96 Sites. The Nuclear-AI Nexus

When Epoch AI released their Frontier Data Centers Hub in late 2025, they revealed something most investors missed: satellite imagery can detect data centre construction eight months before public announcements. Cooling towers, substation installations, and ground clearing patterns tell the story long before press releases.

Our analysis extends this methodology to nuclear facilities worldwide. Our Nuclear Infrastructure Atlas, currently in beta, tracks 96 sites aligned with the nuclear-AI nexus, focusing on facilities with potential to serve hyperscaler demand. The results challenge assumptions about project timelines, capacity predictions, and investment opportunities.

The IAEA PRIS database tracks 416 operational reactors globally, with 63 under construction. The Curtiss-Wright Nuclear Power Plant Directory provides comprehensive static data on operating facilities. WANO, representing 130 members operating approximately 460 civil reactors, shares performance indicators across its network.

Yet these valuable resources share a common limitation: they report what has happened, not what is happening. Traditional reporting relies on operator announcements, often months behind physical reality.

Consider the data. China's 32 reactors under construction represent £27bn ($34bn, €32bn) in active investment. Between 2015 and 2024, Chinese facilities achieved average build times of 6.3 years, with the fastest at 4.1 years. Meanwhile, Western projects struggle with decade-long timelines.

Hinkley Point C illustrates the gap perfectly. Originally expected in 2025, the UK's flagship project now targets 2030 for Unit 1. Costs have escalated from £18bn to potentially £46bn ($58bn, €54bn) in current prices. Yet satellite imagery shows continuous progress: the second reactor dome was installed in July 2025, with 26,000 workers currently on site.

The UAE's Barakah plant tells a different story. All four APR-1400 units achieved commercial operation by September 2024, generating 40 TWh annually. Each subsequent unit was delivered faster than the previous. Unit 4 reached commercial operation 40% faster than Unit 1.

First, official timelines lag reality by months. When Georgia Power announced Vogtle Unit 4's commercial operation in April 2024, satellite imagery had shown operational cooling patterns weeks earlier. Investors relying on press releases missed critical windows.

Second, construction progress metrics vary enormously between operators. China's definition of "construction start" differs from European standards. Without visual verification, comparing projects across regions remains guesswork.

Third, grid infrastructure development often indicates project completion before reactor announcements. National Grid invested 4 million man-hours upgrading substations around Hinkley Point C. These patterns are visible from orbit, providing leading indicators of commissioning schedules.

The monitoring methodology we're developing through the Vistergy Atlas adds a dimension that static directories and periodic performance reports cannot capture: real-time visual intelligence.

Where the Curtiss-Wright directory tells you a plant exists, and WANO indicators tell you how it performed last quarter, satellite analysis shows what's happening this week.

Thermal Analysis: Operational nuclear plants emit distinctive thermal signatures. Cooling water discharge creates visible plumes detectable from space. By analysing these patterns across selected facilities, capacity factors can be estimated before official reports. The UAE's Barakah facility, for example, showed consistent thermal signatures months before commercial operation announcements.

Construction Progress Tracking: High-resolution imagery identifies key milestones. Reactor dome installation, cooling tower completion, and turbine hall construction follow predictable sequences. At Hinkley Point C, the Unit 2 dome lift in July 2025 validated EDF's construction claims.

Grid Substation Monitoring: Substations serving nuclear facilities require distinctive equipment configurations. New transformer installations and high-voltage line construction indicate approaching commissioning. This technique provided eight-month lead time for data centre announcements in parallel analysis.

Institutional investors allocate billions to nuclear projects using quarterly reports and management guidance. Yet physical infrastructure changes continuously. The temporal gap between reality and reporting creates systematic mispricing.

Japan provides a compelling example. The country has 33 operable reactors, with restart programmes progressing unevenly. Traditional analysis tracks regulatory approvals. Satellite monitoring reveals which facilities are actually preparing for restart through maintenance patterns and staffing changes.

Similarly, China's construction programme appears uniform in official statistics. Our analysis identifies significant variation. Some facilities progress 20% faster than fleet averages. Others face unlisted delays. This granularity transforms investment timing.

Regulators are beginning to recognise satellite monitoring's value. The IAEA's 2025 AI Symposium addressed digital monitoring applications, including remote verification of construction progress.

In the UK, the Office for Nuclear Regulation has engaged with satellite monitoring providers for security applications. The methodology that tracks construction progress also detects unauthorised changes. Dual-use capabilities accelerate regulatory acceptance.

France's nuclear regulator ASN has explored thermal monitoring for operational assessment. Discharge water temperatures correlate with reactor efficiency. Independent verification complements operator reporting.

Three principles guide effective satellite intelligence deployment.

First, combine multiple data sources. Satellite imagery alone shows physical changes. Integrated with permit data, grid operator filings, and thermal analysis, it tells complete stories.

Second, establish baselines before changes occur. Monitoring Vogtle throughout construction revealed the progression from foundation to completion. New facilities benefit from day-one tracking.

Third, focus on leading indicators. Substation upgrades precede commissioning. Ground clearing precedes construction. Each phase signals what follows.

For infrastructure investors, satellite intelligence offers immediate applications. Construction verification reduces timeline risk. Operators claiming on-schedule progress can be independently validated.

For energy traders, thermal signatures provide operational intelligence. Capacity factor variations affect wholesale prices. Early detection of outages or performance changes creates trading opportunities.

For policymakers, independent monitoring supports accountability. The £28bn ($35bn, €33bn) gap between Hinkley Point C's original and current cost estimates might have been anticipated earlier with continuous construction tracking.

The hyperscaler wave has driven satellite monitoring innovation for data centres. That same capability now extends to nuclear facilities. Our Atlas platform, currently in beta, focuses on 96 sites aligned with the nuclear-AI nexus because not every reactor can serve hyperscaler demand.

The nuclear industry built to different standards across regions. China delivers in 4 to 7 years. The West struggles with 10 to 15. Satellite monitoring quantifies these differences, identifies exceptions, and reveals opportunities that traditional reporting misses.

Next week: 5 Researchers. 1,785 Hours. What We're Investigating Next.

Stop typing reproductions and start vibing code. Wispr Flow captures your spoken debugging flow and turns it into structured bug reports, acceptance tests, and PR descriptions. Say a file name or variable out loud and Flow preserves it exactly, tags the correct file, and keeps inline code readable. Use voice to create Cursor and Warp prompts, call out a variable like user_id, and get copy you can paste straight into an issue or PR. The result is faster triage and fewer context gaps between engineers and QA. Learn how developers use voice-first workflows in our Vibe Coding article at wisprflow.ai. Try Wispr Flow for engineers.

Start flowing free