- The Vistergy Brief
- Posts
- Examining the Decommissioning Arbitrage
Examining the Decommissioning Arbitrage
How Shuttered Nuclear Sites Offer the Fastest Path to New AI Infrastructure Through Brownfield Advantages Everyone Overlooks
The Hook That Changes Everything
While the industry debates grid connections and new reactor designs, Microsoft quietly signed a 20-year deal to restart Three Mile Island Unit 1, a facility that's been shuttered since 2019. In Michigan, Palisades secured $1.52 billion to become the first nuclear plant ever restarted after decommissioning. And in Illinois, Constellation just rezoned 524 acres around Byron for what insiders suggest will be the largest behind-the-meter AI facility in North America.
The pattern is clear: decommissioned nuclear sites aren't liabilities. They're the fastest path to AI-ready power infrastructure.
Interactive: Global Decommissioning Opportunity Map
Click any facility to see the restart potential. Toggle filters to compare operating vs. decommissioned capacity.
The reactors at Palisades and Three Mile Island would be the first ones ever restarted after decommissioning. This isn't just a technical achievement; it's a fundamental shift in how we value nuclear infrastructure.
Consider what decommissioned sites already possess:
Grid connections: 1-3 gigawatt (GW) capacity already in place
Water rights: Cooling infrastructure worth £500m+ ($630m+, €585m+) to replicate
Security clearances: Nuclear-grade perimeters already established
Community acceptance: Decades of safe operation history
Skilled workforce: Many former operators are still local
The maths is compelling. Building new grid connections alone takes 5-7 years and costs £100-300 million ($126-378m, €117-351m) per gigawatt. Decommissioned sites bypass this entirely.
The Decommissioning Arbitrage Explained
Here's what everyone misses: decommissioning doesn't mean demolition. Most shuttered plants maintain their turbine halls, cooling systems, and switchyards in working condition for decades. They're essentially power plants waiting for fuel.
Interactive: Byron Site Development Visualisation
Byron, Illinois: 524 acres rezoned for AI infrastructure adjacent to 2,300 MW nuclear capacity
Take Byron in Illinois. In August 2024, the Ogle County Board approved a rezoning of 524 acres surrounding the site. Constellation Energy (owner and operator of the Byron plant) requested a change from agricultural to industrial. This isn't random land development. It's a strategic positioning for AI infrastructure that can tap directly into existing nuclear capacity.
The arbitrage opportunity breaks down like this:
Acquisition cost: £50-200m ($63-252m, €59-234m) for a decommissioned site
Restart cost: £800m-1.5bn ($1-1.9bn, €936m-1.8bn) vs £8-12bn ($10-15bn, €9.4-14bn) for new build
Time to power: 2-3 years (vs 8-12 years new build)
ROI timeline: Positive cash flow in year 4 (vs year 15 for new)
Engineering Reality Check: Not All Sites Are Equal
Before rushing to buy every shuttered plant, understand the technical constraints:
Viable candidates must have:
Intact reactor pressure vessels
Functional steam generators
Less than 20 years since shutdown
No major component degradation
Active operating licenses (or clear path to renewal)
Currently, only 12 sites globally meet all criteria. But those 12 represent 18 GW of potential AI-ready power, enough for 6 million graphics processing units (GPUs).
Interactive: Viable Restart Candidates
Green markers indicate sites meeting all restart criteria. Click for detailed viability scores.
The First-Mover Advantage
Although nuclear has been gaining traction for AI data centre needs, there are a few key downsides, most notably time to power and cost: High capex requirements: Capex for nuclear power plants is estimated to be 5x to 10x that of using natural gas, unless you're restarting existing infrastructure.
Companies moving now gain:
Pre-approved sites: Environmental assessments already complete
Workforce availability: Former operators eager to return
Political support: Jobs returning to economically depressed areas
Infrastructure leverage: Existing roads, water, transmission
Italy gets it. Westinghouse proposed using decommissioned nuclear sites for SMR installation, targeting construction by 2030. They're not building from scratch. They're building on foundations that already exist.
The Counter-Narrative Nobody Discusses
The nuclear industry's obsession with new technology (SMRs, fusion, advanced reactors) blinds them to the opportunity hiding in plain sight. We have 30+ GW of decommissioned capacity globally that could serve AI needs within 36 months.
Compare this to the SMR timeline:
First commercial SMR: 2029 (optimistic)
Regulatory approval: 5+ years
Supply chain development: Unknown
Cost: Still theoretical
Meanwhile, restarting Palisades will deliver 800 megawatts (MW) by 2026. That's not a projection. That's steel and concrete already in place.
Strategic Implementation Framework
For infrastructure leaders, here's your brownfield development checklist:
Phase 1: Site Assessment (Months 1-3)
Technical viability study
Regulatory pathway analysis
Community engagement plan
Partnership structuring
Phase 2: Deal Structuring (Months 4-6)
Power purchase agreement (PPA) negotiation
Restart funding arrangement
Grid interconnection rights
Water rights confirmation
Phase 3: Execution (Years 1-3)
Regulatory approvals
Component refurbishment
Workforce recruitment
AI facility construction
The Portfolio Approach
Smart money isn't betting on one site. It's assembling portfolios. Consider Constellation's strategy: Byron for new development, Dresden for potential restart, and existing operations for immediate needs. This staged approach de-risks while maximising optionality.
Interactive: Build Your Restart Portfolio
Use our interactive tool to model restart scenarios. Filter by viability score, capacity, and location.
The Monday Morning Action Plan
Stop waiting for SMRs. Stop fighting grid queues. Start here:
Map decommissioned sites within 100 miles of your data centres
Assess technical viability using public decommissioning reports
Engage current owners (many are eager to avoid demolition costs)
Structure innovative deals that share restart costs for dedicated power
Move fast: only 12 viable sites exist globally
The Bottom Line
The nuclear-AI revolution won't be built on new reactors. It'll be built on the bones of the old ones. While competitors wait for SMRs or fight for grid connections, the winners are quietly restarting shuttered plants and capturing 10-year advantages.
Decommissioning arbitrage isn't just a strategy. It's the only strategy that delivers gigawatt-scale AI power before 2030.
What shuttered infrastructure could transform your AI ambitions? The clock is ticking on these brownfield opportunities.
Explore the full interactive analysis at atlas.vistergy.com?filter=decommissioned