for more than 60% of global demand for Nvidia's GB and VR series AI systems in 2026. TrendForce estimates their AI training compute could grow more than 56% year over year, while inference computing power could rise about 122% — a gap that underscores the growing importance of inference.

China Is Deploying Millions of AI Accelerators

China is one of the few markets with recent unit-level accelerator estimates. IDC data reported by Reuters shows that more than 2 million AI accelerators were shipped in China in the first half of 2026, with Nvidia shipping about 678,000 units and Huawei about 572,000.

Nvidia's share of China's data-centre AI accelerator market fell to about 34%, while Huawei reached about 28%. The shift toward domestic chips reflects U.S. export controls that limit Chinese access to Nvidia's most powerful products.

Cloud Capex Is Approaching $900 Billion

TrendForce estimates that combined 2026 capital expenditure by Amazon, Microsoft, Alphabet, Meta, Oracle, ByteDance, Alibaba, Tencent, and Baidu will exceed $886.7 billion, an increase of about 90% from a year earlier. The five North American hyperscalers account for almost 90% of the total. Stanford's 2026 AI Index likewise shows hyperscaler capex more than doubling since ChatGPT launched.

Across the wider industry, AI infrastructure spending could reach about $811 billion in 2026. Amazon, Microsoft, Alphabet, and Meta together are expected to spend about $630 billion on AI infrastructure.

Company budgets are now measured in the hundreds of billions of dollars. Amazon has outlined about $200 billion of capital expenditure for 2026, with AI infrastructure making up much of it. Alphabet raised its 2026 capex forecast to $195 billion to $205 billion, and Microsoft is expected to spend more than $175 billion during calendar 2026.

Power May Become the Biggest Constraint

AI companies first worried about securing enough GPUs; increasingly, the binding limit is electricity. A new AI campus can need more than 1 GW of power — the demand of hundreds of thousands of homes — and grid connections can take years to secure. Other bottlenecks include transformers, gas turbines, transmission lines, cooling equipment, land, permits, water, and skilled construction workers.

Companies are therefore building or financing their own power. In October 2026, American Electric Power announced a $4 billion, 702 MW fuel-cell project to supply Google operations in Wyoming.

Because the IEA publishes several scenarios rather than a single forecast, how quickly these bottlenecks clear is one of the variables that will determine which of its projections proves closest.

What Powers AI Data Centres?

The IEA estimates today's data-centre electricity supply at roughly 30% coal, 27% renewables, 26% natural gas, and 15% nuclear. Renewables are expected to meet nearly half of the added data-centre demand through 2030. Natural gas is also expected to play a large role because it can provide steady output when renewable generation fluctuates, and nuclear could become more important toward 2030 and beyond.

This is why Microsoft, Google, Amazon, and others have signed deals covering nuclear plants, next-generation reactors, renewables, and gas.

The Biggest Uncertainty Is Efficiency

Forecasting AI electricity use is difficult. New chips improve performance per watt, and models are becoming more efficient through methods such as quantization, distillation, and mixture-of-experts designs. But cheaper AI tends to invite far heavier use — a dynamic known as the rebound effect. Consider a model that becomes ten times cheaper to run: if usage rises 50-fold, total electricity use still rises fivefold. For this reason the IEA uses several scenarios rather than a single precise figure for 2030.

The Bottom Line

Global data centres used about 485 TWh of electricity in 2025, and the IEA expects consumption to approach 950 TWh by 2030, with AI-focused facilities growing far faster than conventional computing. Almost 100 GW of new capacity could be needed worldwide between 2026 and 2030 at a cost of about $3 trillion. North America already has about 66 GW under construction, roughly 95% of it committed.

Hardware is expanding at a similar pace: AI server shipments are expected to reach about 2.8 million units in 2026, custom chips are gaining share from GPUs, and the nine largest cloud providers are expected to spend more than $886 billion in 2026.

The central question may no longer be whether companies can build more capable models. It is whether electricity, chips, data centres, transformers, cooling systems, and financing can keep up. Following that answer means watching the same series compiled here: IEA scenario updates, construction and absorption figures from JLL and CBRE, TrendForce shipment and capex estimates, and hyperscaler budget disclosures.

Sources

• International Energy Agency: Key Questions on Energy and AI, 2026
• International Energy Agency: Energy Demand From AI
• International Energy Agency: Energy Supply for AI
• JLL: 2026 Global Data Center Market Outlook
• JLL: North America Data Center Report, Midyear 2026
• CBRE: North America Data Center Trends H1 2026
• JLL: EMEA Data Centre Mid-Year 2026
• TrendForce: Global AI Server Shipments 2026
• TrendForce: AI Server Shipments and CSP Capex, August 2026
• TrendForce: Nvidia Blackwell GPU Shipments
• Stanford University: 2026 AI Index Report, Economy Chapter
• Reuters: Microsoft plans huge deployment of its own AI chips
• Reuters: US AI boom faces electric shock
• Reuters: Data centers' flexible power usage could save grid billions
• Source: CoinCentral — AI Data Centre Statistics 2026