The surging electricity appetite of artificial intelligence data centers is thrusting a long-dormant clean energy sector back into the limelight. Tech behemoths are intensifying their investments in long-duration energy storage, leveraging large-scale procurement and project partnerships to open up markets for novel storage technologies that were previously hampered by high costs and sluggish commercial adoption.

According to a September 18 report, since the start of 2025, at least 150 gigawatt-hours of long-duration energy storage projects in the United States have been announced, approved, or fully financed. This marks a more than fivefold increase compared to the total planned capacity nationwide at the end of 2024, with the majority of this new demand stemming from AI data centers.

Unlike conventional lithium-ion batteries, which typically provide only a few hours of storage, long-duration energy storage can hold electricity for dozens of hours or even days. This makes it better suited to meet the escalating power requirements of AI data centers and helps bridge the temporal gap between generation and consumption for renewable sources like wind and solar.

The involvement of tech giants is also delivering the much-needed commercial validation that has been scarce in this space. Alphabet is collaborating with Italian firm Energy Dome to deploy a carbon dioxide-based storage system with a 10-hour capacity in Arizona. Meanwhile, Meta Platforms, Inc. has signed an agreement with California-based Noon Energy to install a system capable of storing energy for up to 100 hours. As order books swell, long-duration storage startups are poised to secure further financing and scale up their production capabilities.

AI Demand Revitalizes Long-Duration Storage, Accelerating Technology Pathways

Long-duration energy storage encompasses a variety of technological routes, enabling the sustained storage and discharge of electricity for hours or even days. Historically, this field has been constrained by high costs and a lack of large-scale validation, leading to slow commercial progress. Now, the expansion of AI infrastructure is quickening the pace of deployment in this sector. Cully Cavness, co-founder and president of Crusoe, noted that long-duration storage is becoming a vital tool for scaling AI infrastructure.

The technology strategies adopted by the tech giants differ markedly: Alphabet is partnering with Energy Dome to achieve 10-hour storage using carbon dioxide, Crusoe is working with Energy Vault to store power through gravitational potential energy, and Meta Platforms, Inc.'s approach with Noon Energy extends storage duration to an impressive 100 hours.

Amanda Yang, head of clean and renewable energy at Meta Platforms, Inc., emphasized that "four-hour storage capacity is nowhere near sufficient." Lucia Tian, director of advanced energy technologies at Alphabet, added that some long-duration storage solutions do not depend on critical minerals like graphite, helping to reduce reliance on such supply chains.

Tech Titans' Orders Open Financing Windows for Startups

For long-duration storage startups, the significance of large data center clients extends beyond procurement volumes; it lies in aiding new technologies to gain market acceptance.

Aric Saunders, executive vice president at Noon Energy, revealed that AI data centers now account for roughly half of the company's planned deployments, a figure that was nearly zero just two years ago. He noted that while other customers are also growing, few can match the procurement speed and financial muscle of AI data centers, calling them "accelerants for the industry."

Tom Sisto, co-founder and chief executive of XL Batteries, stated that a procurement order from data center developer Prometheus Hyperscale has provided crucial market validation. For innovative hardware firms looking to expand production, such orders also help alleviate financing challenges.

Yiyi Zhou, an energy storage analyst at BloombergNEF, suggested that as these technologies first take root in AI data centers, real-world operational data and scaled manufacturing are expected to drive down costs, paving the way for broader adoption across other industries.