AI Demand Catalyzes US LDES Expansion
Since early 2025, the US long-duration energy storage (LDES) market has witnessed a monumental surge. At least 150 GWh of LDES projects have been announced, approved, or financed, marking a more than fivefold increase from the end of 2024. This explosive growth is largely propelled by the insatiable energy demands of AI data centers.
Tech Giants Lead the Charge
Major technology corporations are aggressively securing ultra-long-duration storage to keep their AI operations running seamlessly. Google recently partnered with Form Energy for a massive 300 MW / 30 GWh iron-air battery system in Minnesota. Estimated at $1 billion, this project boasts a 100-hour discharge capability, making it the world's largest by energy capacity. Additionally, Google is collaborating with Energy Dome to deploy a 10-hour carbon dioxide energy storage system in Arizona, and planning a 23 MW / 200 MWh CO2 battery project in Ireland for 2028.
Meta is also making significant strides, signing an agreement with Noon Energy to reserve up to 1 GW / 100 GWh in ultra-long-duration storage. The arrangement begins with a 25 MW / 2.5 GWh pilot project targeted for 2028. Noon Energy reports that AI data centers now account for about half of its planned deployments.
Diversifying Storage Technologies
The push for LDES is also diversifying the technology landscape. Crusoe is collaborating with Energy Vault to utilize gravitational potential energy for storage. Furthermore, Google's Mammoth Solar project in West Virginia integrates 10 MW / 100 MWh of zinc-based long-duration storage provided by Eos Energy.
"Some LDES technologies do not rely on critical minerals like graphite, reducing supply chain dependencies," highlighted Lucia Tian, Google's Advanced Energy Technology Director, emphasizing the strategic advantage of these alternative systems.
As AI continues to reshape global energy consumption, these mega-projects underscore the critical role of LDES in building a resilient, decarbonized grid.
This article was assisted by AI analysis. Please refer to the original source for official information.