CEG After the Run-Up: Is the AI Power Trade Still a Buy, or Already Priced In?

CEG still has a credible AI-power thesis, but the burden of proof has shifted

The AI-power thesis still holds up, but Constellation now needs execution to justify the story. The demand side is tangible, yet the stock is no longer a hidden idea.

Demand is real, but valuation already reflects it

U.S. data center electricity demand could nearly triple by 2030. That is a real physical shift, not just a tech buzzword. The problem for investors is that the market has already noticed. After the run-up, the question is no longer whether AI needs power; it is whether Constellation can convert that need into durable, contracted earnings.

The next proof points are operational, not rhetorical

Management needs progress on about 1 gigawatt of nuclear uprates and must show that Constellation's existing portfolio of over 5,650 megawatts of long term clean energy agreements can support lasting revenue quality. If that happens, the premium can hold. If not, waiting for clearer execution would be reasonable.

AI demand is becoming a grid problem, and that helps Constellation's story

The demand is measurable in megawatts

Some hyperscale AI campuses are now being designed to use hundreds of megawatts of electricity, comparable to the load of a medium-sized city. In addition, U.S. data center electricity demand is expected to climb to 35 gigawatts in 2030 from 19 GW in 2023. That makes power infrastructure one of the more concrete ways to express the AI build-out.

Constellation's role in that story is straightforward. It operates a fleet of nuclear reactors capable of supplying stable baseload electricity around the clock, along with renewable and grid-management assets. That combination matters because AI buyers are looking for reliable power, not just intermittent clean energy.

Why Constellation is in the frame

The company is not selling an abstract theme. It is trying to monetize existing clean-power assets and expand them through uprates. Its plan for about 1 gigawatt of nuclear uprates, combined with over 5,650 megawatts of long term clean energy agreements, is the clearest link between AI demand and potential earnings.

There is also a policy tailwind. The current administration has issued executive orders aimed at expediting reactor testing and approvals. That does not guarantee faster project execution, but it could reduce friction when developers and grid users are competing for limited capacity.

The real debate is earnings quality, not demand existence

Rising PPA prices make the case more credible

The market for clean power is tightening rather than softening. In North America, long-term corporate PPA prices rose: average solar PPA prices increased 9% year over year to $61.7/MWh, and wind prices rose 9% to $73.7/MWh. That matters because it suggests buyers are paying up for clean power, not just chasing headlines.

Against that backdrop, Constellation's pitch looks more interesting than a plain utility story. The company has been working to lock in premium, carbon-free energy revenues over many years, including through AI data center contracts. If customers truly value firm, clean power, those agreements can do more than create interest; they can support better earnings visibility.

The bear case is mainly about execution risk

The bearish case does not require doubting AI demand. It can simply argue that the shares already reflect too much optimism before all the execution risks are resolved. Constellation still faces cost and regulatory pressure on its aging nuclear fleet, and that risk sits alongside a plan centered on about 1 gigawatt of nuclear uprates and US$3.9 billion in capital spending.

That is the crux of the debate. Bulls see uprates plus AI contracts as a path to higher-quality, long-duration earnings from existing assets. Bears see more capital intensity, more regulatory exposure, and revenue projections that may look cleaner on paper than in practice.

What would justify buying, holding, or waiting on CEG now

The checklist for the next few quarters

Watch these developments:

• Whether management converts interest into signed deals that support premium, carbon-free energy revenues over many years.
• Whether about 1 gigawatt of nuclear uprates stay on schedule and translate into more dependable output.
• Whether contract activity is becoming more substantial as AI buyers seek larger, more complex power agreements.

Signals that would support buying now

• New AI-backed contracts that look durable rather than exploratory.
• Uprates that advance without fresh regulatory friction, with expediting reactor testing and approvals helping execution.
• Evidence that Constellation can compete on speed, capacity, and reliability.

Signals that would argue for waiting

• More AI-power headlines without new contracting proof.
• More complexity without a clear earnings translation.
• Worsening cost or regulatory pressure on its aging nuclear fleet.

The issue is not whether AI wants electricity. It does. The issue is whether Constellation can turn that demand into clean, repeatable earnings well enough to justify a stock that has already been re-rated.