Power for AI: Grid Queues, Nuclear PPAs, Gas Turbines — 2026-09-12
EIA forecasts confirm US electricity demand will hit record highs in 2026-2027 due to AI data centers and manufacturing, while Texas pauses new data center interconnections. Meanwhile, Google’s 22-year nuclear PPA in Finland marks a milestone for sustaining existing nuclear plants via AI compute revenue, and gas turbine shortages are forcing data centers to pivot to industrial boilers.
Power for AI: Grid Queues, Nuclear PPAs, Gas Turbines — 2026-09-12
Top developments
EIA Forecasts Record US Power Demand Through 2027
The U.S. Energy Information Administration (EIA) reported that US electricity use will rise 2% higher this year and next, driven primarily by data centers and manufacturing expansion. This surge is pushing national power use to record highs, straining grid reliability in key regions. The forecast underscores the urgent need for new capacity additions to meet the "supercycle" of demand projected through 2027.
Google’s 22-Year Nuclear PPA Saves Finnish Plant
Google has signed a 22-year power purchase agreement with Fortum to keep the Loviisa nuclear plant in Finland operational beyond its scheduled 2030 closure. This deal marks the first time in Europe that an AI company’s compute revenue serves as the financial mechanism sustaining a national nuclear asset. The agreement highlights a growing trend of hyperscalers using long-term PPAs to prevent the retirement of existing nuclear capacity, securing low-carbon baseload for data centers.
Gas Turbine Backlogs Force Pivot to Steam Turbines
With gas turbine orders now facing delivery dates as far out as 2032, AI data centers are increasingly turning to industrial boilers and steam turbines for power generation. This shift allows operators to deploy proven, faster-to-install technologies to bridge the gap while waiting for combined-cycle gas turbine deliveries. The move signals a significant change in behind-the-meter generation strategies as the turbine supply chain struggles to keep pace with AI-driven demand.
Holtec Seeks $10B IPO Valuation on AI Nuclear Hype
Holtec International is seeking a valuation of up to $10.2 billion in its upcoming US IPO, capitalizing on the narrative that AI data centers require nuclear power. The company plans to offer 50 million shares at $15–$18 per share. This valuation reflects investor optimism about small modular reactors (SMRs) and nuclear restarts, despite industry warnings that SMRs may not be commercially viable at scale for several years.
Local view
Germany: Siemens Energy Order Book Hits €162 Billion German financial media reports that Siemens Energy’s order backlog has reached €162 billion, driven by global demand for gas turbines and grid technology linked to AI data centers. Analysts note that the company holds 12.5 GW of new gas turbine orders, positioning it as a market leader as utilities race to secure capacity. The surge in orders reflects the "gas turbine shortage" becoming a critical bottleneck for global AI infrastructure expansion.
Japan: Mitsubishi Heavy Doubles Turbine Capacity Japanese press highlights Mitsubishi Heavy Industries’ plan to double its large gas turbine production capacity by FY2030 compared to FY2024 levels. The company is investing over ¥100 billion in Japan and the US to meet surging demand from data centers. This move aims to alleviate the global supply crunch, where lead times for major turbines have extended well into the next decade.
Context & numbers
- US Electricity Demand: Projected to rise 2% annually in 2026 and 2027 due to data centers and manufacturing.
- Siemens Energy Backlog: €162 billion total order book, including 12.5 GW of new gas turbine orders.
- Holtec IPO Target: Seeking up to $10.2 billion valuation with a share price range of $15–$18.
- Gas Turbine Lead Times: Major OEMs like GE Vernova report delivery schedules extending to 2031–2032 for new orders.
On the radar
- ERCOT Interconnection Pause: The Electric Reliability Council of Texas (ERCOT) has paused statewide interconnection of new data centers while it audits its large-load queue, a process expected to influence how other RTOs handle similar backlogs.
- FERC Tariff Reforms: Stakeholders are monitoring FERC decisions on show-cause responses (EL26-67 through EL26-72) regarding large load integration rules, which could reshape how data centers pay for grid upgrades.
This content was collected, curated, and summarized entirely by AI — including how and what to gather. It may contain inaccuracies. Crew does not guarantee the accuracy of any information presented here. Always verify facts on your own before acting on them. Crew assumes no legal liability for any consequences arising from reliance on this content.
Powered by
