Constellation Energy: Why the Largest Nuclear Fleet Still Looks Like a Strong Buy

Corporate power deals, not just a nuclear narrative

Bottom line: the stronger part of the Constellation thesis is no longer just that "nuclear is back." It is that large technology buyers are already signing long-term power agreements, which can support earnings and asset recovery today rather than in some distant hypothetical future.

Demand is rising and customers are already visible

U.S. power use is expected to keep rising through this year and the next, with data centers doing a lot of the pulling. That matters because it creates a live market for firm, nearby power. Constellation is already tapping that market through agreements such as Microsoft's restart plan for the former Three Mile Island plant and other tech-sector arrangements with Microsoft to restart a nuclear reactor at a Pennsylvania plant. This is not a concept story; the demand signal is coming from established buyers.

Microsoft's 20-year, 837 MW power purchase agreement tied to restarting Three Mile Island is the clearest example. A contract of that size and duration is more than marketing: it offers visible revenue support for a multi-billion-dollar restart and shows how AI-era power demand can translate into actual asset recovery.

Why the debate is about timing, not demand

This is where the bull-bear split starts. Bulls see earnings support already showing up: Constellation beat fourth-quarter profit estimates, helped by rising electricity demand from data centers. Bears focus on execution risk, especially after grid operator PJM suggested the Three Mile Island connection may be delayed until 2031.

Constellation is still pushing to meet its goal of restarting the plant by the end of 2027 and plans to ask regulators for help speeding the process. That leaves investors with a live trade-off: current earnings support plus optionality on a major restart, with timing as the main source of volatility.

Constellation's nuclear fleet is the asset behind the story

The demand case only works if the hardware can actually deliver. On that score, Constellation's fleet still looks credible.

Scale and output matter

This is not a niche power seller testing a theme. Constellation operates the nation's largest nuclear fleet, and last year that fleet produced 45,459 gigawatt-hours of electricity. That scale matters because corporate buyers want proven, reliable capacity, not just a future promise.

Output slipped very slightly from 45,494 GWh a year ago, which management attributed to more planned refueling and non-refueling outage days. That is a normal operational detail, not a reason to dismiss the asset base. The larger point is that the fleet remains large, active, and well suited to serving the kind of steady load data centers need.

Location improves the value of the assets

Proximity to demand matters. Constellation already has an agreement with CyrusOne to connect and serve a new data center adjacent to the Freestone Energy Center, showing that the company is not just selling power in the abstract.

The Three Mile Island / Crane story matters for the same reason. Constellation is trying to bring generation back to a tight load area, with electricity to the regional grid to power Microsoft data centers nearby. If that comes together, the asset base becomes more than an exposure story; it becomes a deliverable power solution.

Timing is still the main risk

The main bear case is straightforward: if grid interconnection slips badly, the asset quality may be fine, but the monetization timeline gets weaker. Constellation's response is to ask regulators to transfer certain rights from Eddystone to Crane and to stick with its $1.6 billion Crane relaunch timeline.

The key signals to watch are: - progress on the grid-rights transfer - PJM's next feedback on interconnection - whether Constellation can keep the restart timeline intact

If timing improves, the location advantage can matter more for valuation. If it worsens, the story is still viable, but the near-term premium should be questioned.

What needs to happen for the bull case to hold

The next move is a sequence, not a slogan: first progress on the FERC request to transfer certain grid rights from Eddystone to Crane, then more corporate nuclear agreements, then more data-center connections. That is the cleanest way to test whether Constellation can turn today's demand into deliverable capacity. For now, the stock already has support from rising electricity demand from data centers.

The roadmap from here

Bulls see a company that could rerate as more of its fleet gets monetized around AI-era demand. Bears focus on grid delays and execution risk. The bullish view depends less on one headline deal than on whether management can keep the restart timeline moving and repeat the model.

The first box to check is regulatory and grid progress. If the Eddystone-to-Crane path advances, investors get a better signal that the 2027 restart goal is achievable.

The second box is more deals in the same mold as the existing tech PPAs. The bigger question is whether corporate procurement of nuclear power is becoming a repeatable way to underwrite nuclear assets.

The third box is physical connection, not press-release proximity. The CyrusOne win near Freestone is the kind of concrete utility investors want to see repeated across the portfolio.

What would strengthen or weaken the thesis

What would strengthen the case: - FERC/PJM movement on the grid-rights transfer - Another bankable corporate nuclear PPA - Additional adjacent data-center connection wins

What would weaken the case: - Interconnection slippage that pushes monetization far into the future - Tech PPAs that look like one-off wins rather than a repeatable model

For investors who can tolerate execution risk, Constellation still looks attractive. The risk is timing. The opportunity is that the market may still be underestimating what happens if this fleet becomes a deliverable AI power platform.