CEG: Why Nuclear Is The Trade Of The Decade
Constellation Energy owns the largest fleet of nuclear reactors in the United States. Microsoft is paying premium PPAs to restart Three Mile Island for AI data centers. Every other hyperscaler is now in line. Here is the full bull case on the cleanest pure play on the AI power crunch.
For 30 years nuclear was a dirty word in finance. The reactors were too expensive. The politics were too messy. The build cycles were too long. The fuel chain was complicated. And renewable subsidies made it easier to build solar and wind than to defend an existing nuclear plant. The result was an industry that quietly carried the heaviest baseload, the cleanest carbon profile, and the lowest market valuation of any electricity source in America.
That story is over. The single biggest demand shock in the history of the US electric grid is AI data center buildouts. Hyperscalers need 24/7 firm power. Renewables alone do not solve that problem. Natural gas helps but is carbon-intensive. The only large-scale, dispatchable, carbon-free, available-today baseload solution is nuclear. And in the United States, the largest fleet by a wide margin belongs to one company.
Constellation Energy, ticker CEG, is the company that owns the answer to the most expensive question in the AI buildout.
Why this matters now
CEG operates the largest nuclear fleet in the United States at a moment when hyperscalers are paying premium long-duration PPAs for firm carbon-free power, and the fleet's economic value has fundamentally repriced in a way the market is still digesting.
What CEG actually owns
Twenty-one reactors across the United States. Roughly 22 gigawatts of nameplate capacity. Adjacent gas, hydro, solar, and wind assets. A retail and commercial energy services business. A nuclear fuel services arm. A growing portfolio of long-duration data center power contracts.
The nuclear fleet is the crown jewel. The rest is icing.
The watershed moment was Three Mile Island
The Three Mile Island Unit 1 restart deal with Microsoft was the moment the model changed forever. Microsoft signed a 20-year, take-or-pay power purchase agreement at a meaningful premium to the prevailing wholesale market. CEG agreed to restart Unit 1, which had been shut for economic reasons in 2019. The implications cascade through the entire industry.
It established that hyperscalers will pay above-market rates for firm, dispatchable, carbon-free power.
It established that long-duration PPAs are the contracting model for nuclear and AI.
It established that previously uneconomic reactors are now economic.
It established that the relicensing path for existing reactors is a credible source of additional capacity.
It established a precedent that every other hyperscaler now wants their own version of.
The engine of the trade
Engine one: structural shortage of firm clean power. Hyperscalers have made aggressive net-zero commitments. They need 24/7 carbon-free electricity. Wind and solar with batteries can carry a load but cannot fully replace baseload at hyperscaler scale. Nuclear is the only proven option. The supply of nuclear is finite. The demand is exploding. Price follows.
Engine two: long-term PPAs reprice every megawatt. Every reactor that signs a long-duration AI PPA at premium pricing turns lumpy merchant revenue into stable contracted cash flow. That changes the multiple the market puts on the asset. The repricing has barely begun.
Engine three: license renewals and capacity uprates. Existing reactors are getting 20-year license renewals from the Nuclear Regulatory Commission. Some are getting capacity uprates that add meaningful generation without new construction. Each renewal extends asset life by decades. The accounting value lags the economic value.
Engine four: small modular reactors and new build optionality. Constellation has partnerships and stated interest in advanced reactor deployments. None of that is required for the thesis to work but each successful deployment is upside. The base case is the existing fleet plus contracted PPAs. The bull case adds new build.
The dashboard I track
Realized energy price per megawatt-hour. Capacity contracted under long-duration PPAs. Adjusted EBITDA growth. Free cash flow conversion. Capital allocation, especially buybacks and dividends. Nuclear production tax credit utilization. Regulatory developments at the NRC and at the state level.
The single metric that matters most is the average realized contracted price across the fleet. As long as that climbs, the operating story compounds.
What would change my mind
The first risk is regulatory. Nuclear lives and dies by federal and state policy. A hostile administration, a state-level pricing intervention, or an NRC ruling could pressure economics. The current direction of policy is supportive but politics changes.
The second risk is operational. Reactors are complex. Unplanned outages cost money. A meaningful safety event in the industry would create headline risk regardless of who experienced it.
The third risk is power price normalization. If the AI buildout disappoints, wholesale power prices ease, and PPAs do not get signed at premium rates, the thesis weakens.
The fourth risk is competition for the same customers. Vistra, NextEra, and other utilities are aggressively chasing the same hyperscaler PPAs. CEG's fleet is the largest but the market is competitive.
The fifth risk is valuation. CEG has rerated meaningfully from a sleepy utility multiple to something closer to a growth utility. Further multiple expansion is not guaranteed.
Where it sits in my framework
CEG is a core long-term position in my framework. It is one of the cleanest ways to express a long-duration view on AI power demand without taking AI hardware or AI customer risk. The downside is bounded by the underlying utility cash flows and the embedded value of the existing fleet. The upside is bounded by how many hyperscaler-scale PPAs get signed.
I size it as a sleep-well-at-night position. Bigger than my speculative names. Smaller than my growth core.
The political layer most investors miss
Nuclear has bipartisan support in 2026. The right likes it for energy independence and national security. The left likes it for decarbonization. Multiple states have moved from anti-nuclear to actively supportive in the past three years. The Inflation Reduction Act production tax credit established a price floor that protects the fleet's economics even in low-power environments. This is the most favorable political backdrop nuclear has had since the 1960s.
That tailwind is unlikely to disappear in one election cycle.
Run the numbers yourself
Read the Three Mile Island PPA announcement. Read CEG's most recent earnings transcript. Look at the contracted versus merchant revenue mix slide. Now look at the projected AI power demand chart from any major consulting firm. The math is simple. Either the AI power crunch is real, in which case CEG is one of the cleanest expressions of the trade. Or it is not, in which case this name is not for you. Decide for yourself.
Read next: NVDA: Why The King Is Not Done | APLD Stock Bull Case
*Disclaimer: MentorSurge is not a financial advisor and this is not financial advice. This post is for educational and entertainment purposes only. Nothing here is a recommendation to buy or sell any security. Utility and power stocks carry regulatory and commodity price risk. Investing involves substantial risk of loss. Numbers cited were accurate when written and change constantly. Always do your own research and consult a licensed professional before making decisions with real money.*
Educational commentary only, not individualized investment, tax, or legal advice. Investing involves risk, including loss of principal. Examples and scenarios are illustrations, not forecasts or recommendations to buy or sell a security. Older articles reflect the date they were written.
