Talen Energy
Owns and operates a ~10.7 GW generation fleet (Susquehanna nuclear core) and earns money three ways: (1) long-term contracted PPAs to hyperscalers (AWS, $18B through 2042), (2) PJM/ERCOT merchant energy + spark spreads, and (3) PJM capacity-market revenue. Capital-intensive, asset-heavy; thesis is contracting scarce carbon-free baseload to AI data centers.
The thesis on this name
State of Data-Center Power
AVOID / least-preferred of the IPP trio rather than a clean short. Talen has a real 2.2GW Susquehanna nuclear anchor (Q1-2026 adj EBITDA $473M, adj FCF $350M, fact), but it has faced stiffer FERC headwinds on its behind-the-meter datacenter interconnection (the amazon co-location precedent) than Constellation or Vistra, and attempted a $3.45B gas acquisition to keep pace (fact). The market prices the full datacenter-PPA optionality while the regulatory path to monetizing behind-the-meter nuclear is the most contested of the three. Prefer CEG (premier fleet/PPAs) or VST (cheaper, gas+nuclear) for the demand-anchor; Talen carries the most regulatory tail-risk per unit of upside.
State of Data-Center Power
The Susquehanna-nuclear/Amazon PPA story ($18B through 2042) — purest behind-the-meter colocation call, but valuation is hot and binary on FERC.
State of Data-Center Power
$18B Amazon PPA, purest behind-the-meter nuclear story; small clip on single-asset + FERC binary risk.
Earnings, margins, COGS & capex
FY2025 revenue grew 22% to $2,581M and Adjusted EBITDA reached $1,035M, but GAAP showed a $(219)M net loss on derivative mark-to-market and refinancing charges (fact, FY2025). Momentum inflected hard in Q1 2026: Adjusted EBITDA more than doubled to $473M and Adjusted FCF quadrupled to $350M YoY as the Freedom/Guernsey gas plants (closed Q4 2025), higher spark spreads, June-2025 capacity/RoR revenue, and the AWS PPA ramp all landed (fact). The June-2026 close of the $3.45B Cornerstone gas deal (2.6 GW) adds >15% to cash flow per share and pushes the line of sight to >$40/sh FCF by 2028 (fact). The model is a leveraged bet on PJM tightness: rich when capacity/power prices stay high, exposed if they revert.
Income statement — where each revenue dollar goes
% of revenueOf every $1 of revenue, ~0¢ is cost of goods and ~0¢ operating expense, leaving ~100¢ of operating profit (~2025¢ net).
Revenue trend
Margins
up — Q1 2026 run-rate sharply higher ($473M)
up — Q1 2026 turned positive (+$63M)
up sharply (quadrupled YoY in Q1)
mixed
COGS structure
COGS for an IPP is fuel + purchased power + plant O&M + nuclear fuel amortization. Nuclear (Susquehanna) has very low marginal fuel cost — the spread vs. PJM power price is the profit engine; the gas fleet's COGS is natural-gas cost, so spark spread (power price minus gas-implied cost) drives gas-plant margin. Capacity revenue and the contracted AWS PPA carry high incremental margin. Mark-to-market on hedges/derivatives swings GAAP COGS/earnings quarter to quarter.
Capex
Capex funds nuclear fuel reloads, the Susquehanna refueling outage (elevated 2025) and uprate studies, gas-fleet maintenance, the AWS colocation/grid reconfiguration (executed during the spring-2026 Susquehanna outage), and growth options (up to 300 MW digital-infrastructure/crypto campus; SMR exploration with Amazon). Company frames near-term capex in the hundreds of millions; the dominant capital event of 2025-26 is M&A ($3.8B Freedom/Guernsey + $3.45B Cornerstone), debt-funded.
Latest earnings
Mixed/noisy: revenue beat Zacks consensus by ~11%; headline GAAP EPS of $1.33 'missed' a $5.01 consensus, but that gap is mark-to-market/derivative-driven noise, not an operating miss — the operating metrics (EBITDA, FCF) beat and management reaffirmed guidance. Market treated it as a beat.
Reaffirmed FY2026 Adjusted EBITDA $1,750M–$2,050M and Adjusted FCF $980M–$1,180M (excludes Cornerstone). Cornerstone (closed June) + debt redemption adds >15% to CF/share and ~$1.00 FCF/share from interest savings; >$40/sh FCF targeted by 2028 (fact).
- Adjusted EBITDA (Q1 2026)
- $473M (>2x YoY)
- Adjusted FCF (Q1 2026)
- $350M (~4x YoY)
- Net leverage target (YE2026)
- <3.5x net debt/EBITDA incl. Cornerstone
- PJM 2026/27 capacity clear
- 6,702 MW at $329.17/MW-day
Growth drivers
- AWS PPA ramp — up to 1,920 MW from Susquehanna, ~$18B revenue through 2042; staged 840-1,200 MW by 2029, 1,680-1,920 MW by 2032 (fact)
- PJM capacity-price tightness — cleared 6,702 MW at $329.17/MW-day in the 2026/27 BRA, a multiple of historical clears (fact)
- Accretive M&A — Freedom/Guernsey (2.6 GW, closed Q4 2025) + Cornerstone (2.6 GW gas, closed 15 Jun 2026) compound FCF/share toward >$40 by 2028 (fact)
- Spark-spread expansion as PJM data-center load growth outpaces new supply
- Susquehanna uprates + potential SMRs with Amazon adding net-new carbon-free MW
- Share-count leverage: rising FCF/share with buyback capacity once leverage is below 3.5x
Reported financials — SEC EDGAR
Audited GAAP figures pulled from SEC filings · latest filing 2026-02-26. The audited primary-source spine — not financial advice.
Revenue — annual (GAAP)
Margins & balance sheet — FY’25
Bull & bear
TLN is the purest listed way to own scarce carbon-free baseload at the center of the AI power build-out: a 2.5 GW nuclear plant, an $18B AWS contract, and a PJM fleet leveraged to record capacity prices, with FCF/share compounding toward >$40 by 2028.
- Susquehanna nuclear is a strategically scarce, irreplaceable asset — hyperscalers are paying up for exactly this, and TLN already locked AWS for $18B through 2042
- Operating inflection is real and accelerating: Q1 2026 Adjusted EBITDA >2x and FCF ~4x YoY; FY2026 guidance reaffirmed with Cornerstone upside on top
- PJM tightness is structural — 2026/27 capacity cleared at $329/MW-day, and data-center load growth in PJM keeps energy + capacity revenue elevated
- Accretive, debt-funded M&A (Freedom/Guernsey + $3.45B Cornerstone, 5.2 GW combined gas) compounds FCF/share and diversifies beyond the single nuclear asset
- FERC's Dec-2025 colocation order de-risks the regulatory path that the whole behind-the-meter/data-center model depends on; Street is near-uniformly Buy with targets to $499-576
The valuation already prices a flawless AI-power build-out: TLN trades at a rich multiple on merchant cash flows that swing with PJM power and capacity prices, with elevated leverage and a regulatory regime (FERC/PJM colocation) that is still being written.
- Valuation is hot — forward P/E ~24-28x and a high EV/EBITDA on cyclical merchant earnings; little margin for error if any pillar slips
- Binary regulatory risk: FERC rejected the original behind-the-meter AWS ISA, forcing a restructure; the still-pending PJM colocation rules could land unfavorably
- Capacity-price reversion is the killer scenario — a drop from $329 toward ~$100/MW-day would gut a major earnings pillar
- Leverage is high and rising (~$6.1B net debt + $2.6B from Cornerstone); integration + balance-sheet strain layer onto volatile earnings
- Demand-timing risk: data-center construction can slip and new supply can clear faster than expected, deflating the scarcity premium the stock embeds
What it is worth
Multiple-based (EV/EBITDA + FCF/share) cross-checked against sell-side targets; merchant-IPP cash flows are scenario-dependent on PJM power/capacity prices, so a scenario band is more honest than a point estimate.
~$250-330 (estimate)
PJM capacity reverts toward ~$100/MW-day and/or FERC-PJM colocation rules disappoint; the rich multiple compresses on cyclical merchant earnings and elevated leverage.
~$455 (Street average 1-yr target)
guidance met, Cornerstone integrated, leverage to <3.5x, capacity prices stay firm but don't expand further.
~$500-576 (sell-side high
Goldman $499, Morgan Stanley $498, Street high $576) — assumes AWS ramp on schedule, sustained high PJM capacity prices, favorable colocation rules, and accretive M&A compounding to >$40/sh FCF by 2028.
At ~$412 the stock prices a high-execution AI-power scenario: ~24-28x forward P/E on >$40/sh FCF-by-2028 math; bull case extends the scarcity premium, bear case applies capacity-price reversion (~$100/MW-day) and a regulatory haircut.
SWOT
Strengths
- Owns Susquehanna — 2.5 GW of dispatchable, carbon-free, low-marginal-cost nuclear baseload, the scarcest asset class for AI data-center buyers
- Anchor $18B AWS PPA through 2042 converts merchant volatility into long-dated contracted cash flow
- PJM-concentrated ~10.7 GW fleet sits in the tightest US power market with the highest data-center load growth
- FCF inflecting hard: Adjusted FCF quadrupled YoY in Q1 2026; line of sight to >$40/sh by 2028
Weaknesses
- Still heavily merchant-exposed — energy + capacity revenue swings with PJM power and capacity prices
- Elevated leverage (~$6.1B net debt pre-Cornerstone; +$2.6B from the June deal) magnifies downside
- GAAP earnings are noisy (FY2025 net loss) from derivative mark-to-market — hard to read on headline EPS
- Single-asset concentration: an extended Susquehanna outage or nuclear incident is an outsized hit
Opportunities
- FERC's Dec-18-2025 order directing PJM to write colocation/BTM rules clarifies the path for behind-the-meter and grid-connected data-center deals
- Susquehanna uprates + SMR development with Amazon add net-new carbon-free MW
- Replicate the AWS template with other hyperscalers; up to 300 MW digital-infrastructure campus
- Continued PJM capacity-price strength and tightening reserve margins lift merchant upside
Threats
- FERC/PJM colocation rules could land less favorably than priced; original BTM ISA was rejected in 2025
- PJM capacity prices reverting toward ~$100/MW-day from $329 would compress a key earnings pillar
- Hyperscaler nuclear/gas competition (Constellation-Microsoft, Vistra-Meta, NRG) for scarce baseload
- Data-center demand timelines slip or new supply clears faster than expected, deflating the scarcity thesis
Moats, dependencies & bottlenecks
Moats
owning the 2.5 GW Susquehanna nuclear plant high (multi-decade asset, near-impossible to replicate; licensed nuclear) This is the core moat — hyperscalers need 24/7 carbon-free power and there are very few large nuclear assets to contract.
high (17-year term) Locks a large slice of Susquehanna output to an investment-grade counterparty; reduces merchant volatility on the contracted MW.
Sited in the tightest US power market with the most data-center demand; interconnection queue position and grid access are scarce.
~10.7 GW post-acquisitions; Freedom/Guernsey/Cornerstone are among PJM's most efficient CCGTs, low on the dispatch curve.
Dependencies
$18B PPA is the defining contract; single-counterparty concentration, though AWS is investment-grade and the term is long.
Original BTM ISA was rejected in 2025; the entire data-center power model hinges on the rules PJM files under FERC's Dec-2025 order.
Merchant energy and capacity revenue swing with PJM; capacity cleared at $329/MW-day but could revert sharply.
Gas-fleet margin depends on the spread between power and gas; a compressed spark spread hits the CCGTs.
Single large nuclear asset — license, safety, and outage performance are concentrated risks.
Acquisition-driven model relies on cheap debt issuance ($3.8B + $4.0B notes); rates and access matter for accretion.
Advantages
- First-mover hyperscaler nuclear contract — the AWS deal set the template and is already signed at $18B
- Lowest-marginal-cost large baseload (nuclear) in a market paying premium prices for carbon-free 24/7 power
- PJM concentration aligns the fleet with the highest-growth US data-center demand region
- Demonstrated accretive M&A engine — three large PJM gas deals (Freedom, Guernsey, Cornerstone) executed at attractive multiples (~6.7x EV/EBITDA)
- Strong near-uniform sell-side support and equity-market access to fund growth
Weaknesses
- High merchant exposure makes earnings cyclical and sensitive to PJM power/capacity reversion
- Elevated and rising leverage (~$6.1B net debt + Cornerstone) amplifies downside
- Customer concentration in AWS and asset concentration in Susquehanna
- GAAP earnings volatility from derivative mark-to-market obscures the underlying trend (FY2025 net loss)
- Rich valuation leaves little cushion if regulatory or demand assumptions disappoint
- Regulatory dependence on rules that are still being written (FERC/PJM colocation)
Bottlenecks
- PJM interconnection queue and grid-upgrade timelines gate how fast new colocation/data-center load can be served
- Susquehanna refueling-outage schedule caps available nuclear MW windows (the AWS grid reconfiguration was timed to the spring-2026 outage)
- Regulatory clock — PJM compliance filings (Jan/Feb 2026) and FERC review pace gate when behind-the-meter/colocation deals can scale
- Balance-sheet capacity / leverage ceiling (<3.5x target) limits how much further debt-funded M&A can run before equity is needed
- Skilled nuclear/plant operating labor and long-lead equipment for uprates and any SMR build
Top signals & trends
Top signals
The single biggest binary — favorable rules unlock behind-the-meter scaling; unfavorable ones cap the model. Watch the final orders, not the directive.
$329.17/MW-day in 2026/27 is a pillar; a sustained high clear confirms the thesis, a drop toward ~$100 breaks it.
On-schedule ramp converts contracted MW to cash; delays would push out the $18B realization.
Hitting the leverage target while integrating 2.6 GW validates the M&A engine and frees buyback capacity.
Net-new carbon-free MW is the next leg of growth beyond the existing fleet.
A second anchor contract would diversify customer concentration and re-rate the multiple.
Trends
Structural load growth tightens PJM and lifts both energy and capacity prices — the core tailwind.
Meta-Vistra/Oklo/TerraPower, Microsoft-Constellation, Google-Kairos/NextEra — validates the asset class and TLN's scarcity premium.
Provides clarity (positive) but the rejected original ISA shows the rules can constrain the most lucrative BTM structure.
$329/MW-day clears are a multiple of historical norms; durable if tightness persists, a headwind if it normalizes.
Has driven TLN's triple-digit rally and rich multiple; leaves valuation vulnerable to sentiment reversal.
Supportive gas prices help CCGT margins; a compressed spark spread would pressure the gas fleet.
Ecosystem & competitor graph
Suppliers feed the company; customers pull from it. Line thickness shows the strength of each tie (supply-chain dependency, customer earnings contribution). Hover to isolate a tie.
Cameco, Centrus) Fuel for Susquehanna's two units; low marginal cost but a critical input.
Fuel for the Freedom, Guernsey, Cornerstone (Waterford/Darby/Lawrenceburg) CCGT fleet — drives gas-plant COGS and spark spread.
Turbines, parts, and maintenance for the gas fleet; equipment for uprates/SMRs.
Funded $3.8B (Freedom/Guernsey) + $4.0B notes (Cornerstone); the M&A model depends on bond-market access.
Energy Capital Partners / Caithness / BlackRock (asset sellers) Sources of the acquired gas plants — the inorganic growth pipeline.
Anchor customer — $18B PPA through 2042, up to 1,920 MW from Susquehanna; also SMR/uprate exploration partner.
PJM Interconnection (capacity + energy market) Buys capacity and energy via the wholesale market; cleared 6,702 MW at $329.17/MW-day for 2026/27.
Microsoft, Meta, Google contracting nuclear/gas elsewhere — TLN's addressable next-anchor pipeline.
Delivers AWS's grid-connected power across the PJM grid under the restructured FTM arrangement.
Largest US nuclear fleet; signed Microsoft (Three Mile Island restart) and Meta deals — the most direct nuclear-to-hyperscaler competitor.
Large nuclear + gas IPP; 2.1 GW Meta nuclear deal (Beaver Valley/Perry/Davis-Besse). Direct PJM/data-center competitor.
Large gas-heavy IPP expanding into data-center power supply; competes for hyperscaler load and capacity revenue.
PJM nuclear owner exploring data-center colocation; regional baseload competitor.
Next-gen reactor developers (Oklo, TerraPower) contracting future hyperscaler nuclear — longer-dated competitive threat to the scarcity premium.
Enables rival gas capacity additions that could erode PJM tightness and capacity prices over time.