Artificial intelligence is reshaping the electricity sector as rapidly as it is transforming the digital economy. The explosive growth of hyperscale data centres, advanced semiconductor manufacturing and electrified industry is creating a new category of electricity demand: large, concentrated loads requiring extremely high reliability. For utilities and grid operators, this is creating a challenge that goes beyond simply adding new generation capacity. The issue is whether electricity systems can deliver reliable power at scale, in the right locations, and within the timelines demanded.
A new report from the National Hydropower Association (NHA) argues that pumped storage hydropower (PSH) could play a central role in meeting that challenge. The report – Winning the AI Race: Tapping into Pumped Storage Hydropower – identifies more than 60GW of proposed PSH capacity currently moving through the US Federal Energy Regulatory Commission (FERC) licensing pipeline, across approximately 80 projects nationwide. Importantly, around 85% of those projects are located in the Western US, where electricity demand is expected to grow fastest and where many new data centre developments are being planned. For developers and system planners, the alignment between emerging demand centres and proposed storage projects highlights the potential role of long-duration storage in supporting the next phase of grid expansion.
Structural pressures on the grid
The western US is expected to experience some of the most dramatic changes in electricity demand over the next decade. According to the report, electricity consumption across the Western Interconnection is projected to grow by more than 20%, driven primarily by data centres, advanced manufacturing and broader electrification. At the same time, more than 24GW of existing coal, gas and nuclear generation is scheduled to retire across the region. For grid operators, this combination of rising demand and declining conventional generation presents a structural reliability challenge.
The replacement capacity coming online is expected to consist largely of variable renewable generation and short-duration battery storage. While these resources play a critical role in decarbonisation strategies, they do not replicate the operational characteristics of the synchronous generation that has historically stabilised electricity systems. For large data centres powering AI workloads, this issue is particularly acute. These facilities require extremely stable power supplies and cannot tolerate extended outages or significant voltage or frequency deviations. The NHA report argues that PSH can directly address this emerging reliability gap. In addition to providing long-duration energy storage, PSH facilities contribute synchronous inertia, voltage support and black-start capability – services that are becoming increasingly scarce as conventional generation retires.
Long-duration storage at scale
In the US, pumped storage already represents the dominant form of long-duration energy storage. According to the report, PSH accounts for nearly 90% of installed long-duration storage capacity in the country.
While battery storage has expanded rapidly over the past decade, most installations provide only a few hours of energy capacity. By contrast, PSH facilities are typically designed for eight hours or more of generation, enabling them to manage extended periods of system imbalance.
This capability is becoming increasingly valuable as power systems integrate higher shares of wind and solar generation. Long-duration storage allows operators to shift large volumes of energy across daily or multi-day cycles, smoothing variability and reducing curtailment.
PSH also provides operational characteristics that remain difficult to replicate with other storage technologies. Mechanical inertia, for example, remains an inherent feature of rotating hydro turbines and generators.
As renewable penetration increases and conventional synchronous generation declines, this form of physical inertia is becoming increasingly important for system stability.
Transmission constraints
The geography of electricity generation is also shaping the role of pumped storage in future grid planning.
Across the western US, many new renewable projects are being developed in remote areas with strong solar and wind resources. However, these locations are often far from major demand centres such as Phoenix, Salt Lake City and California’s coastal metropolitan regions.
Transmission congestion along major corridors already limits the ability to move low-cost electricity across the region. The NHA report highlights the potential for pumped storage to address this challenge by acting as both a storage resource and a form of network support.
Where facilities are located close to demand centres, they can operate as dispatchable generation assets, reducing reliance on imported electricity across constrained transmission corridors. Conversely, projects located near renewable generation hubs can absorb surplus energy that might otherwise be curtailed.
In effect, PSH can increase the effective transfer capability of existing transmission networks without requiring the lengthy planning and permitting processes associated with new high-voltage infrastructure. Given that major transmission projects in the US can take a decade or more to complete, this flexibility is increasingly attracting attention from planners and policymakers.
Development pipeline stalled
Despite the scale of the development pipeline, the report notes that no new pumped storage projects are currently under construction in the US, even though several projects have already secured licences. For developers, the primary obstacles are financial and regulatory rather than technical. PSH projects typically involve high upfront capital costs and long construction timelines. In electricity markets that do not fully compensate long-duration storage or reliability services, developers often struggle to secure long-term revenue certainty.
The report argues that existing electricity markets undervalue the full range of services provided by pumped storage, including inertia, frequency response and black-start capability. These attributes have historically been supplied by conventional synchronous generators, but market structures were not designed to compensate them separately. As those generators retire, the value of these services is increasing—but the mechanisms to monetise them remain limited.
Regulatory reform
The report calls for action from the Federal Energy Regulatory Commission to address these market design challenges. FERC has previously intervened to modernise electricity markets as new technologies have emerged. Past orders have required system operators to accommodate variable renewable generation and later opened wholesale markets to battery storage resources.
According to the NHA, long-duration energy storage technologies – including pumped storage – are now reaching a similar inflection point.
The report suggests that a dedicated regulatory initiative could update market participation models, establish compensation mechanisms for grid-forming services and revise capacity accreditation frameworks to better reflect long-duration performance.
Such reforms would not favour a specific technology but would allow different long-duration storage solutions to compete on equal terms based on their reliability value.
Financing large infrastructure
In addition to regulatory reform, the report highlights the potential role of federal financing tools in supporting pumped storage development. One option involves the US Department of Energy’s Title XVII loan guarantee programme, which can provide financial backing for large infrastructure projects. Under recent revisions, the programme now explicitly recognises infrastructure that supports grid reliability as an eligible category.
By offering loan guarantees or construction-risk backstops, the programme could help reduce financing risks during the multi-year construction phase of pumped storage projects. The report also highlights the Department of Energy’s Grid Resilience and Innovation Partnerships (GRIP) programme, which has already been used to support grid-enhancing technologies.
With more than US$10.5bn in funding, the programme could potentially support pumped storage development through revenue-stabilisation mechanisms during the early operational phase of projects. These measures could improve project bankability by ensuring that developers can service debt during the initial years of operation.
Policy alignment required
Ultimately, the report frames pumped storage as a strategic infrastructure resource for the emerging digital economy. AI data centres and advanced manufacturing facilities require continuous, high-quality power supplies, and many developers are already seeking locations with access to reliable electricity infrastructure.
For policymakers, the challenge is ensuring that regulatory frameworks and electricity markets evolve quickly enough to support the construction of large reliability assets. As the report concludes, the US has already entered an era in which electricity generation capacity influences economic competitiveness in artificial intelligence and advanced manufacturing.
If demand from AI continues to grow at its current pace, long-duration storage technologies such as pumped storage hydropower may become an increasingly important component of grid planning, particularly in regions where renewable integration, transmission congestion and rising load are converging.
For the hydropower sector, the question is no longer whether the technology can deliver these services, but whether the policy and market frameworks required to support new development can keep pace with the energy demands of the digital age.
