Constellation Energy is locking in nuclear power contracts that stretch a decade or longer, securing commitments from both the federal government and major technology companies racing to power artificial intelligence infrastructure. The General Services Administration awarded Constellation a 10-year electricity contract to procure nuclear energy and attributes for specified megawatt-hours over the contract life. Separately, Constellation signed a 20-year power purchase agreement with Microsoft to supply electricity from Three Mile Island Unit 1 to Microsoft data centers. Together, these deals signal that long-duration nuclear commitments are becoming the preferred instrument for buyers who need firm, carbon-free electricity at scale.
Why federal and corporate nuclear deals are converging now
The GSA contract and the Microsoft PPA represent two distinct demand signals arriving at the same time. Federal agencies need stable electricity pricing and clean energy attributes to meet government sustainability targets. Hyperscale data-center operators need round-the-clock power that solar and wind cannot guarantee on their own. Both buyers landed on the same solution: binding, multi-year nuclear supply agreements with Constellation.
The federal side is anchored by the GSA’s 10-year award, which purchases nuclear electricity and associated attributes for specified megawatt-hours across the full contract term. That structure gives the government price certainty and a predictable stream of clean-energy benefits, while guaranteeing Constellation a revenue floor that supports continued operation of its existing reactor fleet. The deal is notable because it marks a direct procurement of nuclear-specific electricity attributes rather than a generic renewable energy credit transaction, signaling that federal buyers are willing to differentiate among zero-carbon sources when reliability is at stake.
On the private side, the 20-year agreement with Microsoft links nuclear generation directly to the buildout of AI infrastructure. The EIA’s discussion of the Microsoft PPA highlights that the contract is tied to restarting Three Mile Island Unit 1 and delivering its output to data centers, framing it as part of a broader pattern in which data-center owners are turning to nuclear as a potential electricity source. A 20-year commitment is unusually long for a corporate PPA, reflecting both the capital intensity of bringing a shuttered reactor back online and the buyer’s need for decades of reliable supply to underpin cloud and AI services.
These two deals also reveal how different policy and market pressures are converging on similar solutions. Federal agencies face statutory and executive mandates to decarbonize their operations without compromising mission-critical reliability. Technology companies, meanwhile, are under pressure from investors and customers to grow AI capacity while keeping emissions in check. Nuclear power, with its combination of high capacity factors and low operational emissions, offers a way to satisfy both sets of constraints, provided buyers are willing to sign long enough contracts to cover the fixed costs.
The hypothesis that Constellation’s contracting pace depends on additional hyperscaler nuclear PPAs deserves scrutiny. Both of the confirmed deals closed without waiting for a wave of copycat agreements from other technology companies. The GSA contract shows that federal procurement alone can anchor a nuclear supply arrangement. The Microsoft PPA shows that a single corporate buyer can justify a reactor restart. Constellation does not appear to need two or three more 15‑year‑plus hyperscaler commitments before it moves. What both deals share is a buyer willing to commit for a long enough period to cover the fixed costs of nuclear generation, whether that buyer is a government agency or a cloud computing giant.
Contract structures tying Constellation to data-center load growth
The two confirmed contracts differ in duration and purpose but share a common architecture: specified megawatt-hours over a fixed term, purchased by a creditworthy counterparty. The GSA contract runs 10 years and procures nuclear electricity and attributes for general federal use. The Microsoft PPA runs 20 years and directs output from Three Mile Island Unit 1 to data centers. Both lock in volumes rather than relying on spot-market pricing, which insulates Constellation from short-term power price swings and gives the company predictable cash flow to support capital-intensive assets.
Three Mile Island Unit 1 is the more dramatic case. The reactor shut down in 2019 after years of operating at a loss in competitive wholesale markets, where nuclear plants have struggled against cheap natural gas and subsidized renewables. A 20-year PPA from Microsoft changes the economics by guaranteeing revenue over a timeline that can encompass restart work, regulatory approvals, and sustained operation. The EIA’s analysis places this deal alongside other nuclear and data-center arrangements, treating it as evidence that computing-driven electricity demand is reshaping the commercial case for nuclear power in the United States and could influence decisions about whether to extend or revive other reactors.
On the federal side, established procurement rules make it possible to structure electricity supply in ways that resemble corporate PPAs. The government’s contracting mechanisms allow agencies to enter fixed-term energy contracts with specified quantities and attributes, giving Constellation the same kind of long-horizon revenue certainty that a private offtaker provides. The difference is that the GSA contract serves a portfolio of federal facilities rather than a single data-center campus, spreading the benefits of firm, carbon-free power across multiple agencies and locations.
Both contract models effectively tie Constellation’s nuclear output to long-term load growth. For Microsoft, that load is the expanding footprint of AI and cloud computing, which requires dense clusters of servers with minimal tolerance for outages. For the federal government, it is a combination of existing building stock, new facilities, and electrification initiatives that increase electricity consumption even as agencies seek to cut emissions. In each case, the buyer is betting that its demand will persist or grow over the contract term, making a fixed nuclear commitment a rational hedge.
Open questions about Constellation’s nuclear pipeline beyond two deals
Two confirmed contracts, however significant, do not constitute a fully transparent trend. No primary records in the public domain confirm how many additional data-center PPAs Constellation has signed beyond the GSA and Microsoft agreements. The company operates the largest U.S. nuclear fleet, which means it has capacity to serve more buyers, but the pace of future contracting depends on variables that neither deal resolves, including regulatory conditions, regional grid needs, and the willingness of other hyperscalers or industrial customers to sign similarly long contracts.
First, reactor restarts are expensive and slow. Three Mile Island Unit 1 is one case; whether Constellation can replicate the model at other shuttered plants depends on site-specific engineering work, safety upgrades, workforce availability, and the timelines associated with licensing reviews. Each additional restart would likely require its own anchor buyer willing to accept a 15‑ to 20‑year obligation. Without such commitments, the financial case for reviving dormant reactors remains uncertain, especially in markets where wholesale prices are volatile or policy support is limited.
Second, the GSA contract does not explicitly mention AI data centers. The agency describes the purchase as nuclear electricity and attributes for federal use, with no reference to particular applications such as cloud computing or machine learning. Linking this deal to AI-driven demand therefore requires inference from the broader market context rather than from the contract itself. While it is reasonable to assume that some portion of federal electricity consumption supports digital services, the public documentation does not tie the nuclear procurement directly to that trend.
Third, the financial terms of both agreements remain opaque. Neither the GSA materials nor the EIA discussion of the Microsoft deal specify detailed pricing, escalation clauses, or performance penalties. That lack of transparency makes it difficult for outside observers to judge how generous the contracts are relative to prevailing wholesale prices or to alternative clean-energy options. It also limits the ability of other potential buyers to benchmark their own negotiations against these pioneering deals.
Even with these uncertainties, the GSA and Microsoft contracts point toward a future in which nuclear power is increasingly underpinned by long-term, bilateral agreements rather than relying solely on commodity markets. For Constellation, that shift offers a path to monetize its existing fleet and potentially justify additional investment in restarts or life extensions. For buyers, it provides a way to secure firm, carbon-free electricity at a time when both reliability and emissions performance are under unprecedented scrutiny. How many more such deals emerge-and how quickly-will determine whether these two contracts are early outliers or the leading edge of a broader realignment in the U.S. power sector.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
