TL;DR · 30-second read
The Short Version
- Constellation, a large US power company that runs nuclear plants, signed a 20-year deal that pays to have power available during the busiest hours of the year, not just a total amount over the year.
- Google also signed a deal for power matched to its hour-by-hour use.
- Google and Amazon are paying to get more electricity out of existing reactors. Together that adds about as much power as a large new nuclear plant.
- Why it matters: the race to power artificial intelligence is turning electricity into a deal about timing, and the busiest hours are where grids get tight.
Constellation Energy disclosed its first Peak+ transaction in its third-quarter 2026 earnings release, filed with an 8-K on November 9. The deal is a 20-year agreement for 270 megawatts (MW) of unit-contingent energy, capacity and nuclear attributes from Constellation’s existing nuclear fleet, starting in 2028. The customer was not named. Yahoo Finance reported the launch as nuclear-and-storage contracts. Constellation describes Peak+ as an integrated product that combines new capacity, “including new generation and demand response,” with energy products for large customers.
The same release recapped a 20-year, 690 MW power purchase agreement (PPA) with Amazon tied to more than $3 billion of investment at Calvert Cliffs. It also covered a 20-year PPA with Google that will fund more than $4.3 billion of uprates adding about 890 MW across 11 reactors, plus a separate 15-year, 2,700 MW load-following agreement with Google. Constellation also reported a conditional Department of Energy (DOE) loan guarantee commitment of up to $5 billion for uprates and a $715 million deal to buy the 609 MW Rhode Island State Energy Center.
Executive Summary
Peak+ is small next to Constellation’s hyperscaler deals, but its structure matters. Most clean-power deals for large buyers have been about volume: a set number of carbon-free megawatt-hours over a year. The first Peak+ contract adds capacity, which is payment for being available when the grid is most stressed. Constellation pitches the product as support for “peak demand growth.” In the same quarter, Google signed a 2,700 MW load-following agreement, which shapes energy to the buyer’s hourly consumption. Taken together, the two deals show large buyers paying for when power arrives, not only how much.
The rest of the quarter shows where the supply is coming from. More than $7.3 billion of investment tied to Amazon and Google contracts will fund roughly 1,080 MW of added nuclear output at existing plants, arriving between 2028 and 2032. A conditional federal loan guarantee of up to $5 billion would help finance more uprates. Constellation reported adjusted operating earnings of $5.06 per share, up from $3.04 a year earlier, and raised the midpoint of its full-year guidance.
For data center developers, the practical point is that contract terms are getting more detailed. Capacity, hourly shape and delivery risk now have to be negotiated, not just a clean-energy volume. The first Peak+ deal is unit-contingent, which means peak-hour firmness has limits that buyers will need to read closely.
From Annual Volume to Peak-Hour Value
The standard clean-power contract for a large corporate buyer has been a PPA. The buyer commits to an annual quantity of carbon-free energy and claims the environmental attributes, whether or not those megawatt-hours line up with the hours its servers are drawing power. Constellation’s first Peak+ deal adds a different line item. The earnings release describes 270 MW of “energy, capacity, and nuclear attributes” over 20 years. It frames Peak+ as combining “new capacity to support peak demand growth with energy products tailored to customer needs.” Capacity is paid for availability during the grid’s most stressed hours, usually a few extreme-weather peaks each year. Those peaks, more than annual energy totals, decide how much generation and transmission a region must build to take on a new load.
Google’s second agreement this quarter approaches the same issue from another direction. It is a 15-year financial energy load-following agreement for 2,700 MW in PJM, the grid operator covering much of the Mid-Atlantic and parts of the Midwest. Load-following means the energy is shaped to match the buyer’s consumption hour by hour, rather than delivered as a flat annual block. Peak+ prices availability at the peak, and the Google agreement prices hourly shape. Neither is defined by an annual total. That is the mechanism behind the headline: the products Constellation is now writing for its largest customers put a value on timing.
Three groups are affected. Data center developers and their procurement teams will need contracts that specify when power arrives, not only its source. Grid planners size systems to peak demand, so contracts that bring new capacity matter more to them than attribute-only deals. Competing generators may face pressure to offer capacity alongside clean attributes. One 270 MW contract is not yet a market. Constellation calls it its first Peak+ transaction, and the real test is whether more follow and at what prices.
The Firmness Has Fine Print
The first Peak+ deal is unit-contingent. In power contracting, that usually means deliveries depend on specific generating units operating. If a reactor is offline during a heat wave, the seller is generally not obliged to buy replacement power. The buyer gets capacity value tied to nuclear units, which run continuously most of the time, but not an unconditional guarantee of supply at the exact hour it matters most.
Constellation’s own description hints at how it may close that gap. Chief executive Joe Dominguez described Peak+ as bringing together “new capacity, including new generation and demand response.” Demand response means paying customers to reduce their use during peak hours. That suggests firmness assembled from a portfolio rather than a single plant. Constellation has not said which resources beyond existing nuclear back the 270 MW deal, or whether batteries are part of the Peak+ toolkit.
Timing matters as well. Deliveries begin in 2028. The contract helps a buyer planning load for the second half of the decade, not one trying to energize a campus next year.
Uprates Are the Supply Behind the Pitch
An uprate raises the output of an existing reactor through equipment and licensing changes. Here is what the uprate deals add up to:
- Google: more than $4.3 billion across 11 units in Illinois, Pennsylvania and New Jersey for about 890 MW. By simple division, that is roughly $4.8 million per added megawatt, with the first uprate expected in 2028.
- Amazon: more than $3 billion at Calvert Cliffs, including a 190 MW expansion expected between 2030 and 2032, along with continued operation and future relicensing of the plant.
Together that is about 1,080 MW, consistent with Dominguez’s statement that the agreements add more than 1,000 MW, “the equivalent of a new large-scale nuclear plant.” Because uprates build on existing sites and grid connections, they avoid the siting fights that new plants face.
Financing is being lined up to match. The conditional DOE commitment would guarantee up to $5.0 billion of Federal Financing Bank lending for uprates over 30 years, at 0.375 percentage points above comparable Treasuries. It still depends on a final loan agreement. Meanwhile, the near-term picture still includes gas. Constellation agreed to pay $715 million, roughly $1.2 million per megawatt, for a 609 MW combined-cycle plant selling into ISO New England’s competitive market.
Earnings Momentum and the Cost-Shift Claim
The commercial backdrop is strong. Adjusted operating earnings rose to $5.06 per share from $3.04, about a 66% increase. GAAP net income rose to $5.19 from $2.97. Constellation raised the midpoint of its guidance for the second time this year, to a range of $12.50 to $12.80. The company attributed the quarter to favorable market and portfolio conditions and the addition of Calpine, partly offset by higher interest costs and its nuclear production tax credit portfolio.
Dominguez said the new megawatts come online “with no costs being borne by existing customers” and “all with private investment.” Parts of that claim are well supported. Named customers are underwriting specific investments under long-term contracts. One element deserves precise framing: the uprate program may also draw on a federally guaranteed loan. That is debt Constellation would repay, but the guarantee is a public backstop. How costs are allocated between new large loads and existing ratepayers is a live question across PJM, and the claim would be stronger with published detail on how Peak+ capacity is accounted for in that process.
Background
Constellation Energy operates a large US nuclear fleet. With its acquisition of Calpine Corporation it has also added a substantial natural gas generation business. Much of its fleet sits in PJM, the grid operator for much of the Mid-Atlantic and parts of the Midwest, where fast-growing data center demand has made long-term power contracts with hyperscalers such as Amazon and Google central to the company’s strategy.
In wholesale markets like PJM and ISO New England, generators earn money in two main ways. They sell energy, the electricity actually produced, and they sell capacity, a commitment to be available when demand peaks. Corporate clean-energy buyers have historically focused on energy volumes and environmental attributes. Peak+ and load-following agreements move those contracts toward capacity and timing. Source: Constellation Energy (CEG) Launched Peak+ Nuclear And Storage Contracts (Yahoo Finance), on the launch of Constellation’s Peak+ contracts. See also Constellation pursues hybrid capacity contracts alongside nuclear uprates (Utility Dive). Primary sources: Constellation Energy Form 10-Q for the quarter ended September 30, 2026 (filed November 9, 2026); Constellation Energy Form 8-K (filed November 9, 2026); Exhibit 99.1: Constellation Reports Third Quarter 2026 Results; Exhibit 99.2: Third Quarter 2026 Earnings Conference Call presentation.Sources