Constellation Just Sold Google a 20-Year Power Floor — and CEG's Stock Already Front-Ran the Trade

Headline capacity is 890 MW of new nuclear output — produced through uprates at 11 Constellation reactors, with the first increments expected by 2028 and full delivery before the end of 2032. Per the joint announcement, the 20-year PPA also bundles demand-response capabilities into the pricing, letting Constellation curtail non-critical load during high-stress grid events and effectively monetize the flexibility premium Google is willing to pay for firm power. The accompanying 15-year agreement covers a further 2,700 MW across Constellation's broader PJM fleet, bringing the total package to 3,590 MW — roughly 16% of Constellation's ~22,000 MW nuclear fleet.

Heading into the deal, Constellation Energy was the worst-performing large-cap in the AI-power complex — down 24% year-to-date per pre-market coverage, sitting at a 50-day moving average of $271.47 versus a 200-day of $287.79, with a 52-week range of $228.63–$412.70. The compression came from a mix of soft Q2 2026 earnings (revenue of $7.506B, EPS of $1.45 vs. $11.12B and $4.49 in Q1), FERC rate-case uncertainty, and a market that was treating CEG as a regulated utility caught between capex and rate-base recovery. The Google deal reset the frame: an investment-grade counterparty with multi-decade demand is exactly the cash-flow visibility a rate-cycle utility does not have.

The post-jump setup: Constellation Energy traded at $267.62 against a consensus 12-month analyst target of $341.53 (19 covering analysts: 6 Strong Buy, 13 Buy, 3 Hold, 0 Sell), implying ~28% upside even after the rally. Trailing P/E is 25.2x on TTM EPS of $10.22, EV/EBITDA is 15.2x, and net debt has surged to $23.62B at June 30, 2026 — up from $5.24B at year-end 2025 — reflecting the closing of the Calpine acquisition, which roughly doubled total assets to $98.25B. The deal matters precisely because it locks cash flow against that balance sheet.

Across the public tracker of hyperscaler nuclear deals, 13 transactions committing 9.8 GW of capacity have been signed through mid-2026 — a meaningful shift from one-off contracts to multi-decade anchors. The Google-Constellation agreement is the second announced in 2026 alone that runs the full 20-year term, joining the September 30 Amazon-Calvert Cliffs agreement (690 MW, $3B Maryland investment) and lining up behind the Microsoft-Constellation Three Mile Island restart (835 MW, 20-year PPA, targeted 2027) and the Meta-Constellation Clinton Clean Energy Center deal (1,121 MW, June 2025).

The pattern matters more than any single deal: every hyperscaler that has signed nuclear PPAs in 2025–2026 has either gone 17–20 years or contracted firm, 24/7 output. That is structurally different from the renewable PPAs that have dominated corporate clean-energy procurement for a decade, which typically run 10–15 years and match intermittent wind or solar. Investors who were treating nuclear PPAs as renewable-equity equivalents were mispricing duration risk — and CEG's compression through 2026 reflects that mispricing.

Constellation Energy is now the only operator with four hyperscaler nuclear contracts — Microsoft (TMI), Meta (Clinton), Amazon (Calvert Cliffs), and Alphabet (multi-state uprates). The combined contracted envelope is ~4,500 MW of incremental or preserved output, plus additional energy-supply agreements on top. Strategically that creates three compounding advantages: (1) regulatory optionality, because each anchor de-risks license renewals and relicensing at multiple sites; (2) supply-chain leverage, because the same uprate engineering firm, turbine supplier, and fuel cycle can be amortized across more reactors; and (3) commercial pricing, because each new anchor raises the marginal rate Google, Meta, Amazon, and Microsoft have implicitly accepted.

The Google announcement also includes a Google Cloud Gemini Enterprise deployment to help Constellation optimize plant operations, accelerate uprate planning, and streamline construction — the first time a hyperscaler is bringing software stack directly onto a nuclear operator's production environment. That signals hyperscalers are moving from \"offtake buyer\" to \"operational partner,\" which lengthens the strategic moat beyond the contractual term.

The pre-market tape tells the real story: Vistra traded up ~9% and Talen Energy ~11% on the Google announcement, not because either company is in the headline deal but because the deal validates the entire nuclear-data-center thesis. The peer multiples have not yet converged — Vistra trades at 23.6x trailing P/E and 10.4x EV/EBITDA on $19.21B of TTM revenue with a $48.85B market cap, while Constellation Energy is at 25.2x P/E and 15.2x EV/EBITDA on $31.27B TTM revenue with a $96.1B market cap. Talen Energy is a different bet: $16.42B market cap, $3.74B revenue, 10.7 GW of mostly gas-adjacent and nuclear capacity, and a forward P/E of 9.9x on analyst earnings expectations. The Google-Constellation deal pulls the whole curve higher by re-anchoring what a nuclear PPA is worth.

Constellation operates ~22,000 MW of nuclear — but that capacity runs on fuel and parts from a much smaller upstream ecosystem. The Google deal's 11-reactor uprate program puts incremental pressure on a fuel-supply chain that is already capacity-constrained: Cameco controls the largest share of Western uranium production, with $3.47B of TTM revenue and a market cap of $38.1B; Centrus Energy holds the only operational U.S. uranium-enrichment capability (HALEU for advanced reactors) and trades at $2.68B market cap; Oklo sits on the advanced-reactor side with the Aurora microreactor design and a $6.73B market cap. NuScale Power and Nano Nuclear Energy cover the SMR / microreactor pipeline that hyperscalers are also funding.

The transmission is concrete: each incremental MW of uprated nuclear capacity requires additional low-enriched uranium and enrichment services, but the leverage is asymmetric. Centrus Energy has been the most volatile beneficiary in 2025–26 (52-week range $135.75–$464.25) because its HALEU franchise is the supply bottleneck for advanced reactors. Cameco sits closer to spot-uranium exposure (52-week range $77.55–$135.24) and benefits more slowly as inventories are drawn down. Oklo is the speculative upside — the Aurora-INL project is targeted for 2027 with Meta's 1.2 GW commitment, but is still pre-revenue.

Short-term (days–quarters), three things move first: (1) the next round of hyperscaler nuclear announcements, with Meta and Alphabet both publicly working through additional SMR pipelines; (2) FERC rate-case clarity on Constellation Energy's post-Calpine rate base; and (3) the Q3 2026 earnings cycle, where the company's ability to frame the Google PPA as accretive to 2027 EBITDA will determine whether the post-jump hold-up is durable. The 2,700 MW energy-supply agreement is the under-discussed piece — pricing on that tranche has not been disclosed and represents the most plausible source of upside revisions in 2027.

Long-term (1–3 years), the structural shift is what matters: hyperscalers are locking 17–25 year nuclear capacity as core infrastructure spend, not as corporate clean-energy marketing. That changes how the universe of Vistra, Talen Energy, and the Oklo / NuScale / Nano Nuclear advanced-reactor pipeline will be valued — incumbent nuclear gets re-priced toward infrastructure-like multiples, while merchant-power exposure gets compressed. The risks are equally concrete: PJM capacity pricing volatility (already evident in the 2024–25 reform debate), NRC license-renewal delays at aging plants, and the political risk that a future administration rolls back DOE loan guarantees that have underwritten both Crane and Calvert Cliffs uprates.

Putting it together: the Google-Constellation deal is the cleanest evidence yet that hyperscaler demand has moved from short-cycle procurement to multi-decadal infrastructure. Constellation Energy is the highest-conviction expression of that thesis because it is the only operator with four hyperscaler anchors on one nuclear fleet; Vistra is the next-best expression through its gas-plus-nuclear flexibility in PJM and ERCOT; Talen Energy is the highest-beta expression through its concentrated Susquehanna exposure to Amazon. Upstream, Cameco and Centrus Energy capture the fuel-cycle leverage, while Oklo and NuScale Power capture the optionality on advanced reactors. For investors, the trade that just printed in CEG is the beginning — the rest of the curve catches up over the next two earnings cycles.