Why Nuclear Power Deals for AI Data Centers Keep Stalling After the Announcement
Major tech companies have announced billions in nuclear power agreements to fuel AI data centers, yet the real challenge isn't securing the deals,it's getting them financed, built, and insured on time. Meta has stacked up agreements worth more than 6.6 gigawatts of capacity by 2035 with Constellation, Vistra, and Entergy. Google has ordered small modular reactors from Kairos Power. Microsoft is bringing Three Mile Island back online. But behind these headline-grabbing announcements sits a quieter crisis: the financial and insurance infrastructure needed to actually deliver these projects doesn't yet exist at the scale required.
What's Blocking Nuclear Power From Scaling for AI?
The barrier to nuclear expansion isn't technology or safety,it's structure. Not a single large-scale nuclear plant is currently under construction in the United States despite more than 400 operating reactors globally, according to industry analysis. The problem is that nuclear plants are capital-intensive, slow to build, and have historically been financed on utility balance sheets in regulated markets that no longer exist in the same form.
The proposed solution involves layering infrastructure funds, pension equity, long-term bank and insurer debt, and Department of Energy guarantees alongside hyperscaler commitments to buy the power. This represents a fundamentally different capital structure than utilities have used for decades, and it's still being assembled in real time.
Until that repeatable model emerges, financiers are pricing a technology that hasn't yet proven it can be delivered on budget at scale."Natural gas is still kind of the king right now," said David Williams, vice president of nuclear business development at Kiewit Nuclear Solutions. "Nuclear's economics improve as programmes move from one-off builds to repeatable ones and a supply chain forms around them."
David Williams, Vice President of Nuclear Business Development at Kiewit Nuclear Solutions
Where Is the Insurance Bottleneck Tightening?
If financing is one gap, insurance is arguably tighter. Builders' risk coverage, the policy protecting a plant during construction, is in critically short supply relative to the pipeline. Industry discussion at a recent Financial Times Global Insurance Summit revealed that only 2 to 3 percent of planned data center projects are actually in construction, with builders' risk capacity as the specific bottleneck. Even as that capacity has grown by half over nine months in response to demand, it remains insufficient.
Nuclear adds complexity most underwriters haven't priced before. Long lead times on transformers and switchgear, delay-in-startup exposure running into millions of dollars per day, and third-party liability structures borrowed from a handful of historic nuclear builds all sit outside standard data center insurance programs.
The sector is only just working out how to approach nuclear risk."It's going to be something that is going to test the market a little bit," said Jon Tellekamp, chief underwriting officer for construction and energy at Axa XL.
Jon Tellekamp, Chief Underwriting Officer for Construction and Energy at Axa XL
How Should Companies Approach Nuclear Power Strategy?
For operators planning to use nuclear power, the practical takeaway is clear: site selection and power strategy must now be underwritten alongside financing and insurance from day one, not bolted on once a nuclear power purchase agreement is signed. This integrated approach addresses multiple constraints simultaneously rather than treating them as sequential steps.
- Financing Structure: Develop a capital stack combining infrastructure funds, pension equity, bank debt, and Department of Energy guarantees before signing power purchase agreements, rather than after.
- Insurance Planning: Secure builders' risk coverage and third-party liability structures early in the project timeline, as capacity remains limited and underwriting standards are still being established.
- Community Acceptance: Address local concerns and regulatory requirements in parallel with commercial negotiations, as community acceptance has emerged as the binding constraint on deployment speed.
The investment opportunity lies not in another gigawatt announcement, but in whoever builds the first repeatable underwriting and financing model for small modular reactor-backed data center capacity. Whoever solves this problem stands to control the pace at which the rest of the market can move. Banks, insurers, and infrastructure funds that can standardize these deals rather than pricing each project from scratch will unlock the capital needed to match AI's accelerating power demand.
The Metro Connect Fall conference in September will bring together legal advisers, insurers including Aon and Lockton, construction firms, and investors alongside fiber and hyperscale buyers for the first time under one roof. Expect the Investment and Finance Summit and Nuclear for AI Forum tracks to surface which institutions are actually underwriting these projects, what pricing looks like once repeatable deals close, and whether the capital stack can move fast enough to keep pace with AI infrastructure growth.