Amazon has committed to purchasing nuclear power for the next two decades under a landmark agreement with Constellation Energy, a deal that will channel more than $3 billion into energy infrastructure and reflects the growing pressure that data centers and artificial intelligence are placing on the American electricity grid. The 20-year contract, announced Wednesday, covers 690 megawatts of generation from Constellation’s Calvert Cliffs Clean Energy Center in Maryland — the state’s only nuclear facility — and includes support for a roughly 190-megawatt expansion expected to come online between 2030 and 2032. The two companies also entered a retail electricity supply agreement covering Amazon’s operations across the 13-state PJM regional grid, the largest wholesale electricity market in the United States. That arrangement, the companies said, will help Amazon stabilize energy costs across its footprint in the region. The scale of the commitment is striking. It signals not just a corporate preference but a structural shift in how the technology industry intends to power itself going forward.
To understand how this deal came together, it helps to trace the arc of Amazon’s energy strategy over the past several years. The company had long positioned itself as a leader in renewable energy procurement, signing wind and solar agreements at a pace few corporations could match. But as its data center portfolio expanded — and as the computational demands of generative AI began straining those projections — it became clear that intermittent renewables alone could not guarantee the around-the-clock reliability that industrial-scale computing requires. Nuclear power, which generates electricity continuously regardless of weather conditions, began to look less like a legacy technology and more like a strategic necessity.
Amazon moved deliberately. In 2023, the company announced a $500 million investment in X-energy, an advanced nuclear developer working on next-generation reactor designs. Around the same time, it struck an agreement with Energy Northwest to support the development of small modular reactors in Washington state. These were not passive financial bets — they were attempts to help build the supply chain for a technology that does not yet exist at commercial scale. The Constellation deal is different in character: it anchors Amazon to proven, operating capacity right now, while the longer-horizon projects mature.
Calvert Cliffs is a significant piece of infrastructure. Its two reactors can produce up to 1,790 megawatts of electricity — enough to power the equivalent of more than 1.3 million homes — and the plant employs more than 800 people in Maryland. The reactors are currently licensed through 2034 and 2036, and Constellation has been seeking 20-year license extensions from federal regulators. Amazon’s long-term revenue commitment materially strengthens the economic case for those extensions, giving Constellation the financial visibility it needs to justify the investment. Joe Dominguez, Constellation’s chairman, president and chief executive, framed the deal in explicitly infrastructural terms, arguing that it demonstrates how private capital can reinforce critical energy systems that the public depends on.
One important technical detail shapes how this arrangement actually works. Unlike some corporate nuclear agreements — most notably the deal Microsoft struck last year to reopen Three Mile Island’s Unit 1 and route its power directly to a data center — the electricity generated at Calvert Cliffs will continue flowing into the PJM regional grid rather than being funneled exclusively to Amazon facilities. Amazon is, in effect, paying to sustain and expand generation capacity that benefits the broader grid, while receiving credit for that carbon-free electricity against its own consumption. Whether that model delivers the same environmental integrity as a direct physical connection is a question that grid regulators and energy analysts continue to debate.
The Bigger Picture: Who Bears the Cost of Big Tech’s Power Appetite?
The surge in corporate nuclear procurement does not exist in a vacuum. It is happening against a backdrop of rapidly rising electricity demand driven almost entirely by the technology sector, at a moment when grid operators across the country are warning that supply may struggle to keep pace. Amazon, Microsoft, Google and Meta are collectively signing power agreements at a pace that would have seemed implausible five years ago, and nuclear — once dismissed as too expensive and too slow — has become the preferred solution for companies that need firm, clean capacity and have the balance sheets to commit to 20-year contracts. The political economy of this shift deserves scrutiny. When a single corporation can lock in two decades of generating capacity from a regional grid, the implications for other ratepayers and for public planning are not trivial.
The deal also arrives at a moment of genuine policy consequence for nuclear energy more broadly. The Biden administration invested heavily in keeping existing nuclear plants online through the Civil Nuclear Credit Program, recognizing that premature closures were eliminating irreplaceable low-carbon generation. The Inflation Reduction Act extended and expanded production tax credits for nuclear power, creating the financial conditions that make deals like this one more viable. Those policy choices — made by a government that believed in active industrial strategy — are now bearing fruit in the form of private capital flowing toward nuclear infrastructure. The lesson is not that the market solved this problem on its own. The lesson is that public policy created the conditions for the market to act.
Amazon’s vice president of AWS Global Operations, Kerry Person, described the agreement as a commitment to carbon-free energy that benefits both the grid and the communities where the company operates, noting that it supports hundreds of jobs and brings significant new generation capacity to the region. Those claims are largely accurate, as far as they go. But the broader question — how society ensures that the infrastructure costs and grid reliability risks associated with Big Tech’s explosive power demand are distributed fairly, rather than quietly shifted onto ordinary ratepayers — remains unanswered. That is a question for regulators and elected officials, not corporate press releases. And it is one that deserves a serious answer before the next 20-year deal is signed.