What Big Tech’s nuclear power strategy means for construction and delivery
April 2026A series of recent announcements highlight a clear trend: major technology companies are turning to nuclear energy to meet the escalating power demands of AI and data centres. While the strategic rationale is compelling, the implications for project delivery and the construction industry are complex.
A step change in demand for nuclear infrastructure
In the UK, this shift aligns with government efforts to accelerate nuclear deployment through Great British Nuclear (GBN), the small modular reactor (SMR) selection programme and updated National Policy Statements for energy infrastructure. Similar momentum is building in Europe as states revisit their energy security strategies, creating additional opportunities but also increased competition for specialist nuclear resource.
The scale of Big Tech’s commitment is striking. In 2025, Meta signed a 20-year power purchase agreement (PPA) with Constellation Energy for the output of the Clinton Clean Energy Center to provide 1,121 megawatts of emission-free nuclear energy. Meta has similarly agreed deals with Vistra, TerraPower and Oklo to support up to 6.6 GW of nuclear capacity by 2035, positioning it among the largest corporate purchasers of nuclear power in US history.
Other technology companies are pursuing similar strategies.
Google has entered into agreements to power data centres via SMRs. Microsoft is restarting the former Three Mile Island Unit 1 reactor, now known as the Crane Clean Energy Center. Amazon recently revealed plans, in partnership with X-Energy, for the construction of 12 SMRs to be named “the Cascade Advanced Energy Facility” as part of its drive to secure reliable, low-carbon power for its expanding digital ecosystem, including its data centres, cloud storage systems and AI operations.
This trend is reinforced by a wider industry commitment: in 2025, Google, Amazon and Meta signed a pledge with the World Nuclear Association to support tripling global nuclear capacity by 2050.
The direction of travel is clear – AI and data centres require continuous, high-density power, and nuclear offers a low-carbon baseload solution at scale.
From strategy to delivery: the reality gap?
Despite the momentum, a gap remains between ambition and deliverability.
Goldman Sachs estimates that Big Tech may require 85–90 GW of new nuclear capacity to meet AI-driven demand, yet only around 10% of that is expected to be available by 2030. This mismatch underscores the scale of the delivery challenge.
Persistent barriers include high capital costs, lengthy development timelines and complex regulatory frameworks:
- Programme risk: Nuclear projects can span a decade or more, with delays in planning, permitting, financing or design having significant downstream impacts.
- Regulatory burden: Licensing regimes are extensive and subject to policy change, increasing uncertainty.
- Supply chain constraints: Shortages of specialist skills and components may drive cost escalation.
- Technology risk: Many projects rely on SMRs or advanced designs that are not yet proven at scale, creating first-of-a-kind risk.
Contracting and risk allocation challenges
These risks translate directly into contractual complexity.
- Design responsibility must be carefully considered where design maturity is limited. Traditional design-and-build approaches may be unsuitable without adaptation.
- Change in law provisions will be critical given evolving regulatory frameworks and project duration, with clear mechanisms required for time and cost relief.
- Programme and delay risk must reflect the interdependence between approvals, funding and construction activity. Unrealistic assumptions remain a common source of disputes.
- Interface risk is heightened by the number of stakeholders involved and disaggregation of contracting structures, requiring clear allocation of responsibility and coordination.
Financing and delivery
Nuclear projects are highly capital-intensive, and financing structures, often underpinned by long-term power purchase agreements, play a central role in viability.
Delivery models are also evolving. For SMR projects, developers are increasingly favouring integrated delivery structures in which the reactor vendor plays a central role. Collaborative models such as alliance contracting may be well suited if there is design uncertainty or supply chain immaturity.
For contractors, there may be increased scrutiny of programme certainty, milestone-driven payments, and enhanced funder/sponsor oversight.
Opportunities for the construction sector
Notwithstanding the risks, the opportunity is significant. Tech-sector investment is likely to drive new-build projects, life extension of existing plants, and associated infrastructure such as grid upgrades and co-located data centres.
Contractors with experience in regulated sectors will be well placed, particularly where they can engage early and adopt collaborative delivery models.
Insurance considerations
Nuclear work raises unique insurance issues, including nuclear exclusions in professional indemnity policies, limitations in contractors’ all risks cover, and the potential need for state‑backed indemnity under the Nuclear Installations Act regime. Early engagement with brokers and insurers is essential to ensure that required activities can be properly insured without creating gaps in cover.
Conclusion
Big Tech’s move into nuclear energy will lead to further growth in the sector over the coming years. However, the sector’s historic challenges, including cost, complexity and regulation, remain.
For construction professionals, success will depend on realistic risk allocation, robust contract structuring and early engagement with delivery challenges. Those who can navigate this complexity will be well positioned to capitalise on what may become a defining infrastructure trend.
If you are involved in data centres and/or nuclear procurement, design, engineering, construction or delivery, please contact Andrew Croft and David Nitek to discuss how we can support your next project.
Download PDF
