AI’s race for power could turn big tech into energy system innovators

AI's unprecedented demand for power is turning big tech companies into builders of their own energy infrastructure, and, under the right policies and market rules, data center investment could help accelerate innovation and deployment of clean energy technologies, argue researchers from WU Vienna University of Economics and Business.

A growing race for electricity

Global electricity demand for new data centers will double by 2030, according to the International Energy Agency, outpacing the deployment of renewables. In the US, data centers could consume up to 17% of US electricity by 2030 – 60% higher than 2024 projections.

The critical constraint is how quickly data centers can access significant levels of reliable power, rather than the cost of electricity. Technology companies, like Google, need vast amounts of reliable power within a few years – far faster than conventional infrastructure can be planned and built. To help data centers meet demands, energy storage, like new battery technologies, is emerging as a potential solution.

Data centers as part of the solution

Together with colleagues from the International Institute for Applied Systems Analysis (IIASA) and King Abdullah University of Science and Technology, the WU researchers Behnam Zakeri und Kavita Surana argue that data center plans should be judged not only as energy consumers but by the flexibility, energy storage, and clean-firm capacity they bring to the system. With the right policies in place, AI’s growing demand for power could accelerate investment in new energy and storage technologies, bringing benefits to the wider electricity system.

Behnam Zakeri: “AI is already helping to discover new materials, accelerate battery development, and optimize energy-storage systems. At the same time, the growing demand for reliable clean electricity for data centers is creating a powerful market for batteries, long-duration storage, and other flexibility technologies. Under the right conditions, the digital infrastructure boom could help accelerate the clean-energy transition beyond data centers themselves, creating spillover effects across other sectors.”

Batteries could help ease grid constraints

To understand how much on-site energy storage could help data centers meet peak demand, the researchers analyzed energy consumption at 96 UK data centers, using data from UK Power Networks. They show that if on-site batteries are used to help meet peak energy requirements, the peak imported electricity from the grid would reduce by 10–15%. This suggests that equipping data centers with batteries could ease pressure on constrained grids. With additional clean electricity generation on-site or nearby, a data center’s reliance on the grid would be even less, say the researchers.

Policy reforms to unlock the AI–energy storage nexus

The paper, published in Energy and Climate Change, proposes a series of policy reforms for system operators and market rules designed to incentivize and reward the deployment of energy storage technologies and other clean energy infrastructure in data center plans:

• Reform grid-connection rules to reward projects that provide measurable net benefits to the energy system.

• Make it easier for data centers to deploy batteries and clean generation on-site and remove market and regulatory barriers that prevent them from providing grid services.

• Treat computing load as a flexible, rather than fixed, resource by rewarding data centers for reducing demand during grid stress and matching computing activity to periods of clean electricity.

• Support innovation and make it easier to finance so that AI’s demand for reliable, clean electricity can help scale the next generation of energy-storage technologies.

Detailed study results

The AI-energy storage nexus: opportunities for clean energy transitions. Behnam Zakeri, Kavita Surana, Julian David Hunt, Volker Krey, Keywan Riahi. Energy and Climate Change. July 2026. Availbale at: https://doi.org/10.1016/j.egycc.2026.100259