$CEG EXECUTIVE CALL SUMMARY: CONSTELLATION ENERGY (03/31/26)
The March 31 call was far more important as a strategic reframing event than as a conventional quarter-only earnings discussion. The company used the session to move the narrative away from near-term binary data-center contract timing and toward a more modelable long-duration earnings framework built around 2026 adjusted operating EPS guidance of $11.00-$12.00, 20%+ base EPS growth through 2029, a $5.0B buyback authorization, and $3.9B of growth capital deployment. The central message was that substantial upside remains available from AI, contracting, commercial optimization, and capital allocation, but that the baseline itself is now strong enough to support the equity story without requiring immediate new hyperscaler announcements.
The near-term market impact was negative because the call failed on the one catalyst that appeared most anticipated by the market. Dominguez stated, “we’re not going to be announcing a new data economy deal today,” and also acknowledged that management had previously expected to have “an important transaction” done by this call. At the same time, the 2026 EPS midpoint of $11.50 came in slightly below the $11.6 LSEG consensus cited by Reuters. Reuters reported a 2.2% premarket decline, and by 13:50 UTC the shares were down 5.2% intraday. The stock reaction indicates that the market treated the absence of a new AI/data-center deal as more important than the long-term framework improvements.
WHAT THE CALL CHANGED
The call changed 3 things materially. First, it formalized a new earnings architecture. Pages 21-23 of the presentation divide earnings into “base” and “enhanced,” with base earnings defined as more visible, easier to calculate, and supported by long-term contracts, PTC-supported nuclear output, minimum expected fossil earnings, and average commercial margins. Enhanced earnings remain material and reflect upside from power prices, commercial outperformance, and volatility capture. Second, the call clarified that the published 20%+ base EPS CAGR excludes buyback accretion and leaves 2028-2029 free cash undeployed in the baseline. Third, it explicitly linked future AI/data-center contracts to a more complex regulatory and political environment in PJM and at the federal level, making deal timing slower but potentially making Constellation’s combined nuclear-plus-gas-plus-storage platform more valuable.
One of the strongest features of the call was disclosure quality. Pages 30-32 of the presentation provide a bottoms-up modeling toolkit across contracted clean, available nuclear, natural gas and oil, wind/solar/hydro, capacity, commercial margin, O&M, interest expense, and tax assumptions. For a merchant and quasi-contracted power business, that degree of transparency is unusual and appears deliberately designed to reduce the valuation discount typically applied to earnings streams that investors perceive as hard to underwrite. The company is effectively inviting the market to independently rebuild the base case and decide how conservative it is.
QUARTERLY PERFORMANCE VERSUS HISTORY
The underlying quarter was mixed. Fourth-quarter 2025 adjusted operating EPS was $2.30 versus $2.44 in 4Q24, a modest year-over-year decline, while GAAP EPS fell to $1.38 from $2.71. Operating revenue rose to $6.074B from $5.382B, but operating income fell to $598M from $972M. Management attributed the adjusted EPS decline primarily to “unfavorable nuclear PTC portfolio results,” partly offset by favorable market and portfolio conditions. That mix matters because the quarter’s weakness came more from portfolio monetization and mark-to-market type effects than from a thesis-breaking deterioration in fleet value.
The full-year picture was stronger on an adjusted basis and weaker on a GAAP basis. Full-year 2025 adjusted operating EPS rose to $9.39 from $8.67 in 2024, while GAAP EPS fell to $7.40 from $11.89. Operating revenue increased to $25.533B from $23.568B, but operating income declined to $3.086B from $4.352B. The reconciliation shows that unrealized fair value adjustments were a $2.26/share drag in 2025 versus a $3.25/share benefit in 2024, which explains much of the GAAP deterioration. That divergence reinforces why management is trying to get investors to focus on adjusted and especially base earnings rather than raw GAAP volatility.
Operationally, the quarter was not pristine, but the full-year record remained excellent. In 4Q25, nuclear generation was 45,459 GWh, effectively flat with 45,494 GWh in 4Q24, but ownership capacity factor slipped to 93.1% from 94.8% because non-refueling outage days rose to 30 from 3, even though planned refueling outage days improved to 63 from 66. The more important full-year data were stronger: annual nuclear capacity factor was 94.7% versus 94.6% in 2024, and the presentation shows that the fleet still outperformed the industry’s 90.5% capacity factor benchmark and 38 average refueling outage days by a wide margin. Gas and pumped-storage dispatch match rate improved to 99.4% in 4Q25 from 93.2% a year earlier, while renewable capture improved to 97.2% from 95.7%. The quarter therefore showed nuclear noise, not a deterioration in broad fleet quality.
GUIDANCE, BASE EARNINGS, AND COMPARISON WITH PRIOR FRAMES
The most important quantitative update was the 2026 adjusted operating EPS range of $11.00-$12.00. Against 2025 actual adjusted operating EPS of $9.39, that implies a 17.1%-28.0% year-over-year increase, or 22.5% at the midpoint. That is strong in absolute terms. More importantly, it came despite management explicitly citing 2 acquisition-related headwinds: more divestitures than expected under the DOJ/FERC remedy and higher depreciation from purchase accounting after marking Calpine’s assets to fair value at close.
Relative to prior guidance, the message was better than the stock reaction suggested. In the January 2025 Calpine acquisition deck, Constellation showed 2026 standalone guidance of $8.70-$9.70, with base earnings of $5.95-$6.05 and “at least $2.00” of annual EPS accretion from Calpine, which implied at least $10.70-$11.70 of combined 2026 earnings power. The current $11.00-$12.00 range is therefore slightly better, even after absorbing the larger-than-expected remedy package. The March 18 agreement to sell approximately 4.4 GW of predominantly gas-fired PJM generation to LS Power for $5B confirms that the remedy package was substantial, not cosmetic.
The other major guidance change was philosophical. In January 2025, management framed the post-Calpine long-term outlook as at least 13% base EPS growth from 2024-2030, with enhanced earnings 20%-25% on top of base and at least $2/share of annual accretion from Calpine. The March 2026 framework is more front-end loaded and more explicit: 20%+ base EPS CAGR from 2026-2029, then 10%+ rolling 3-year base EPS growth thereafter. The periods are not identical, so the comparison is not perfectly apples-to-apples, but the new framework clearly raises the visibility and near-term steepness of the compounding profile.
The risk that remains underappreciated is earnings mix. Page 21 of the presentation explicitly says base earnings are 60%-70% of total and enhanced earnings are 30%-40% of total. Page 22 then shows 2026 base earnings of $6.65-$6.75, with enhanced earnings still comprising about 40% of total. By 2029, base rises to $11.40-$11.90, but enhanced earnings still represent 30%-35% of total. The valuation implication is clear: the call materially improved earnings visibility, but it did not transform Constellation into a fully contracted utility-like earnings stream. A large fraction of the earnings stack remains exposed to power prices, spark spreads, commercial margin outcomes, and volatility capture.
At the same time, the Q&A made the published framework look more conservative and therefore more credible. Management explicitly said, “There is no benefit in the 20% base EPS CAGR from capital allocation for the share repurchase,” and said there is “nothing planned with regard to accretion” from 2028-2029 free cash in the baseline, beyond earning interest income. Management also confirmed that 2027 share count is held flat for modeling purposes. This means the reported CAGR is not using financial engineering to hit the target. It is an operating baseline with undeployed capital optionality layered on top.
A further positive nuance is that the 2026 disclosures include earnings from assets to be divested in 2H26, which means 2027 base growth is being shown after the remedy headwind has largely rolled through. That makes the 2026-to-2027 base step-up more impressive than the market may initially assume.
DATA CENTER, PJM, AND REGULATORY IMPLICATIONS
The single biggest near-term disappointment was the absence of a new nuclear or major hyperscaler deal. That matters because management itself had primed the market for one. Dominguez acknowledged that he had previously indicated the company expected to be done with an important transaction by this call, then said, “we’re not going to be announcing a new data economy deal today.” The reasons cited were not demand collapse. They were process and policy complexity: more scrutiny on data-center development and the need for stakeholders to be aligned, plus the March 4 White House Ratepayer Protection Pledge, which management said forced renegotiation of some PPA terms.
That White House pledge is central to the call’s real meaning. The proclamation says hyperscalers and AI companies must “build, bring, or buy” the generation needed for their data centers, must pay for all new delivery infrastructure upgrades, must negotiate separate rate structures, and must pay those rates whether or not the electricity is ultimately used. FERC’s December 18 PJM order had already moved in a similar direction by telling PJM to create clear rules for AI-driven and other large co-located loads, and FERC’s large-load ANOPR is specifically seeking comment on whether flexible or curtailable large loads should move through interconnection faster and pay the full cost of upgrades. The implication is that future Constellation AI contracts are likely to be broader, more infrastructure-inclusive, and more negotiated across energy, capacity, and interconnection than the market may have expected 6 months ago.
Management’s response was not retrenchment but reconfiguration. Dominguez said “regulatory clarity helps deals get done,” but also said Constellation is “not waiting on regulatory clarity or certainty.” The company described multiple structures that can work now: pairing nuclear with batteries, demand response, uprates, or gas-fired generation; letting customers buy a PJM backstop capacity product while buying energy and clean attributes from Constellation; or relying on customer-side curtailment and AI-enabled load shifting to manage peak obligations. This is where the Calpine transaction matters strategically. It converts Constellation from a pure clean baseload seller into a full-stack reliability and structuring platform, which is arguably more valuable under the new White House/FERC framework than under the older and simpler co-location narrative.
The pipeline itself still looks real beneath the missing headline. The presentation shows that Constellation has executed 5,650+ MW of long-term clean energy deals, ~880 MW of contracted storage, ~2,300 MW of long-term agreements at natural gas plants, and >1,100 MW of data-center agreements under contract at Thad Hill and Freestone. The February 24 earnings release added further evidence by disclosing a new 380 MW CyrusOne agreement at Freestone plus an exclusive arrangement for another 380 MW phase, on top of previously announced 400 MW Thad Hill agreements. In clean baseload, over the last 12 months the company increased 2030 contracted output under long-term agreements from 12M MWh to 48M MWh, yet still has 147M MWh available for contracting. The problem is therefore timing and structure, not a lack of commercial inventory.
The call also made a broader policy argument that is favorable to the thesis if it becomes accepted. Management pointed to PJM’s load-duration curve and the Brattle analysis in the presentation showing large amounts of idle generation and wire capacity for most hours of the year, arguing that efficiently integrated data-center load can lower, not raise, bills for incumbent customers by improving system utilization. That is strategically important because the political narrative around data centers had been drifting toward “cost causer.” Constellation is clearly trying to reframe them as “cost reducer,” provided peak obligations are properly managed. If that reframing takes hold with policymakers, it would materially improve the durability of the data-center contracting opportunity.
ASSET QUALITY, OPERATIONS, AND CAPITAL ALLOCATION
The call’s strongest long-term argument remains asset scarcity. Constellation now controls roughly 55 GW of nuclear, gas, geothermal, hydro, wind, solar, and storage capacity, and management argued that replacement cost would exceed 3x current enterprise value. The portfolio is geographically diversified, with strong positions in PJM, ERCOT, CAISO, and NYISO, and includes the largest U.S. nuclear fleet, the largest U.S. natural gas fleet in the combined company, and the largest geothermal fleet. The presentation also argues that the company is the largest private-sector power producer in the world, producing 298M MWh, and the lowest-carbon-intensity major U.S. generator. In an environment where replacement gas capacity is expensive, new nuclear remains uncertain, and interconnection queues are long, incumbent dispatchable assets with fuel diversity, transmission access, and permitting history deserve a premium.
The gas portfolio may be the most underappreciated part of the story. Management said around 80% of the ~27 GW natural gas fleet is efficient CCGT or cogeneration capacity, and the presentation argues those assets are worth roughly $65B at replacement cost. Yet the fleet is still underutilized. The presentation’s illustrative ERCOT view says CCGTs had some excess capacity about 90% of the time in 2025 and could be fully utilized about 80% of the time by 2030 under higher load growth. The hedge structure is equally attractive. Page 34 shows 2026 CCGT output is 20% contracted offtake and 80% hedged, leaving virtually no open downside; open exposure then rises to 10% in 2027, 20% in 2028, and 45% in 2029. Meanwhile, page 35 shows only about 28%-33% of natural gas/other energy base gross margin is currently attributable to contracts, which means much of the gas earnings stack still rests on average spark-spread assumptions and can be upgraded through long-term agreements. That asymmetry is favorable: downside is largely protected near term, while repricing upside builds over time.
The capital allocation framework was the other major source of tangible support. The 2026-2027 bridge shows $13.6B of available cash composed of $0.8B of starting cash net of Calpine cash uses, $8.4B of cumulative FCFbG, and $4.5B of after-tax asset sale proceeds. Uses are mapped to $3.9B of identified growth capital, $1.3B of dividends, $3.4B of deleveraging, and $5.0B of authorized share repurchases. This is not a vague authorization. It is directly embedded in a funding bridge. The company also maintained a 10% dividend growth target and emphasized that growth capital must meet a double-digit unlevered return threshold.
The balance-sheet evidence was also unusually concrete. In January 2026 Constellation issued $2.75B of senior notes, including $800M of 40-year unsecured notes at a sub-6% coupon. The presentation says more than 95% of Calpine’s corporate debt has already been retired or exchanged into Constellation debt. Moody’s and S&P reaffirmed Baa1/BBB+, Calpine’s ratings were upgraded to investment grade, and the presentation says Moody’s lowered its downgrade threshold to 25% FFO/debt from 30% for the Baa1 rating. Management also said 2x debt/EBITDA is a reasonable long-run leverage anchor. At the current intraday market cap of about $102.8B, the $5.0B repurchase authorization equals roughly 4.9% of equity value. At the current share price, it would equate to about 17.7M shares, or roughly 4.9% of the 361M diluted share count used in 2026 guidance. None of that buyback accretion is included in the published 20% base EPS CAGR.
One of the more important tempering disclosures was around new nuclear. Dominguez said, “I am not yet at a confidence level where I could say to you that we are committed on a path to new nuclear.” That removes one speculative leg of the bull case in the near term. The growth stack for the next several years is Crane, relicensing, uprates, batteries, demand response, gas utilization, and contracting, not a greenfield SMR decision. That is strategically sensible because it prioritizes nearer-dated, higher-certainty returns, but it also means that the call did not validate any near-term step change from new-build nuclear.
INVESTMENT IMPLICATIONS
The short-term implication is negative to mixed. The stock had clearly been positioned for a data-center announcement and instead got a base-case framework presentation, a modest consensus miss on 2026 EPS, and a management acknowledgment that an expected transaction was not ready. Until a major nuclear PPA, co-location agreement, or gas-powered land deal is actually signed, the shares are likely to trade more on contract cadence and PJM/FERC headlines than on the newly disclosed base earnings math. The $5.0B buyback authorization and the clearer floor under earnings should limit downside, but the call by itself was unlikely to create immediate upside.
The medium-term implication is more constructive. Management quantified a large menu of 2029 base EPS opportunities on page 23: a 1 GW nuclear PPA at a $20-$50/MWh premium to PTC floor adds $0.40-$1.00 per share; a 1 GW natural-gas-powered land agreement adds $0.20-$0.50; a 1%-2% gas capacity factor increase adds $0.10-$0.20; a commercial-margin improvement to average adds $0.10-$0.30; 3% versus 2% PTC inflation adds $0.30; and capital allocation adds $0.50+. Management notes these are not additive, but the sizing is still important because only partial execution is needed to create meaningful upside above the published base case. The call therefore improved the asymmetry of the long-term thesis even as it weakened the near-term catalyst calendar.
The longer-term implication is that Constellation is increasingly becoming a hybrid asset-scarcity, policy-floor, and commercial-optionality story. The federal nuclear PTC remains in force through 12/31/32 and, as shown in the appendix, inflation mechanically lifts the floor over time. The company’s base case assumes only 2% inflation, but the appendix shows that 3.5% inflation would raise the 2031 PTC price to $56/MWh and would be worth about $1.55/share of base EPS accretion in that year. New York’s ZEC extension preserves more than 3,000 MW through 2050. DOE is backing Crane with up to a $1B loan guarantee. FERC is actively designing a more explicit regime for co-located and flexible large loads. That policy mosaic materially reduces downside. The remaining question is execution. The most important signposts are closure of the remaining divestitures, at least 1 incremental large data-center deal, proof that Calpine integration expands commercial margins, continued progress at Crane, and evidence that the 147M open nuclear MWhs can be monetized materially above the PTC floor.
Netting the positives and negatives together, the call was strategically stronger than tactically stronger. Strategically, it improved transparency, formalized a very attractive compounding algorithm, highlighted substantial undeployed free-cash-flow optionality, and reinforced the uniqueness of the combined nuclear-plus-gas platform. Tactically, it delayed the most visible AI catalyst, left the stock vulnerable to regulatory headline risk, and reminded the market that roughly 30%-40% of earnings still sit in enhanced rather than fully contracted buckets. The appropriate interpretation is not that the thesis weakened. It is that the timing of monetization became less immediate and the structure of monetization became more complex, while the floor under long-term value became more explicit.
COMPANY EMPLOYEES ON THE CALL
• Joseph Dominguez, President and Chief Executive Officer
• Shane Smith, Executive Vice President and Chief Financial Officer
• Tim Flottemesch, Vice President, Investor Relations
• Other members of Constellation’s senior management team were said to be present for Q&A, but were not individually identified.
RESEARCH ANALYSTS ON THE CALL
• Angie Storozynski, Seaport Global
• David Arcaro, Morgan Stanley
• James West, Melius Research
• Julien Dumoulin-Smith, Jefferies
• Steve Fleishman, Wolfe Research
