When Google and Xcel Energy announced plans for a new data center in Pine Island, Minnesota earlier this year, the most interesting number was not necessarily the gigawatts of clean power behind it. It was the hours.

The agreement includes a 300 MW / 30 GWh iron-air battery system from Form Energy, designed to discharge for up to 100 hours. Xcel Energy described it as the largest battery project by gigawatt-hour energy capacity announced to date.

Beyond its scale, the project signals something broader. The next phase of the energy transition will not be defined only by how much clean electricity can be generated, but by whether it can be made available when data centers, factories, cities and households actually need it.

That makes long-duration energy storage more than an efficient grid technology. It is becoming a test of whether clean power systems can deliver not just low-cost electricity, but reliable electricity over time.

The Storage Boom Has A Duration Challenge

Solar and wind have already won much of the cost argument. Battery storage is now scaling at remarkable speed. Not just because it can absorb excess renewable generation, shift electricity to periods of higher demand and help stabilize the grid, but also because of declining battery costs and opportunities for improved returns.

According to BloombergNEF, global energy storage additions, excluding pumped hydro, reached 112 GW in 2025, crossing the 100 GW threshold for the first time. 1 The U.S. alone added a record 57.6 GWh of battery energy storage capacity, up 30% from 2024. 2

Yet, most of that storage is still short duration . Lithium-ion batteries, especially lithium iron phosphate, dominate the market and are exceptionally good at shifting solar power into the evening peak, balancing intraday fluctuations and supporting grid stability. What they are less suited for is sustaining power systems through prolonged periods of low renewable generation or multi-day supply-demand imbalances.

That challenge becomes increasingly important as grids absorb more variable renewable generation while electrified industry - and particular AI infrastructure and data centers - drive demand for continuous clean power availability. The sheer speed at which new data centres are being deployed is creating concentrated demand for power in specific locations, putting pressure on local grids and, in some cases, contributing to higher electricity prices. As a result, access to reliable clean power is increasingly becoming a competitive factor rather than simply an energy consideration.

This is where long-duration energy storage, or LDES, enters the picture.

Long-duration energy storage (LDES) is broadly defined as eight hours or more of discharge, up to multi-day. 3 LDES installations rose by 49% in 2025 to more than 15 gigawatt-hours. The Long Duration Energy Storage Council estimates that up to 8 TW of LDES capacity could be needed globally by 2040. However, reaching that scale would require deployment to accelerate by as much as fifty-fold over the next fifteen years. 4

Not All Storage Scales The Same Way

It is tempting to talk about LDES as if it were one technology category waiting for its lithium-ion moment. But that would be misleading.

The sector spans very different technologies, from flow and sodium batteries to compressed air, pumped hydro, hydrogen-based systems and thermal storage. 5 But the more important distinction is not the technology itself. It is whether these systems scale like products or like infrastructure.

Some of these technologies behave less like batteries and more like large energy infrastructure assets. Their role in the power system resembles that of power plants or major industrial facilities, while their deployment often depends on specific geographical, geological or engineering constraints. Pumped hydro requires suitable geography with natural elevation differences. Compressed-air storage often depends on underground formations or engineered storage sites. Hydrogen storage requires pipelines, natural caverns and industrial integration.

These technologies can be powerful and durable, but each project is highly site-specific. In the United States, pumped hydro still accounts for 88% of installed utility-scale storage , yet most natural suitable sites have already been developed.

Other LDES technologies sit closer to the industrial logic of batteries. Flow batteries, sodium-based systems, iron-air and metal-air batteries and other modular electrochemical solutions can be manufactured, transported and deployed closer to the point of use. They can be paired with renewable generation, installed behind the meter and expanded incrementally as demand grows. Unlike lithium-ion systems, where longer duration typically requires adding more cells, several LDES technologies allow storage capacity and power output to be scaled separately. In practice, this can make it easier to optimise projects for available space and site constraints. It also means they can deliver energy for a longer periods without a corresponding increase in cost, making them increasingly attractive for multi-hour or multi-day storage needs.

That distinction matters because it creates very different competitive landscapes. Manufacturing-based LDES technologies will compete on production capacity, supply chains, learning curves and cost reductions. Infrastructure-based LDES technologies will compete on permitting, engineering, project development, and access to suitable sites.

Flow Batteries: The Interesting Middle

Among the LDES pathways, flow batteries occupy a particularly strategic middle ground. They are closer to the industrialisation logic of batteries yet are designed for longer-duration applications traditionally associated with other forms of energy storage. Unlike conventional batteries, they store energy in liquid electrolytes held in tanks, while relying on separate, largely standardized power-conversion units, allowing power and energy duration to be scaled independently. Their safety profile, long cycle life and suitability for longer-duration applications as well as being able to provide short time grid supporting services make them particularly attractive for stationary storage.

China is already demonstrating what scale can look like. In Xinjiang, Rongke Power has commissioned a 200 MW / 1,000 MWh vanadium redox flow battery paired with a one-gigawatt solar farm. On the back of strong national and provincial policy support, China now accounts for roughly 93% of cumulative global new generation of LDES installations .

The country also accounts for 70% of global vanadium mining capacity and processes more than half of the world’s vanadium supply. 6 But raw materials are only part of the equation. Flow batteries also depend on key industrial components, including membranes, stacks, piping, pumps and tanks. As the sector scales, manufacturing capacity for these components could easily become concentrated.

Importantly, vanadium is not the only active electrolyte component available for flow batteries. Alternative chemistries based on more abundant materials could enable fully domestic U.S. or European value chains without sacrificing performance.

The lesson from solar and lithium-ion batteries is that regions that fail to build industrial capacity early often end up importing the products while the manufacturing expertise, supply-chain leverage and value creation accumulate elsewhere.

While LDES may increasingly be needed by the system, the system does not yet consistently pay for what it delivers.

Lithium-ion batteries continue to dominate the four- to eight-hour storage market thanks to their mature supply chains and declining costs, while funding for long-duration storage fell sharply in 2025, with venture capital investment down 72% .

Electricity markets still reward short-term arbitrage far better than resilience, multiday adequacy or deferred grid investment.

AI is beginning to change that dynamic. Data centers and electrified industry increasingly require round-the-clock clean power that can be deployed faster and closer to load than many infrastructure-heavy storage solutions can realistically achieve. An EPRI and LDES Council benchmark projects roughly 37% cost reductions by 2030 for intraday electrochemical LDES, against 6 to 25% for compressed-gas systems. 7

Large offtakers could play a decisive role. Data centers, industrial users and utilities can anchor early projects through long-term contracts and clean capacity agreements. The Google–Form Energy agreement may ultimately do for long-duration storage what early hyperscaler PPAs did for utility-scale solar: create bankable demand at scale. It could also help change the financing logic. Long-term contracted revenues make large LDES projects easier to assess as infrastructure assets, where investors are used to high upfront capital costs in exchange for predictable cash flows over time. That distinction matters because high initial capex remains one of the sector’s biggest barriers to scale.

Building The Industry Behind The Demand

In the United States, manufacturing and investment incentives continue to provide important support for domestic storage production and deployment through 2033, even as tighter foreign-content requirements reshape supply chains.

Europe has the ambition . And activity is emerging across the continent. In Switzerland, for example, FlexBase is developing a 2.1 GWh underground flow-battery facility, illustrating both the technological diversity of the sector and the scale of projects now being pursued. 8

But long-duration storage is not one industry scaling along one curve. Some technologies depend on manufacturing capacity and supply chains. Others depend on infrastructure, permitting, engineering and geography. The challenge is not simply to back a winner, but to create the conditions for different routes to scale.

Closing the gap will require market designs that properly value long-duration flexibility, procurement mechanisms that create bankable demand, public-private finance for first commercial projects, and clearer strategies for materials, manufacturing and supply chains.

The urgency is growing. The rapid expansion of AI infrastructure, alongside broader electrification trend, is creating demand for reliable clean power that extends well beyond a few hours of storage. Whether long-duration storage will be needed is no longer in doubt.

The harder question is whether Western economies can also build the manufacturing ecosystems and infrastructure capabilities needed to deploy these technologies at scale.

For now, the window remains open. But as the experience of solar panels and lithium-ion batteries has shown , leadership in clean energy industries rarely waits for latecomers.

1 https://about.bnef.com/insights/clean-energy/energy-storage-enters-the-100-gigawatt-era-three-things-to-know/

2 https://seia.org/news/united-states-installs-58-gwh-of-new-energy-storage-in-2025/

3 https://ldescouncil.com/

4 https://ldescouncil.com/long-duration-energy-storage-companies-form-council-to-accelerate-carbon-neutrality/

5 https://ldescouncil.com/cost-benchmarking-for-long-duration-energy-storage-solutions/

6 https://tamarindo.global/insight/analysis/storage-wars-the-battle-for-vanadium-and-why-china-will-win-again/

7 https://ldescouncil.com/cost-benchmarking-for-long-duration-energy-storage-solutions/

8 https://eandt.theiet.org/2026/06/03/switzerland-building-world-s-largest-underground-battery-store-europe-s-green-energy