AI Demand Helps Make Nuclear Power Financeable Again

The rapid expansion of artificial intelligence is changing the economics of nuclear power as technology companies seek large volumes of reliable, low-carbon electricity for data centres.

Microsoft, Amazon, Google and Meta have moved beyond conventional renewable energy purchasing and are increasingly supporting nuclear projects through long-term power purchase agreements, investments and development partnerships. These commitments give plant operators and developers predictable future revenues, which can help them secure financing for projects that utilities might otherwise consider too risky.

The trend reflects the scale and operating profile of AI infrastructure. Data centres require electricity around the clock, while the largest AI facilities can consume as much power as major industrial sites. Unlike many traditional commercial customers, hyperscale technology companies can also sign contracts lasting 15 or 20 years, providing the revenue certainty needed to support capital-intensive energy projects.

The International Energy Agency reported that global data-centre electricity demand increased by 17% in 2025, compared with growth of 3% in overall global electricity demand. It expects total data-centre consumption to double by 2030, while electricity use by AI-focused facilities could triple. The agency also found that capital expenditure by five major technology companies exceeded $400 billion in 2025 and could rise by another 75% in 2026.

This combination of rising demand and substantial corporate spending capacity is giving nuclear developers access to a new type of customer.

Microsoft Supports a Nuclear Plant Restart

One of the clearest examples is Microsoft’s agreement with Constellation Energy to support the restart of Unit 1 at the former Three Mile Island nuclear site in Pennsylvania.

The reactor, which is separate from the unit involved in the 1979 accident, stopped operating in 2019 because of economic conditions. The facility has since been renamed the Crane Clean Energy Center.

Under a 20-year power purchase agreement, Microsoft will buy electricity from the restarted reactor. Constellation expects the plant to provide approximately 835 megawatts of carbon-free generating capacity. The company has said the restart could create 3,400 direct and indirect jobs and generate more than $3 billion in state and federal tax revenue.

The agreement provides Constellation with a long-term customer and a predictable revenue stream, reducing some of the market risk associated with bringing the reactor back into service. The plant could resume generation in 2027, subject to regulatory approvals and the completion of refurbishment work.

For Microsoft, the contract offers access to continuous low-carbon electricity as the company expands cloud and AI services. For the nuclear industry, it demonstrates how corporate demand can support the reopening of an existing asset that could not previously compete in wholesale electricity markets.

Technology Companies Back Advanced Reactors

Other technology companies are supporting new reactor designs rather than restarting existing plants.

Google has entered an agreement with Kairos Power to purchase electricity from a planned fleet of small modular reactors. Amazon has invested in X-energy and is supporting proposed SMR developments in Washington state and other US markets.

Amazon has said its investment in X-energy and related projects is intended to help deploy more than 5 gigawatts of new nuclear capacity in the United States by 2039. Its initial plans include a project developed with Energy Northwest, while partnerships involving Doosan Enerbility and Korea Hydro & Nuclear Power are intended to strengthen reactor manufacturing and deployment capabilities.

Small modular reactors are designed to be built in smaller units than conventional nuclear plants. Advocates argue that factory production, standardised designs and repeat construction could lower costs and shorten delivery schedules. However, most commercial SMR designs have not yet demonstrated these benefits at scale.

The technology sector’s involvement may help address one of the main barriers facing advanced nuclear developers: securing customers before a reactor is constructed. The IEA estimates that the pipeline of conditional offtake agreements involving data-centre operators and SMR projects increased from 25 gigawatts at the end of 2024 to around 45 gigawatts by April 2026.

Such agreements do not guarantee that every project will proceed, but they provide evidence of future demand and can improve negotiations with investors, lenders, regulators and supply-chain partners.

Why Nuclear Is Attractive to AI Operators

Nuclear power offers several characteristics that align with the needs of AI data centres. Plants can operate at high capacity factors, produce electricity continuously and generate very low operational greenhouse gas emissions.

Solar and wind remain central to corporate clean-energy strategies, and technology companies continue to sign large renewable power agreements. However, variable generation must be supported by storage, flexible demand, transmission capacity or other sources of firm power when renewable output is low.

Nuclear energy can complement renewables by supplying continuous generation, although it is generally slower and more expensive to build than new solar or wind capacity. Existing nuclear plants and reactor restarts may therefore offer more immediate opportunities than entirely new facilities.

The IEA expects renewables and natural gas to provide much of the additional electricity required by data centres through 2030 because they can be deployed relatively quickly in many markets. Nuclear projects are more likely to make a larger contribution over longer timeframes, particularly if advanced reactor developers can move from demonstration projects to repeat construction.

Financial Commitments Do Not Eliminate Project Risk

Corporate contracts can improve project bankability, but they do not remove nuclear power’s underlying challenges.

Large reactors have historically faced lengthy construction periods, cost overruns, complex licensing requirements and local opposition. Advanced reactors must also establish manufacturing capacity, secure specialised fuel, demonstrate safety systems and obtain regulatory approval before they can operate commercially.

Long-term corporate agreements may transfer some risks rather than eliminate them. Contract terms determine who is responsible if a project is delayed, exceeds its budget or fails to enter service. Electricity prices must also remain acceptable to technology buyers over agreements that can extend for decades.

There are wider questions about how AI-related infrastructure costs should be allocated. Concentrated data-centre development can require new substations, transmission lines and generating capacity. Regulators and policymakers are increasingly examining whether these investments should be paid for by data-centre operators or distributed across other electricity customers.

The IEA has warned that data centres can create affordability challenges because their loads are large, geographically concentrated and capable of expanding faster than power infrastructure. It recommends closer coordination among governments, grid operators, technology companies and energy suppliers.

Implications for the Net-Zero Transition

The growing relationship between AI and nuclear power could support the preservation of existing reactors, the commercialisation of advanced designs and the development of additional firm low-carbon capacity.

It may also reshape the way energy projects are financed. Instead of relying mainly on regulated utilities or government support, some nuclear facilities could be underwritten by corporations whose future growth depends on securing electricity.

However, nuclear development should not be considered a substitute for investment in renewable generation, transmission networks, storage and energy efficiency. AI operators will probably require a diversified supply portfolio, particularly because new nuclear projects may take several years to deliver.

For policymakers, the immediate challenge is to ensure that corporate energy demand produces additional generation and grid investment without transferring excessive financial risks to households. For technology companies, increasingly direct involvement in power generation brings greater responsibility for project delivery, community engagement, waste management and transparent emissions reporting.

AI may be making nuclear projects more financeable, but long-term success will depend on whether the industry can convert corporate commitments into reactors that are completed safely, on schedule and at an economically sustainable cost.

Source: www.forbes.com