The call: HOLD Constellation (CEG ~$305, +14% intraday). ADD pullback $282–$295 toward $368, stretch $412. Invalidation: daily close below $250. 6–12 months. HIGH conviction.
Why now: Google signed a 20-year PPA funding 890 MW of nuclear uprates at 11 CEG units plus a 15-year 2,700 MW supply deal — the lowest-risk megawatts in AI power.
The disagreement: Market treats this as one more hyperscaler press release. It is the moment CEG's uprate inventory goes scarce — spokesman says few facilities left to tap.
The level that changes everything: PJM's Bring-Your-Own-Power tariff (FERC decision window, requested Oct 12 effective date). Approval turns every data-center PPA into a CEG comp.
Google and Constellation announced Tuesday, Oct 6 a 20-year power purchase agreement enabling 890 MW of new nuclear capacity via uprates at 11 CEG units in Illinois, Pennsylvania and New Jersey, on the PJM grid. Constellation invests more than $4.3 billion; first uprate delivers by 2028, all 890 MW before end-2032. A separate 15-year supply agreement covers 2,700 MW of existing PJM fleet output, plus a 5-year Google Cloud / Gemini Enterprise alliance for an "AI for Energy" blueprint.
The market's verdict was instant: CEG $305.22, +14.0% intraday (open $291.12, high $309.48, prior close $267.62), top of the S&P 500 on a record day (SPY $781.16, +0.8%). Vistra (VST) rode the draft at $160.98, +11.1%. Google itself was flat at $347.82 — the buyer pays nothing today; the seller reprices.
Read the primary source and the shape is clear. This is not a restart (Three Mile Island / Crane), not an SMR option (Kairos / Elementl), not a paper offtake. It is anchor-customer financing for extra output at plants that already run — turbines, steam generators, digital controls — inside existing licenses, sites, workforces and interconnections.
Bernstein's line Tuesday — existing generation and new uprates are the "cheapest, cleanest and firmest" additions — is the whole thesis in one sentence. My opinion: they are right, and the arithmetic favors CEG over every SMR developer.
890 MW across 11 units averages ~81 MW per unit. At a 93% fleet capacity factor (CEG's Q2 print), that is ~7.25 TWh per year of firm, 24/7 carbon-free output — the equivalent of a large new reactor or several SMRs, without a greenfield interconnection queue. The $4.3 billion price tag implies ~$4,800/kW. A new AP1000 in the US has cleared $10,000–$15,000/kW with a decade-plus timeline. Even allowing for NRC review per unit and outage-window staggering (hence 2028→2032 phasing), the uprate delivers this decade at roughly half the capital intensity of new build.
No PPA price was disclosed — do not divide $4.3 billion by megawatt-hours and call it a power price; that is Constellation's capex, not Google's payment. Illustratively, 7.25 TWh at $85–$100/MWh is $616–$725M of annual revenue once fully online. At a mid-40s EBITDA margin that is ~$275–$325M of incremental EBITDA, or roughly $0.60–$0.80 of EPS power on ~340M diluted shares — before the 2,700 MW supply leg, which is pure revenue-duration extension on already-running assets. That is why the stock moved 14%: duration, not next quarter.
Buried in the Bloomberg write-up is the line that matters most for valuation: Constellation spokesman Paul Adams said Tuesday the company now has few additional facilities that could be tapped for similar agreements, after this deal plus last week's Amazon 690 MW / Calvert Cliffs agreement and June's Walmart deal.
This flips CEG from a pipeline story to a scarcity story. If the uprate shelf is nearly full, each remaining megawatt reprices higher, and the 2,700 MW supply agreement becomes the template for monetizing the rest of the 55 GW fleet (largest private power producer in the US, ~10% of national clean energy). My opinion: consensus still models CEG as a merchant nuclear operator with data-center optionality. It should now be modeled as the Federal Reserve of the electron in PJM — the only counterparty that can deliver 500+ MW of firm clean power before 2030.
PJM context seals it. The grid serves 67 million people and is the epicenter of the load-growth fight. PJM's Bring-Your-Own-Power / Interim Resource Adequacy proposal — large loads bring new supply or face curtailment, with an Oct 12, 2026 requested effective date — makes Google's structure the compliance blueprint. Google's pledge that other customers "bear none of the associated costs" is aimed directly at FERC and state regulators. Approval generalizes the CEG comp to every hyperscaler load in PJM.
CEG entered today beaten down: -25% over the prior 12 months vs S&P +17%, 52-week range $228.63 (July 1, 2026) to $412.70 (Oct 15, 2025). Fundamentals had already turned — Q1 revenue doubled on Calpine (closed Jan 7, 2026), full-year 2026 adjusted operating guidance raised to $11.50–$12.50, Q2 beat on Calpine accretion plus firmer PJM capacity prices, 93% nuclear capacity factor — but the stock sat at $267 pre-deal, or ~22× the $12 midpoint.
At $305 that is ~25× 2026E and a 27.4× TTM multiple (EPS TTM $10.26, ROE 14.7%, revenue growth +26% y/y, beta 1.14, yield 0.9%). Expensive against regulated utilities at 18–20×, cheap against the contracted growth it just locked: 890 MW of 20-year paper plus 2,700 MW of 15-year paper, $4.3B of rate-base-like capex funded by a AAA credit, and a Gemini software tie-in carried at zero megawatts. $368 — my base target, the pre-selloff shelf — is 21% above here and 30× 2027E if CEG earns ~$12.30 on capacity plus Calpine synergies. Stretch $412 retakes the 2025 high if FERC blesses BYOP and a third hyperscaler signs the last uprate shelf.
| Flagship call | Level |
|---|---|
| Ticker / action | CEG — HOLD into strength; ADD pullback $282–$295 |
| Target | $368 base; $412 stretch (2025 high) |
| Invalidation | Daily close below $250 (pre-breakout shelf) |
| Horizon / conviction | 6–12 months / HIGH (contracted duration + scarcity + regulatory tailwind) |
| Audience / size | Core equity income-growth sleeve; half-size on chase, full on pullback |
Do not chase the +14% spike with a full ticket. The premarket was $276; the open was $301. Intraday momentum can overshoot to $315–$320, then fade as arb funds sell the headline. The add zone $282–$295 is the breakout retest. This updates — not replaces — open call EQ-CEG-1 (HOLD, ADD $262–$285, tgt $368, kill <$250): the lower edge of that zone is stale after today; roll adds up to $282–$295.
Bull 25% ($412+): FERC approves PJM BYOP in October, a third hyperscaler takes the last uprate shelf at a higher $/MWh, and 2027 guidance embeds contracted escalators. CEG retakes its $412.70 high and screens as a growth utility at 30×+.
Base 50% ($340–$368): 890 MW funds on schedule, first electrons 2028, 2,700 MW supply leg extends duration, capacity prices hold. Stock grinds to $368 as 2027 estimates rise to $12–$13. Pullback entries work; chase entries tread water for weeks.
Bear 25% (back to $250): NRC uprate reviews slip, capex overruns past $4.3B without PPA escalators, or PJM BYOP stalls and data-center backlash reprices load growth. A daily close below $250 breaks the thesis — the market is saying the contracts are value-destructive.
Load-bearing assumption: the undisclosed PPA price covers Constellation's ~$4,800/kW plus a utility-like return. If the $/MWh is a concession to win Google's cloud alliance (software bundled for electrons), incremental ROE collapses and my $368 is $40 too high. Watch the 8-K when filed — price redaction vs. escalator language is the tell.
One measurable condition: a daily close below $250, or Constellation filing an 8-K / 10-Q that shows uprate capex funded without matching PPA escalators (margin dilution on the $4.3B). Either breaks scarcity-pricing. A slower signal: FERC rejects or guts PJM's BYOP framework — then Google's blueprint has no regulatory multiplier and VST/TLEN comps fade.
Downstream consequences run three ways. First, VST is the sympathy long with a cleaner entry — $160.98 after its own $4.2B DOE loan headline for Pennsylvania/Ohio nuclear life-extensions; open call EQ-VST-1 (BUY $140–$152, tgt $210–$225, kill <$132.66) stays live, and today's +11% confirms the sector bid. Second, utilities as a duration hedge are breaking: XLU +2.4% today while the 10Y sits above 5.3% — if BYOP passes, IPPs (CEG, VST, TLN) take share from regulated wires; keep AEP ($120.79, HOLD/ADD $118–$125) but do not add XLU here. Third, Google de-risks its own AI capex story — 890 MW here plus Finland's Loviisa plus Duane Arnold plus Southern uprates totals a multi-gigawatt firm-power moat; that supports the mag-7 multiple even as ad-tech litigation ($3.2B headline) lingers.
No contradiction with open calls: CEG, VST, AEP and XLE longs all lean the same way — electrons over duration. The live conflict to respect is TACT-SPX-FADE-1 (fade 7,730–7,780 toward 7,620): CEG adds are stock-specific, sized half until the index fade resolves.