AI's electricity appetite is enormous. It may also be exactly what renewable energy needed.

Global electricity consumption by data centers will more than double by 2030, reaching roughly 945 terawatt-hours, equivalent to Japan's entire annual electricity use.¹ The International Energy Agency projects that U.S. data center power consumption alone will account for nearly half of all electricity demand growth through the end of the decade.² These numbers have understandably set off alarms. But a growing body of research suggests something counterintuitive: the same demand surge raising concerns about emissions may be creating the largest private financing mechanism for clean energy deployment in history.

A demand signal without precedent

Goldman Sachs projects a 165% increase in data center power demand by 2030.³ Morgan Stanley estimates AI data centers will add 126 GW of demand by 2028, equivalent to Canada's entire power consumption.⁴ Sightline Climate's pipeline tracker counts 190 GW across 777 announced projects since 2024.⁵

BlackRock CEO Larry Fink framed AI as fundamentally an energy story in his 2026 annual letter, arguing that electrical power, not chips, is the binding constraint on AI expansion.⁶ McKinsey estimates the total infrastructure investment required at $6.7 trillion through 2030. The five largest hyperscalers plan roughly $400 billion in capital expenditure in 2026 alone.⁶

That capital has to flow somewhere. Increasingly, it flows toward clean energy.

Anchor customers for technologies the grid couldn't finance alone

The mechanism is straightforward. Data centers are creditworthy, long-term electricity buyers. When a hyperscaler signs a 20-year power purchase agreement for a solar farm or geothermal plant, it de-risks the project for investors and developers. Technologies that previously lacked sufficient market pull suddenly have a guaranteed customer. The U.S. Department of Energy has explicitly framed this dynamic as "an opportunity to accelerate the build-out of clean energy solutions."⁷

A March 2026 study from the University of Zurich and Swiss Finance Institute provides the strongest empirical evidence for this claim. Using ChatGPT's November 2022 release as a natural experiment, the researchers found that regions with high data center concentration saw significantly more new power plant investment afterward, and the increases were overwhelmingly in renewables and battery storage, not fossil fuels. Gas investment showed essentially zero net growth. The authors concluded that "growing electricity demand from AI data centers need not undermine decarbonization" when the right contractual structures are in place.⁸

The case studies reinforce this finding:

• Microsoft has contracted 40 GW of new renewable capacity across 26 countries since 2020, the largest corporate clean energy portfolio in history. Of that, 19 GW is already online.⁹
• Google and Fervo Energy scaled enhanced geothermal from a 3.5 MW pilot to 115 MW of commercial capacity, a nearly 30x increase, because Google's PPA provided the offtake certainty Fervo needed to raise $462 million in funding.¹⁰
• Google and Xcel Energy announced the world's largest battery by energy capacity: a 300 MW / 30 GWh iron-air system from Form Energy in Minnesota, paired with 1,400 MW of wind and 200 MW of solar. Google absorbs all costs through a dedicated tariff; residential ratepayers are not impacted.¹¹
• Tech companies have signed contracts for over 10 GW of nuclear capacity in roughly 18 months, including Microsoft's deal to restart the Crane Clean Energy Center (formerly Three Mile Island) and Google's partnership with Kairos Power on advanced small modular reactors.¹²

In December 2025, Alphabet acquired Intersect Power, a renewable energy developer with a 10.8 GW portfolio, for $4.75 billion. It was the first acquisition of a major clean energy developer by a Big Tech company, marking a shift from purchasing clean energy to building it directly.¹³

A cost crossover accelerated by demand

BloombergNEF's 2026 cost analysis documents a historic inversion. Solar paired with four-hour battery storage now costs $57 per megawatt-hour. New combined-cycle gas plants cost $102 per megawatt-hour, a record high driven partly by equipment demand from data centers themselves.¹⁴ Lithium-ion battery costs have fallen 93% since 2010. Solar PV costs have fallen 97%.¹⁵

Data center procurement is accelerating both sides of this divergence: pushing gas costs up through supply constraints while pulling renewables and storage costs down through manufacturing scale. Hyperscalers accounted for 49% of all global corporate power purchase agreements in 2025.¹⁴ That volume drives further cost reductions through learning curves, a dynamic that benefits every electricity consumer, not just the data centers signing the contracts.¹⁵

Acknowledging the gap between ambition and emissions

These investments have not yet translated into lower emissions. Greenhouse gas output at Google, Meta, Amazon, and Microsoft has risen across the board, in some cases by more than 50%. Natural gas still generates over 40% of the electricity powering U.S. data centers, and some companies are building on-site gas plants while purchasing solar offsets elsewhere.¹⁶

The Union of Concerned Scientists estimates that without strong clean energy policy, data center growth could drive a 23% increase in power plant CO₂ emissions by 2035 and $1.6 trillion in climate and health damages over the following decade.¹⁷ The trajectory from here depends heavily on regulatory choices.

On that front, global momentum is building. Germany will require 100% renewable electricity for data centers by January 2027. Ireland mandates 80% from new renewable projects. China requires 80% renewable sourcing in national hub regions by 2030. In the U.S., 22 states introduced more than 60 data center energy bills in 2025.¹⁸ Together, these market forces and regulatory requirements create reinforcing pressure that makes sustained reliance on fossil fuels increasingly costly.

What this means for the SDGs

The implications extend beyond SDG 7 (Affordable and Clean Energy). When data centers co-invest in grid modernization, that infrastructure serves all ratepayers (SDG 9).¹⁹ When iron-air batteries reach commercial scale in Minnesota, the technology becomes available to utilities worldwide (SDG 13).¹¹ When waste heat from Meta's data center in Odense, Denmark warms 11,000 homes,²⁰ or a Microsoft facility in Espoo, Finland supplies 40% of a city's district heating,²¹ data centers function as community energy assets rather than drains on local resources.

Enhanced geothermal, long-duration storage, and advanced nuclear were all considered promising but commercially stalled technologies five years ago. Data center demand is commercializing them at a pace that public funding alone was unlikely to achieve. The pattern (private demand pulling public-good technology to commercial viability through long-term contracts) offers a model worth studying beyond the energy sector.

Policymakers, utilities, and communities now face a design question: how to structure the rules so that this unprecedented wave of private energy investment delivers broad public benefit, not just cheaper electrons for the companies that can afford to write the checks.