Microsoft and Meta have locked in nuclear power for decades through major contracts with Constellation Energy, and the deals are reshaping how the entire energy supply chain gets valued. These aren't one-off purchases; they represent a fundamental shift in how hyperscalers are securing reliable, carbon-free power for AI infrastructure. The contracts span 18 to 20 years and involve physical delivery from existing U.S. reactors, setting a template that's already attracting competitors and reordering investor priorities across three very different companies.

The momentum is real and quantifiable. Constellation Energy, the largest U.S. nuclear operator, signed approximately 920 megawatts of long-term nuclear contracts during the second quarter alone, with an average duration of 18 and a half years. The company declined to identify all customers, but the disclosed deals with Microsoft and Meta provide the clearest picture of what's driving demand. During the same quarter, Constellation's fleet produced 40 terawatt hours at a 93% capacity factor, and during a mid-Atlantic heat wave, the company hit a nuclear capacity factor above 99%.

Why Are Hyperscalers Suddenly Obsessed With Nuclear Power?

AI data centers consume staggering amounts of electricity. Training and running large language models requires continuous, reliable baseload power that solar and wind alone cannot provide. Nuclear offers what hyperscalers desperately need: firm, carbon-free generation that runs 24/7 without weather dependency. By signing multi-decade contracts with investment-grade counterparties, companies like Microsoft and Meta are locking in predictable costs and securing the power they need to scale AI infrastructure without hitting grid constraints.

The framing from Constellation's leadership is blunt and revealing. CEO Joe Dominguez stated: "We have a peak capacity concern, not an energy concern." This signals that the bottleneck isn't total energy availability; it's having enough power available at peak times to support the simultaneous operation of massive AI clusters. Nuclear solves that problem in a way that renewable energy alone cannot.

Joe Dominguez

How to Understand the Three-Layer Supply Chain?

• Reactor Operators: Constellation Energy owns and operates the nuclear fleet, signs the power purchase agreements (PPAs) with hyperscalers, and receives the long-term revenue streams. The company carries a market cap of roughly $99 billion and trades at a forward price-to-earnings ratio of 23, with shares around $290.04 after a 20.81% year-to-date decline. Adjusted operating earnings per share reached $2.55 in the second quarter, and full-year guidance was raised to $11.50 to $12.50 per share.
• Component Manufacturers: BWX Technologies is the sole U.S. supplier of naval nuclear reactors and is aggressively building out domestic commercial nuclear component manufacturing. The company reported Q2 2026 revenue of $902 million, up 18% year over year, with adjusted earnings per share of $1.07. Backlog ended the quarter at $8.4 billion, up 40% year over year, and commercial operations organic revenue grew 33%. The company received a $21 million Department of Energy award to support domestic capacity expansion.
• Equipment Suppliers: GE Vernova supplies the turbines, transformers, switchgear, and other grid equipment that make the entire buildout physically possible. The company's total backlog reached $176 billion, up $13 billion from the previous quarter, with management on track to reach $200 billion in 2027. Data-center electrification orders exceeded $5 billion in the first half of 2026 alone.

These three companies sit at fundamentally different points in the supply chain, and treating them as a single investment basket would misprice all three. Constellation owns the reactors and signs the PPAs, receiving PPA-backed cash flows. BWX builds the components and dominates naval nuclear, anchored by government contracts. GE Vernova sells the equipment and operates on a backlog-driven model with a very different cash flow profile.

What Are the Key Risks and Opportunities?

For Constellation Energy, the bull case is compelling: the company is the only U.S. operator with a fleet of scale to sign multi-decade hyperscaler PPAs, roughly 30% of clean baseload output is now under long-term contract, and PJM (Pennsylvania-New Jersey-Maryland Interconnection) capacity pricing is tightening. However, risks include the Illinois Zero Emission Credit (ZEC) program ending in May 2027, unsettled PJM and Federal Energy Regulatory Commission (FERC) rules for co-located large loads, and commodity hedges creating earnings volatility. Management expects further PJM clarity around November and a FERC order in the first to second quarter of next year.

"The industry is in the early stages of a multi-decade super cycle of growth," said Rex Geveden, CEO of BWX Technologies.

Rex Geveden, CEO at BWX Technologies

For BWX Technologies, CEO Rex Geveden emphasized that there is "a credible opportunity to secure at least one new build nuclear equipment order before the year end." The pipeline includes AP1000s, BWRX-300 reactors, TerraPower, and Rolls-Royce, plus government-backed programs tied to a $17.5 billion Department of Energy Energy Dominance Financing Office loan commitment. Risks include federal budget uncertainty, qualified-tradesperson shortages, and heavy near-term capital expenditures potentially approaching 7% of sales in future years.

Rex Geveden

GE Vernova has seen its share price climb nearly 36% year-to-date, though it has cooled 8.43% over the past month amid headlines around competitive threats. The company's Q2 bookings hit $24.2 billion, an 88% increase year-over-year, with gas power gigawatts under contract rising from 100 to 116 gigawatts sequentially. Management targets at least 125 gigawatts under contract by the end of the year. Electrification equipment backlog is $41 billion, up 69%, and the company secured two more tech selects and early work agreements for small modular reactors (SMRs) in the U.S. during the quarter. Q2 free cash flow reached $5.1 billion, and full-year 2026 free-cash-flow guidance was raised to $11.5 to $12.5 billion. The primary risk is the Wind segment, guided to roughly $400 million of EBITDA losses in 2026, plus tariff and supply chain risk.

The nuclear renaissance is no longer theoretical. It's being priced in real time as hyperscalers compete for firm, carbon-free baseload power. Every dollar of AI capital expenditure eventually has to plug into a wall, and by the time hyperscalers have signed off on GPU orders, the binding constraint stops being silicon and starts being substations, transformers, and the power plants that feed them. The three companies capturing this shift operate at different rungs of the ladder, each with distinct cash flow profiles and risk profiles. Understanding which rung matters most to your investment thesis is the key to avoiding mispricings in what may be the most significant infrastructure buildout of the next decade.