To The Moon

When a Utility Stops Behaving Like One

Wall Street cannot seem to decide what Constellation Energy is. That may be precisely why the opportunity — and the danger — is so interesting.

Is Constellation Energy a utility deserving a utility multiple? A merchant power generator riding an unusually favourable electricity cycle? Or has artificial intelligence transformed its nuclear fleet into scarce digital infrastructure with cooling towers?

I think all three descriptions contain some truth. The mistake is assuming investors must choose only one.

Wall Street cannot decide which game Constellation is playing

At $262.11, $Constellation Energy Corp(CEG)$ has fallen dramatically from its 52-week high of $412.70 despite raising 2026 adjusted operating EPS guidance to $11.50–$12.50. The shares now trade at 25.4 times trailing earnings and roughly 21.2 times forward earnings. Analysts' mean target of $348.30 implies approximately 33% upside.

That hardly resembles a market comfortably agreeing with the analysts.

AI Has Discovered the Electricity Bill

The bull thesis starts with scarcity.

AI data centres consume enormous quantities of electricity, but the important issue is not simply how many megawatts they require. Hyperscalers want dependable power around the clock, preferably carbon-free, and they want it without waiting years for generation and transmission infrastructure.

Nuclear therefore possesses an attribute solar and wind cannot independently replicate: clean baseload generation operating regardless of whether the sun has clocked off or the wind has taken the afternoon off.

Constellation's agreements illustrate the commercial value. $Microsoft(MSFT)$ has a 20-year PPA supporting the restart of the 835 MW Crane Clean Energy Center. $Meta Platforms, Inc.(META)$ a has contracted for 1,121 MW from Clinton for 20 years. During Q2, Constellation added another 920 MW of long-term nuclear PPAs lasting 15–20 years.

Yet here is an underappreciated distinction: these headline nuclear agreements are not simply behind-the-meter deals removing power from the grid. Crane and Clinton are structured to supply power into their respective grids while financially matching corporate demand.

That matters because it makes the bullish story less dependent on regulatory acceptance of one controversial co-location model.

And Calpine adds another card. Its CyrusOne agreements in Texas now exceed 1,100 MW of contracted data-centre development around gas-generation sites. Constellation is consequently becoming less a pure nuclear bet and more a platform for selling something increasingly valuable: reliable megawatts in the right place at the right time.

The AI premium climbed quickly. Gravity eventually requested a meeting

The Numbers Have Their Own Cooling Tower

The financial statements explain why both camps can claim victory.

At $262.11, Constellation trades at 25.4 times trailing earnings and roughly 21.2 times forward earnings. That still places it at a meaningful premium to the roughly 19–20 times earnings traditionally associated with large US electric utilities. The premium is therefore real, but considerably less extravagant than it was when CEG traded above $400.

That distinction matters. The bulls argue Constellation deserves more than a utility multiple because its nuclear fleet, long-duration hyperscaler contracts and exposure to structurally tightening power markets give it an earnings profile conventional regulated utilities cannot easily replicate. The bears see essentially the same assets and reach the opposite conclusion: electricity generation remains capital-intensive, cyclical and increasingly expensive to finance. AI may have changed the customer, but it has not repealed the laws of corporate finance.

The valuation maths exposes the argument neatly. If investors ultimately decide CEG deserves something closer to a conventional 19–20 times multiple, even strong earnings growth could be partially swallowed by multiple compression. Conversely, if long-term power contracts make earnings more predictable and cash generative, today's premium may prove less an AI surcharge than the market recognising that Constellation is no longer an ordinary utility.

Trailing 12-month revenue reached $31.27 billion, up 26%, while operating income jumped to $4.71 billion from $3.03 billion in FY2025. EBITDA reached $7.95 billion, producing a 25.4% EBITDA margin. Net income rose to $3.47 billion and trailing EPS reached $10.33.

Operating cash flow is also healthy at $4.21 billion.

Then I reach free cash flow and the orchestra suddenly drops its instruments.

Trailing free cash flow is only $295 million, versus $1.27 billion during FY2025. Meanwhile, total debt has climbed from $9.50 billion at year-end 2025 to $24.70 billion, with net debt around $24.00 billion. Net debt-to-EBITDA has consequently risen from 1.04 times to approximately 3.02 times.

AI wants megawatts. Megawatts, inconveniently, still require a balance sheet

This does not mean Constellation is financially distressed. It means investors paying a growth multiple cannot lazily treat adjusted EPS as though it were cash arriving by courier.

The Calpine transaction, integration expenditure and capital investment create a substantial gap between accounting earnings and near-term free cash flow. At the same time, interest expense has risen to $783 million on a trailing basis.

With the US 10-year Treasury around 5% and the 30-year above 5.3%, capital has rediscovered that it would quite like to be paid.

That is CEG's valuation problem in one sentence.

Calpine: Dilution Today, Optionality Tomorrow

Constellation paid approximately $22 billion for Calpine, including 50 million newly issued shares and $4.5 billion cash.

Those shares matter. Half of the newly issued stock became free of its lock-up on 30 June 2026; the remaining half remains locked until 30 June 2027. That creates a technical overhang which has little to do with whether nuclear electricity will be valuable in 2030.

There is another overlooked piece. Constellation is selling roughly 4.4 GW of predominantly gas-fired PJM generation to LS Power for $5 billion and the 606 MW Brazos Valley plant for $860 million, subject to closing conditions.

So today's leverage snapshot arguably looks worse than the eventual post-divestiture structure. Successful completion could recycle billions of dollars into balance-sheet repair while Constellation retains Calpine's broader strategic footprint.

That is not glamorous. Balance-sheet plumbing rarely trends on social media. It can, however, materially change equity value.

Competition: The Moat Has Neighbours

Constellation does not own the AI-power trade.

$Vistra Energy Corp.(VST)$ has signed 20-year Meta agreements covering 2,609 MW from its PJM nuclear fleet, including planned uprates. NRG is pursuing its Bring Your Own Power strategy and negotiating a 1.2 GW gas project for a hyperscaler. Utilities and independent generators are racing to monetise generation sites, grid connections and dispatchable capacity.

Constellation's advantage is breadth. Its combined fleet now spans nuclear, gas, geothermal, hydro, renewables and storage, while its commercial business reaches major corporate customers.

Its disadvantage is equally clear: scarcity attracts investment. Today's extraordinary nuclear premium will encourage uprates, restarts, gas construction and eventually new nuclear capacity.

A moat is valuable. A moat with Vistra digging next door deserves watching.

FERC Is the Uninvited Dinner Guest

The largest strategic risk may be regulatory rather than technological.

FERC and PJM have been developing rules governing co-located large loads because regulators must determine who pays for transmission, reliability and capacity when enormous data centres sit beside generating assets.

The issue is politically combustible. If hyperscalers effectively secure privileged access while ordinary customers inherit additional grid costs, regulators will notice. Regulators are rather talented at noticing things once voters receive larger electricity bills.

This could restrict some behind-the-meter economics.

But Constellation's mix of virtual PPAs, grid-delivered nuclear contracts and Calpine-powered co-location means the company is not dependent upon a single regulatory architecture. That diversification is more important than the market often acknowledges.

Scarcity lifts the story. Capital still decides what it weighs

The Multiple Is the Battlefield

I do not think $Constellation Energy Corp(CEG)$ should be valued as an ordinary utility. Its merchant generation exposure, nuclear scarcity, long-duration corporate contracts and data-centre optionality make that comparison increasingly inadequate.

Nor would I value it as software wearing a hard hat.

At $262, the forward multiple around 21 times already reflects much of the brutal re-rating from the $412 peak. The question has therefore shifted. Investors are no longer being asked to pay almost any price for the AI electricity narrative; they are being asked whether approximately 21 times forward earnings adequately compensates them for leverage, regulatory uncertainty and capital intensity while preserving exposure to a genuine structural shortage of reliable electricity.

For me, the most important numbers from here are not another hyperscaler headline or analyst target. They are free cash flow, net debt and the cash realised from asset disposals.

If those improve while long-term contracted power continues expanding, today's sell-off may eventually look like Wall Street mistaking balance-sheet indigestion for structural illness.

If they do not, the bears will have identified the uncomfortable truth: even in the AI age, electricity may travel at nearly the speed of light, but valuation gravity travels faster.

@TigerStars @Daily_Discussion @Tiger_comments @Tiger_SG @Tiger_Earnings @TigerClub @TigerWire