Electric Utilities · Outside the Retail 100

Constellation Energy

CEG

Our rating: hold
12-month target: $319.02 +24.0%
Expected vs. S&P 500: +6.2 ptsSee the math
Next earnings: Nov 9In 4 days

Price caps hit the smaller part of Constellation's profit: about 60% comes from contracts, a tax-credit floor and average margins, a base management expects to grow 20% yearly.

Summary

Constellation runs the largest U.S. nuclear fleet and, since buying Calpine in January, a large gas fleet and retail business too. Analysts raised their 2026 earnings estimates over the past 90 days, yet the stock is down 36% in a year, partly on fears that Washington and PJM, the eastern grid operator, will cap what it earns, and partly with the AI trade. Tech companies keep signing 20-year deals for its reactors. The stock depends on contracts growing faster than caps squeeze its market-priced earnings, about 40% of the total.

$263.30

-$114.41 (-30.29%)Past year

What Constellation Energy does

Constellation owns the largest U.S. nuclear fleet, about 22 gigawatts, over three times Vistra's. Buying Calpine in January made it the biggest U.S. power producer, selling to the grid and 2.5 million customer accounts.^1,2,4,5

What it sells

Nuclear plants49% of power supplied, Q2 2026
25 reactors at 14 sites, the output Microsoft, Meta, Google and Amazon have signed 20-year deals for. Its operating record beats the industry average.
Gas and renewables34% of power supplied, Q2 2026
Calpine's 23 gigawatts, mostly gas plants in Texas, California and the Northeast, plus the Geysers, the largest U.S. geothermal site.
Retail and wholesale supply
The largest U.S. power supplier to businesses and homes by volume, per industry data. Higher customer margins helped raise 2026 guidance.

Who buys it

Grid markets such as PJM buy much of the output and pay for capacity. Microsoft, Meta, Google, Amazon and Walmart hold long contracts, and retail serves 2.5 million accounts, including three-fourths of the Fortune 100.

Where you've seen it

Three Mile Island's undamaged Unit 1 belongs to Constellation, which is restarting it for Microsoft as the Crane plant.

What tech buyers pay for

Tech companies are paying Constellation to keep, restart and enlarge reactors, not just for electricity

A tech buyer gets carbon-free power around the clock for 15 to 20 years, and usually new megawatts it can claim as its own. Constellation funds the work, about $4.3B for Google's uprates and about $3B at Calvert Cliffs, and the long contract repays it. Power still flows into the PJM grid rather than straight to a data center.

The deals also keep plants alive. Meta's contract replaces Illinois state support at Clinton, and Amazon's gives Calvert Cliffs, whose licenses end in 2034 and 2036, the revenue certainty to seek 20 more years.

Prices are not disclosed deal by deal. The average price on contracts signed before 2026 rises from $70 to $88 per megawatt-hour by 2030 as newer deals start, which suggests the newer ones pay more.^{1,2,3,4,6,7,8,9,10,33}

Average contract price, 2026: $70 per MWh
Average contract price, 2030: $88 per MWh
Contracted by 2032: About 30%

Buyer	What they get	Term and timing
Microsoft	Crane restart at Three Mile Island, about 835 MW	20 years; restart targeted for 2027
Meta	Clinton's 1,121 MW plus a 30 MW uprate	20 years from June 2027
Google	890 MW of uprates at 11 reactors, plus 2,700 MW of supply	Uprates 20 years, first online 2028; supply 15 years
Amazon	690 MW at Calvert Cliffs, including a 190 MW uprate	20 years; uprate online 2030–2032
Walmart and others	About 920 MW, including Walmart's 176 MW	15–20 years, starting 2029–2032
CyrusOne	Data centers beside Texas gas plants: 380 MW at Freestone	Announced Feb. 2026; 780 MW signed at Thad Hill

Where we landUprates are the best deals: new megawatts on sites it already runs, sold before they are built, and now backed by a federal loan offer of up to $5B. With only about 30% contracted by 2032, most of the fleet is still for sale.

Against the other nuclear owners

Vistra and Talen sell the same product, but no rival has the sites to match Constellation's volume

Exelon assembled the fleet. It bought reactors for under $30 a kilowatt in 1999 and 2000, merged Unicom and PECO, then added Constellation's plants in 2012, which had cost far more: Nine Mile Point at $439 a kilowatt in 2001 and Ginna at $810 in 2004. By 2014 Exelon ran 23 reactors with over 22 gigawatts.

Size matters most in uprates. Squeezing more power out of existing reactors is the fastest new nuclear supply, and the Energy Department's loan offers show the gap: up to $5B for about 1,100 MW at eight Constellation plants, versus up to $4.2B for 433 MW at Vistra's three.

Rivals have their own strengths. Vistra and NRG lean harder on retail, and Talen's Amazon deal is worth about $18B over 17 years from a single plant.^{4,11,26,27,28,33,34}

Company	Nuclear fleet	Biggest tech deal	Federal uprate loan offer
Constellation	About 22 GW, 25 reactors	Google: 890 MW uprates plus 2,700 MW supply	Up to $5B, about 1,100 MW
Vistra	6,448 MW, six reactors	Meta: 2,609 MW for 20 years	Up to $4.2B, 433 MW
Talen	Susquehanna, about 2.5 GW, 90% owned	Amazon: up to 1,920 MW through 2042	n/a
NRG	About 25 GW, over half gas	None comparable	n/a

Where we landConstellation's edge is breadth: it can sell restarts, uprates and firm supply across many sites at once. Vistra is the closest copy at under a third of the size. The edge lasts until new reactors become cheap to build, a question for the 2030s.

How the business makes money

The backstory

Exelon spun off Constellation in February 2022. AI data centers then made reliable, carbon-free power scarce, and a federal nuclear tax credit set a revenue floor through 2032. CEO Joe Dominguez's plan is to lock that scarcity into 15- to 20-year contracts, add Calpine's gas plants and customers, and buy back stock, while regulators argue over how much power producers should earn.

Constellation sells power three ways: at market prices, under long contracts, and through its retail business. The tax credit puts a floor under the first, contracts add a premium on top, and the cash funds uprates and buybacks.^2,4,7

1 · ScarcityIrreplaceable fleetIts 22 gigawatts of reactors cannot be copied quickly. Management says rebuilding its whole fleet would cost over three times its enterprise value.
2 · FloorTax creditUnsold nuclear output earns at least about $45 per megawatt-hour through 2032, because the federal credit tops up low prices.
3 · PremiumLong contractsTech companies pay above the floor for 15 to 20 years. Management puts the premium at $20 to $50 per megawatt-hour.
4 · UpsideMarket pricesUncontracted output earns PJM power and capacity prices when they exceed the floor. This is the part regulators are capping.
5 · PayoutUprates, buybacksCash funds reactor upgrades and buybacks. About $4.6B has repurchased about 25.5 million shares since the spin-off.

Advantages

Reactors bought cheaply, run better

Major advantage

Constellation wins on a nuclear fleet no rival can assemble now: 22 gigawatts, against Vistra's 6.4 and Talen's single Susquehanna plant. It was bought cheaply decades ago, runs above the industry average, and buyers now pay premiums for it. We see it as durable; after the tax credit ends in 2032, contract prices have to carry it.

1.Bought at the bottom. In 1999 Exelon's AmerGen venture paid $20M for Clinton and $23M for Three Mile Island Unit 1, under $30 per kilowatt. Meta and Microsoft now contract both for 20 years.
2.Others sold, it merged. While other utilities sold reactors, Exelon combined Unicom and PECO in 2000 and Constellation in 2012, passing 22 gigawatts. The 2022 spin-off kept the whole fleet together.
3.Runs harder than peers. Its capacity factor, the share of maximum output, has beaten the industry by about four points yearly since 2013. Q2 refueling outages averaged 23 days, versus the industry's 38.
4.Scale wins bundled deals. Google's deal spreads 890 megawatts of upgrades across 11 reactors at six sites, plus 2,700 megawatts of supply. A one-plant owner like Talen cannot offer that.^{4,11,2,7,26,27,10}

A tax-credit floor through 2032

Strong advantage

Every U.S. nuclear owner gets the credit, so the edge is size: Constellation has the most unsold reactor output underneath it. The credit tops revenue up to about $45 per megawatt-hour in 2026, rising with inflation. It limits the downside on most output, adds nothing when prices are high, and ends after 2032.^2,26

Gas and retail to sell around the clock

Supporting

Calpine's 23 gigawatts and the largest retail book let Constellation sell firm supply, not only one plant's output; Google's 15-year, 2,700-megawatt deal is that kind. The catch: Calpine was paid partly in stock, and Texas gas margins are falling.^2,4,7

What's changing

Contracts pile up while the stock falls

Since January it closed Calpine, won a FERC waiver that keeps Crane's restart on track for 2027, and signed Walmart, Amazon, Google. Shares fell anyway: 9.7% on Jan. 16 when the White House pushed PJM price caps, 15.3% in March as guidance missed forecasts, and 4% after FERC paused a PJM procurement on Sept. 29. On Oct. 27 the Energy Department offered up to $5B in uprate loans.^{1,5,8,9,12,14,16,17}

Competitors

Who else an investor would consider, and how they compare.

Company	Market value	What it does	Compared with Constellation Energy
Constellation Energythis stock	$93.29B	Constellation owns the largest U.S. nuclear fleet, about 22 gigawatts, over three times Vistra's. Buying Calpine in January made it the biggest U.S. power producer, selling to the grid and 2.5 million customer accounts.
VistraVST	—	Texas-based producer with six reactors, coal and gas plants, and TXU Energy retail.	Vistra leans on Texas and gas; Constellation has over three times the nuclear capacity.²⁶
Talen EnergyTLN	$15.97B	Owns Pennsylvania's Susquehanna nuclear plant and sells most of its output to Amazon.	Talen is a one-plant bet; Constellation spreads contracts across 25 reactors.²⁷
NRG EnergyNRG	$19.68B	Retail power seller to about eight million customers, backed by a growing gas fleet.	NRG is mainly retail and gas; Constellation's reactors win long tech contracts.²⁸

Bull vs. bear: two bets on the same stock

1
A share is a small slice of the company. Its value comes from the profits the company will make in the future.
2
Nobody knows those profits yet. Buyers (bulls) pay today's price betting they'll beat what most people expect. Sellers (bears) bet they'll fall short.
3
When results come in, the price moves toward whichever side was right, and that side makes money.

The question they disagree on

Can new long-term contracts grow Constellation's base earnings, the part that doesn't rely on market power prices, to $11.40–11.90 a share by 2029 even if PJM keeps capping prices?^{1,2,3,7,9,12,13,18,30}

Bull case: why buyers buy

If tech companies keep signing premium nuclear contracts, the base keeps growing and the stock's multiple recovers, whatever PJM does to market prices.

Signings picked up this year. Walmart, Google and Amazon signed long deals; management says each gigawatt contracted adds $0.40 to $1.00 a share.
A tax credit floors unsold nuclear output. Through 2032 it tops revenue up to about $45 per megawatt-hour, well below PJM's 2027 forward price near $73.
Profit forecasts are rising while the stock falls. The 2026 consensus (the average Wall Street forecast) rose from $11.75 to $12.16 a share in 90 days.
Retail is beating plan. Higher customer margins and buybacks lifted the 2026 guidance range by $0.50 a share in August.

Bear case: why sellers sell

If regulators keep capping PJM prices and data center demand cools, the market-priced 40% of earnings shrinks, and future contract prices fall with it.

Politics now sets the ceiling. PJM's capacity price is capped near $325 per megawatt-day through the 2029/30 auction, after the White House pushed for caps in January.
Texas margins are thinning. Forward ERCOT spark spreads (power price minus gas cost) were $21.85 for 2026 but $16.16 for 2027, squeezing Calpine's plants there.
Contract prices are secret. Google, Amazon and Meta prices are undisclosed, so investors cannot check management's claimed premium.
Calpine was paid partly in stock. Diluted shares rose from 314 million to 360 million in a year, so per-share growth trails profit growth.

Where Delphi landsholdWe think today's price already weighs both sides fairly.See why

What the caps can touch

Price caps squeeze the market-priced 40% of profit; the contracted 60% keeps growing underneath it

Management splits its forecast in two. Base earnings come from long contracts, unsold nuclear output valued at the tax-credit floor, gas plants at their historical minimum and average retail margins. Enhanced earnings are everything above that: power prices over the floor, better-than-average margins and gains from price swings. For 2026, base is $6.65–6.75 a share against guidance of $11.50–12.50.

Caps hit the enhanced part. Only about 5% of the nuclear fleet sat in the tax-credit zone at June 30, so most reactors earn market prices above the floor, and lower PJM prices cut into that. The floor stops the fall near $45 per megawatt-hour. Capacity payments count toward the credit's revenue test, so for reactors near the floor, lower capacity prices are partly refunded by a bigger credit.

Base grows without price help. Illinois' state support for some reactors ends in 2027, moving that output from about $34 per megawatt-hour to the credit floor, about $49 by 2028. Contracted volume rises by half by 2029, at higher average prices.

What isn't disclosed: the price on any single tech deal, and how enhanced earnings split between PJM, Texas and retail.^2,3,13,30

Base earnings, 2026: $6.65–6.75 a share
Base earnings, 2029: $11.40–11.90 a share
Fleet in tax-credit zone: About 5%

Piece of base earnings	2026	2029
Contracted clean power	36M MWh at $70	54M MWh at $85
Nuclear at tax-credit floor	101M MWh at $44.75	146M MWh at $49.88
Illinois state-supported units	53M MWh at $34.09	None; support ends in 2027
Enhanced share of total earnings	About 40%	30–35%

Where we landCaps can shrink the 40% but cannot break the 60%. The selloff is right in direction, since PJM upside is mostly gone through 2030, but too large for what caps can take.

Why it fell a third

The stock fell 36% in a year while profit forecasts rose: investors cut the multiple, not the earnings.

Constellation peaked near $401 a year ago and closed at $258.08 on Oct. 28. Over the same stretch the business delivered: Calpine closed, Crane's restart stayed on track, and new contracts arrived. Analysts' 2026 forecast rose from $11.75 to $12.16 a share in the past 90 days alone. What fell was the multiple, the price investors pay for each dollar of profit.

The biggest single blow was political. On Jan. 16 the White House and a bipartisan group of PJM-state governors asked PJM to hold an emergency auction and keep its price cap, and the stock fell 9.7%. FERC later approved the cap for two more auctions. Guidance on March 31 came in below consensus, and the stock lost 15.3% that month.

Regulation has not been one-way. In December FERC ordered PJM to write rules letting data centers sit next to power plants, which Capstone analysts called a major victory for nuclear and gas owners. In September, though, FERC paused PJM's backstop procurement, delaying clarity on new supply, and the stock lost 4% the next day.

Part of the fall is the AI trade. On Oct. 22, when worries over AI spending hit tech stocks after Meta's results, Constellation fell 5.7%, leaving it below where it traded before the Google deal two weeks earlier.^{12,14,15,16,17,18,30,32,7}

Date	Event	Stock move
Jan. 16, 2026	White House and governors push PJM emergency auction and price caps	-9.7% that day
March 2026	2026 guidance midpoint $11.50, below the $11.78 consensus	-15.3% for the month
Sept. 30, 2026	Day after FERC suspends PJM's backstop procurement to Feb. 28, 2027	-4.0%
Oct. 6, 2026	Google signs for 890 MW of uprates plus 2,700 MW of supply	+12.2%
Oct. 22, 2026	AI-spending worries hit tech after Meta and Tesla results	-5.7%

Where we landThe fall came from policy and the AI trade, not the business, which is why we don't sell. But PJM caps are real through 2030, so we wait for proof of contract prices before buying.

What to watch

The numbers that will settle the debate.

New long-term nuclear contracts▲ improving

2026 so far: ~920 MW in Q2 (incl. Walmart), Google 890 MW of uprates plus 2,700 MW supply, Amazon 690 MW^1,2,7,9

Each gigawatt adds $0.40–1.00 a share, per management, and grows the base caps can't touch.

PJM capacity price■ flat

$325/MW-day for 2028/29, at the cap; next auction closes Dec. 15^13,30

Drives much of the market-priced profit. Capped through the 2029/30 auction; an extension would cap upside further.

ERCOT forward spark spread▼ worsening

$21.85/MWh for 2026 vs $16.16 for 2027 (North hub, as of June 30)²

Tracks margins at Calpine's Texas gas plants, the main market risk outside PJM.

Next earnings report

Nov 9

In 4 days

Expected earnings per share: $3.88
Expected revenue: —

What to listen for: Whether management raises its 2029 base earnings range of $11.40–11.90 a share for the Google and Amazon deals, the first clue to what those contracts pay.^18,20,2

What would prove this wrong

✕Regulators extend PJM price caps past the 2029/30 auction, or lower them.
✕No new nuclear contract above 500 megawatts in the next two quarters.
✕Management cuts its 2029 base earnings range of $11.40–11.90 a share.
✕Crane's 2027 restart slips again.

Our ratingHoldas of Nov 2 close, $257.25 · what ratings mean

We agree earnings can grow about 22% by 2028, but even after the selloff the model beats the S&P 500 by only about 6 points.

12-month target: $319.02 (+24.0%)
Plus dividend: +0.7%
S&P 500, same method, with dividends: +18.5%
Expected vs the market: +6.2 points

The math behind the recommendation

Three 12-month price scenarios, weighted by probability, compared with the S&P 500 under the same method.

Now $257.25

Target $319.02

bull case25% likely

$378.82 +47%

More gigawatt-scale contracts land, management lifts its 2029 base range, PJM caps lapse after 2029/30, and 2028 earnings reach the top of forecasts.

$16.91 earnings per share × 22.4 = $378.82

base case50% likely

$320.82 +25%

Contracts keep arriving at about the current pace, caps stay through 2029/30, and 2028 earnings match the roughly $16.60 a share consensus.

$16.02 earnings per share × 20 = $320.82

bear case25% likely

$255.62 -1%

Regulators extend or tighten PJM caps, data center demand cools, Texas margins fall further, and 2028 earnings land near the low end of forecasts.

$15.51 earnings per share × 16.5 = $255.62

Why these odds: The 2026 forecast rose over 90 days and 2027 held flat, and contracts keep coming. But PJM caps, FERC's paused procurement and AI-spending worries make policy and sentiment risk two-sided, so bull and bear odds are equal.

How the prices are built: Wall Street expects earnings per share of $16.02 for the year starting a year from now (analysts range from $14.99 to $17.79). Today the stock trades at 19.6 times next year's earnings per share; its five-year middle is 22.14. Because earnings per share is forecast to grow fast, valuations are cut 1%, since fast growth rarely keeps its price.^{17,18,19,20,21,22,23,24,25}

What tests the call: Q3 2026 earnings (Nov. 9, unconfirmed), then PJM's 2029/30 capacity auction closing Dec. 15 (Nov 9)

Buy means expected to beat the S&P 500 by 10+ points over 12 months; Sell means expected to trail it by 10+. Inputs gathered Oct 29; the recommendation updates with the price every day. Model-generated, not personalized investment advice. How ratings work · Track record

What the price assumes

At $257.25, Constellation Energy is priced about like the average stock

Constellation Energy trades at 20 times the earnings analysts expect over the next 12 months. That ratio is the multiple: what investors pay today for each dollar of a year's earnings. That is below its five-year middle of 22.

Analysts expect earnings to grow about 22% the year after, against about 16% for the S&P 500. The S&P 500 trades at 19 times. The price assumes Constellation Energy's earnings grow roughly in line with forecasts, with no premium for doing better.

Constellation Energy multiple: 20×
S&P 500 multiple: 19×
Base case: +25%
Bear case: −1%

The skewOur bear case loses about 3¢ for every $1 our base case makes. That ratio, how much you stand to lose against how much you stand to make, is the skew.

What a gigawatt is worth

This year's contracts could add about as much per share as the worst case inside PJM's price collar takes away

Management gives its own yardsticks for 2029. Each gigawatt of nuclear sold on a long contract at $20–50 per megawatt-hour above the floor adds $0.40 to $1.00 a share. A $1–3 move in power prices is worth $0.45 to $1.35.

In 2026 it signed about 2.5 gigawatts tied to reactors: the second-quarter deals, Google's uprates and Amazon. By our math, at management's premiums that is roughly $1.00 to $2.50 a share once fully delivered around 2032. Uprate deals carry construction costs, so their economics differ.

The collar bounds the other side. Constellation cleared 18,875 MW in PJM's last auction at $325. By our math, a fall all the way to the $175 floor would cost about $2.10 a share after tax, before any tax-credit offset.^{1,2,7,9,13,30}

One gigawatt contracted: $0.40–1.00 a share
$10 PJM capacity move: About $0.14 a share
Power price, $1–3/MWh: $0.45–1.35 a share
Signed in 2026: About 2.5 GW

Where we landContracts are now the bigger lever. If Constellation keeps signing about two gigawatts a year at real premiums, it outgrows any plausible PJM cap. We'd move to Buy on evidence of those prices; Nov. 9 is the first chance.

Last earnings

Q2 2026 results, reported Aug 6

Calpine and higher capacity prices lifted adjusted earnings, and Constellation raised its 2026 guidance to $11.50–12.50 a share.^1,2,17,29

Reported Aug 6	Expected	Actual	Difference
Earnings per share	$2.37	$2.55	+7.6%
Revenue	—	—	—

Beat expectations in 4 of the last 6 quarters

Stock reaction: Down 1.5% that day, closing at $261.10

Why the stock moved

Positive: adjusted earnings per share beat by about 8%, and the guidance midpoint rose $0.50.
Weighing: GAAP profit fell to $1.42 a share, partly on a $340M unrealized hedge loss.
Weighing: extra refueling days cut nuclear capacity factor to 93.0% from 94.8%.

What it means for the stock: Strengthened the base: new contracts, Crane's approvals and Calpine's profit all arrived. It did nothing about PJM price caps, which is what moves the stock.