AI's Relentless Power Thirst Drives Deep Geothermal and 100-Hour Storage Deals

Hyperscalers race to secure 24/7 clean baseload, shifting energy procurement and accelerating long-duration battery deployments amid surging data center demand.

Tradesnaut Quant Research Desk · September 30, 2026 · 6 min read · AI Data Centers

AI's Relentless Power Thirst Drives Deep Geothermal and 100-Hour Storage Deals

Key takeaways

AI data centers' energy demands are skyrocketing, creating an urgent need for reliable, round-the-clock clean power that traditional intermittent renewables alone cannot provide.
Geothermal energy and long-duration battery storage systems are emerging as critical solutions, with major technology companies signing multi-year power purchase agreements and investing in large-scale deployments.
The trend highlights a strategic pivot in energy procurement for hyperscalers, prioritizing firm, dispatchable clean energy to overcome grid constraints and ensure operational continuity, while putting pressure on existing clean energy targets.

What changed

The global surge in artificial intelligence is driving an unprecedented expansion of data center infrastructure, creating immense demand for reliable, carbon-free electricity. Gartner forecasts global data center electricity consumption to reach 565 terawatt-hours (TWh) in 2026, marking a 26% year-over-year increase. This relentless growth has prompted a strategic shift among hyperscalers to secure always-on clean power. Just this year, Ormat Technologies (ORA) solidified its pivot into this market, announcing a 20-year Power Purchase Agreement (PPA) in January 2026 with Switch for approximately 13 megawatts (MW) of geothermal capacity to power data centers. This was followed by a more substantial February 2026 deal where Ormat agreed to supply NV Energy with up to 150 MW of new geothermal capacity to support Google's operations in Nevada, with projects expected online between 2028 and 2030. Concurrently, the long-duration energy storage (LDES) sector is seeing significant commitments. Google in February 2026 pledged to deploy a 300 MW / 30 GWh iron-air battery system from Form Energy, offering an unprecedented 100 hours of storage for a new data center in Minnesota. Similarly, Meta has inked an offtake agreement with Noon Energy for up to 1 GW / 100 GWh of LDES, with initial phases slated for 2028. These agreements signal a market recognizing that intermittent renewables alone cannot meet the unwavering power needs of AI.

The mechanism

The fundamental driver behind this shift is AI's need for continuous, uninterrupted compute power. Unlike traditional cloud workloads, AI training and inference demand a constant, high-density energy supply that cannot tolerate the intermittency of solar and wind power. The International Energy Agency (IEA) projects electricity consumption from AI workloads to grow by 30% annually, accounting for almost half of the net increase in global data center consumption between 2024 and 2030. This demand is pushing power grids to their limits; wholesale electricity costs near U.S. data centers have reportedly risen by 267%, indicating a severe imbalance between supply and the exponential demand from AI workloads. Geothermal energy, which taps into the Earth's heat, provides firm, 24/7 baseload power, making it an ideal candidate for such applications. Recent advancements in enhanced geothermal systems (EGS), leveraging oil and gas drilling technologies, are expanding its geographic viability and reducing costs. Long-duration energy storage solutions, capable of discharging power for 8 to 100 hours or more, are essential to complement variable renewable sources, ensuring a consistent energy flow when the sun isn't shining or the wind isn't blowing. This combination allows hyperscalers to meet aggressive decarbonization targets while maintaining the high uptime critical for AI operations, bypassing grid congestion and delays that can exceed three years.

Who is exposed

Companies positioned to capitalize on this demand for baseload clean energy include geothermal pure-plays and large-scale renewable developers with significant battery storage capabilities. Ormat Technologies (ORA), a geothermal leader, is directly exposed through its recent PPAs with data center operators like Switch and Google. Its focus on expanding both traditional and enhanced geothermal systems positions it to meet this specialized demand. NextEra Energy (NEE), a major utility and renewable energy developer, is another key player. Its NextEra Energy Resources division secured a record 4 gigawatts of new generation projects in Q1 2026, including 1.3 GW of battery storage, and is actively developing projects of 2 to 5 GW specifically for hyperscalers. NextEra also signed 11 PPAs and two energy storage agreements with Meta in late 2025, totaling 2.5 GW of clean energy projects, including 168 MW of battery energy storage system (BESS) capacity. While First Solar (FSLR) primarily manufactures solar modules, its role in enabling large-scale renewable generation means it remains a foundational supplier for the broader clean energy transition. The necessity of pairing solar with long-duration storage to achieve baseload power for AI data centers creates an indirect but significant link, supporting the overall market expansion. The company's ongoing investment of over $5 billion in American manufacturing and R&D infrastructure through 2026 also underscores its commitment to the domestic clean energy supply chain.

Quantitative Outlook

The market data for key players in the clean energy infrastructure space reflects a complex dynamic. Ormat Technologies (ORA) traded at 91.27 today, up 1.01%, yet has seen a decline of 11.42% over the last 30 days and 16.10% over 90 days. Its current price is near the lower end of its 52-week range of 90.36–144.94. NextEra Energy (NEE) similarly rose 0.53% today to 75.89, but is down 7.27% over 30 days and 12.88% over 90 days, trading within a 52-week range of 73.38–96.44. First Solar (FSLR) saw a 2.29% gain today to 176.93, but has experienced more significant declines of 13.46% over 30 days, 25.02% over 90 days, and 19.58% over the past year, sitting at the low end of its 52-week range of 172.16–318.25. These movements, juxtaposed against strong thematic demand and substantial project announcements, suggest broader market pressures or sector-specific headwinds impacting investor sentiment, despite the underlying growth narrative in AI data center power. The S&P 500 closed down 0.17%, and the NASDAQ was down 0.09% today, indicating a slightly negative general market. The sustained high growth projections for data center energy consumption, with US demand expected to double from 31 GW in 2025 to 66 GW by 2027 according to Goldman Sachs, points to a robust long-term demand curve for baseload clean energy and long-duration storage. Continued policy support, such as the US Department of Energy's $99 million investment in geothermal projects, and technological advancements in EGS and LDES will be crucial. Investors should monitor developments in grid infrastructure buildout, regulatory approvals for large-scale energy projects, and the pace at which hyperscalers convert their significant capital commitments—approximately $750 billion in data center expansion for Amazon, Google, and Microsoft alone in 2025 and 2026—into operational clean energy assets. Any acceleration or deceleration in these areas could shift the picture for energy providers.

Tags: Geothermal Energy, Long-Duration Energy Storage, AI Data Centers, Ormat Technologies, NextEra Energy