What did the Department of Energy announce?
On October 27, 2026, the Department of Energy announced a conditional loan commitment of up to $5 billion to Constellation Energy. The commitment came from the department's Office of Energy Dominance Financing. The money would finance nuclear uprates and modernization at eight plants in Illinois, Maryland and Pennsylvania. According to the department, the projects would add nearly 1.1 gigawatts of electricity to the grid.
An uprate raises the maximum power a reactor is licensed to produce, usually by upgrading turbines, generators and control systems. Constellation plans to uprate 11 existing reactors, TechCrunch reported. The listed sites are Braidwood, Byron, Clinton, Dresden, LaSalle and Quad Cities in Illinois, Calvert Cliffs in Maryland and Limerick in Pennsylvania.
The department says the added output would power more than 930,000 homes and support more than 10,000 project-related jobs. It also frames the loan as progress toward a May 2025 executive order. That order calls for 5 gigawatts of uprates and restarts from the existing reactor fleet. A DOE program launched in March 2026, called UPRISE, is meant to speed those projects along. The department says 95% of the 5-gigawatt goal has now been identified.
The commitment is conditional, not final. Constellation must satisfy technical, legal, environmental and financial conditions before definitive financing documents are signed and money is disbursed. The company also played down the commitment. It said the commitment "does not, by itself, represent a final Constellation decision" on any individual uprate.
The Constellation announcement came during a busy week for federal energy lending. A day later, the same DOE office closed two loans totaling $5.1 billion to Alliant Energy. The money covers gas-fired plants, storage and upgrades in Iowa and Wisconsin. Utility Dive reported that the two packages together amount to $10.1 billion in loans for 3.7 gigawatts of capacity.
Which data centers will use the new nuclear power?
Most of the new capacity already has buyers, and they are the largest cloud and AI companies. Constellation said 20-year agreements with Google and Amazon have led it to invest more than $7 billion to add about 1,080 megawatts. A separate agreement announced last year with Meta supports a 30-megawatt expansion at the Clinton plant in Illinois.
Google's deal, announced October 6, 2026, covers 890 megawatts across 11 Constellation units. It is a 20-year power purchase agreement that Constellation says represents more than $4.3 billion of new investment. The company expects the first uprate to be delivered by 2028. Constellation says the agreement sustains about 4,400 existing jobs and creates roughly 7,200 construction jobs.
The Google package also includes a 15-year energy supply agreement for 2,700 megawatts from Constellation's existing PJM fleet. That structure keeps operating plants economically viable while the uprates are built. Google and Constellation also added a five-year technology alliance. Under it, Constellation will use Google Cloud and Gemini Enterprise for grid and plant operations.
Amazon's agreement, announced September 30, 2026, centers on Calvert Cliffs in Maryland. It supports about 190 megawatts of new capacity expected between 2030 and 2032, backed by a 20-year agreement for 690 megawatts. Constellation says the deal enables more than $3 billion of Maryland infrastructure investment. It also supports relicensing the 1,790-megawatt plant for another 20 years.
TechCrunch reported that most or all of the extra power from seven of the eight plants already has buyers. They are Google, Meta and Amazon. One caveat matters here. The electricity still flows onto the regional grid serving PJM and MISO, not down a private line to a single data center. The companies are buying power and its attributes, while the grid absorbs and distributes it.
Constellation argues that this structure benefits everyone. The company says its agreements show how large energy users can enable new, reliable generation. That power, it says, "flows to the regional grid and benefits all customers." That is a claim about market design, not a measured outcome. The contracts are private, and the grid operator, not the buyer, decides how the megawatts are dispatched.
Why uprates instead of building new reactors?
Uprates are incremental, but they avoid the hardest parts of building new nuclear plants. There is no new site to permit, no new transmission corridor to build and no decade-long construction program. Constellation describes uprates as one of the fastest and most cost-effective ways to add firm power from plants that are already running.
The approach is well established. The Nuclear Regulatory Commission says it has approved uprates adding about six gigawatts over the past 20 years. That is the equivalent of six large reactors. The agency says the fleet is interested in requests that could add another two gigawatts. Uprates require NRC approval because they change the licensed power level in a plant's operating license.
That history cuts both ways. It shows regulators can process uprates, and it shows the gains arrive in slices. Six gigawatts over two decades is roughly 300 megawatts a year across the entire US fleet. Constellation's 1.1 gigawatts would be unusually large for one package. That helps explain why the loan spans eight sites and 11 reactors.
The physical work is also substantial. Uprates can involve new turbines, upgraded steam generators, stronger electrical equipment and redesigned digital controls. Each item has its own supply chain and engineering schedule. A plant can only accept so much extra power before other systems need changes. The size of each uprate therefore varies by unit.
The regulator is trying to keep pace. The NRC says it is reviewing its past uprate cases. The aim is to "process these requests as efficiently as possible while maintaining safety." That matters because each request is a license amendment, with its own technical review. A large wave of applications could stretch the agency's schedule, which in turn affects when companies can deliver promised capacity.
The DOE program's 5-gigawatt target is modest against the scale of new demand. It is also a supply-side bet that existing reactors can be pushed harder faster than new plants can be permitted and built. The first Google-backed uprate is due by 2028, while Amazon's Calvert Cliffs expansion is scheduled for 2030 to 2032.
Will the loan lower electricity bills?
The department says yes. Its release says the projects will strengthen the grid and lower costs for millions of Americans. It also says they will add enough power for more than 930,000 homes. Energy Secretary Chris Wright described uprates as "commonsense energy addition policies" that support affordable, reliable power for families and businesses.
The evidence for that claim is thinner than the headline suggests. TechCrunch reported there is no evidence the new nuclear capacity will lower prices for residential ratepayers. Constellation's announcement does not say whether the projects will maintain or lower rates. The company would not confirm the government's claim to TechCrunch.
Grid conditions explain why the question matters. RTO Insider reported that PJM's 2028/29 capacity auction cleared at the $325 per megawatt-day cap. It procured 6,831 megawatts less than the reliability requirement. When capacity is scarce and prices sit at the cap, adding firm supply can ease pressure over time. That is not the same as a guaranteed bill reduction.
The financing structure matters too. A low-cost federal loan can reduce Constellation's cost of capital. The company says that could lower financing costs and let it build more capacity. Whether those savings reach households depends on contracts, regulation and market design. The power purchase agreements with Google and Amazon are commercial deals, and their prices have not been disclosed.
Taxpayers also carry risk. The loan is a federal commitment, and if a project stalls or costs rise, the government's exposure grows. DOE says its conditions are designed to protect that investment, but the terms, interest rate and repayment schedule have not been published. That limits any outside assessment of the deal's value for the public.
How does the loan process work, and what could go wrong?
A conditional commitment is a promise to negotiate, not a check. Constellation must complete due diligence, agree definitive financing documents and satisfy environmental, technical, legal and financial conditions. The department can still decline to fund the loan if those conditions are not met.
The office behind the loan has large resources and a growing track record. Utility Dive reported that the office manages loan programs with combined authority above $300 billion. It operated as the Loan Programs Office under the Biden administration. The Alliant package was the administration's tenth financial close, Utility Dive reported.
That scale explains the pace. Utility Dive reported that the office is using its funds to pursue a "nuclear renaissance." Its strategic plan also emphasizes energy affordability. The Constellation commitment fits that pattern. It is large enough to matter to one company's capital plan, and it serves a policy goal the department can cite publicly.
The Constellation commitment also includes an option to finance future uprate projects at other Constellation facilities. That gives the company a financing channel beyond the eight named sites, if the first projects deliver. It also widens the potential federal exposure beyond the current 1.1 gigawatts.
Several practical risks sit between the announcement and the megawatts. Equipment lead times for turbines and generators can stretch for years. Environmental reviews and NRC licensing take time. Construction costs can move. And Constellation has not committed to every uprate at every site, so the final portfolio may be smaller than the headline number.
Market risk is the other variable. The business case rests on long-term contracts with creditworthy buyers, which Google and Amazon provide. If a hyperscaler changes its data center plans, the demand behind a specific uprate could shift. The 20-year terms reduce that risk, but they do not eliminate it.
What does 1.1 GW mean for AI data center demand?
Data centers are a small but fast-growing share of global electricity use. The International Energy Agency estimates they consumed about 415 terawatt-hours in 2024, or roughly 1.5% of global electricity. The agency projects consumption will reach about 945 terawatt-hours by 2030. The United States accounted for the largest share in 2024, at about 45%.
One gigawatt of nuclear capacity is firm, round-the-clock power. That is the kind of supply data center operators are now contracting for. The DOE says 1.1 gigawatts is enough for more than 930,000 homes. For comparison, Constellation's Calvert Cliffs plant alone is rated at 1,790 megawatts, and the company's entire fleet is about 55 gigawatts.
The timing is the weak point. AI demand is rising now. The first Google-backed uprate is due in 2028, and Amazon's new Calvert Cliffs capacity arrives between 2030 and 2032. In the meantime, utilities are adding gas and storage capacity. The Alliant loans, for example, fund 2.6 gigawatts of gas, storage and upgrades, not nuclear.
Nuclear uprates also have a ceiling. The DOE's 5-gigawatt goal is small next to the projected growth in data center consumption. Uprates cannot be repeated indefinitely at the same plant, and each increment gets harder and more expensive. They are a bridge to new generation and grid investment, not a full solution.
Capacity and energy are also different things. The 1.1 gigawatts describes how much power the plants can deliver at a moment. It does not describe how much electricity they produce over a year. A plant that runs steadily produces far more terawatt-hours than its rating suggests. That is why grid operators value firm nuclear output. Even so, one package of uprates cannot close a demand gap measured in hundreds of terawatt-hours.
Still, the direction is clear. Large technology companies are now paying for new generation rather than only buying renewable energy credits. Google says its agreement will "strengthen the PJM grid," which the company says serves 67 million people. Constellation chief executive Joe Dominguez called the model "pairing private-sector demand and investment with smart federal financing."
What to watch next
The first milestone is the closing. Watch for definitive financing documents, a disclosed interest rate and confirmation that Constellation has committed to specific uprates. Until then, the $5 billion figure is a ceiling, not a transfer.
The second is the NRC docket. Each uprate needs a license amendment, and the agency's review schedule will determine whether 2028 is realistic for the first Google-backed unit. The commission has said the industry may seek another two gigawatts of uprates, so its capacity to process requests matters.
The third is PJM. The grid operator's capacity auctions and its "Bring Your Own Power" proposal will shape how much capacity reaches the market. State approvals will also matter, along with who ends up paying. Constellation says its Google deal is a direct response to that proposal.
The fourth is whether the ratepayer claim survives scrutiny. Regulators, state consumer advocates and independent analysts will compare the DOE's promises with actual bills. If the projects cost more than expected, or if savings stay with shareholders, the political support for federal nuclear lending could narrow.
Finally, watch the next deals. Utility Dive noted that DOE has announced or closed loans to AEP Texas, Southern Co., DTE Gas and others. It also reported an intention to lend Vistra $4.2 billion for nuclear fleet improvements. The Constellation package is a template. Whether it spreads depends on whether the first uprates deliver on time and on budget.