TLN

Talen Energy

Summary

What they do:

Own and operate 13.1 GW of power generation capacity — including 2.2 GW of nuclear from the Susquehanna plant and a significant dispatchable fossil fleet — with the expanded 1,920 MW Amazon Web Services front-of-the-meter nuclear PPA providing ~$18B in contracted revenue over 17 years for carbon-free power delivered anywhere in Pennsylvania to AWS data campuses.

Why they matter:

Talen sits at the intersection of nuclear power and hyperscaler demand. The "Amazon 2.0" PPA (restructured to front-of-the-meter in June 2025) expanded to 1.9 GW is the largest nuclear data center power deal in existence. Talen is aggressively growing through acquisitions — Freedom/Guernsey (+2.8 GW closed Nov 2025), Cornerstone (+2.45 GW expected summer 2026) — building a diversified power generation platform across PJM with 18+ GW at completion.

Recent performance:

FY2025 Adjusted EBITDA $1.035B, Adjusted Free Cash Flow $524M. GAAP net loss $219M (acquisition-related). Q4 Adj. EBITDA $382M, Q4 Adj. FCF $292M (Q4 alone exceeded all of FY2024). 2026 guidance: Adj. EBITDA $1.75B–$2.05B, Adj. FCF $980M–$1.18B (excluding Cornerstone). Liquidity $2B+ ($1.2B cash + $900M revolver). Net leverage 3.0x on 2026 EBITDA midpoint. Stock ~$350, market cap ~$17B, EV ~$23B.

Our Verdict

Play TypeEstablished
Rel. ValueCompelling

The largest behind-the-meter nuclear-data center power play with a 1.9 GW AWS PPA worth ~$18B through 2042, plus aggressive fleet expansion to 18+ GW — at ~17x EV/EBITDA with significant acquisition debt and single-hyperscaler concentration.

Structural trends

Hyperscaler demand for carbon-free baseload powernuclear renaissance for data centersfront-of-the-meter contracted power deliverypower generation fleet consolidationPJM capacity market tightening

Structural

78

/ 100

Moat

7/10

Susquehanna-AWS combination is irreproducible; 17-year PPA creates structural advantage

AI Exp.

High

~40% AI

Play Type

Established

AI Growth

~25%

Rel. Value

84

COMPELLING

The Catch

Talen's investment case still rests heavily on a single customer relationship — Amazon's 1.9 GW PPA for Susquehanna nuclear output. The June 2025 restructuring to front-of-the-meter actually reduces the behind-the-meter regulatory risk that once loomed, but introduces PPA ramp timing risk: full 1,920 MW delivery depends on AWS campus build-out pace, and revenue recognition is back-loaded in 480 MW increments. The Montour zoning rejection demonstrates that data center development faces real local opposition — Mac called it "our fault" for talking about it publicly and has committed to no longer discussing development pipeline in detail, which reduces transparency. The serial acquisition strategy (Freedom/Guernsey $3.8B, Cornerstone pending) creates leverage, though net leverage at 3.0x on 2026 EBITDA is manageable and management targets <3.5x. Susquehanna had a Unit 2 extended outage in spring 2025 and did not receive the PTC in 2025 — nuclear operational risk is non-trivial. This remains a concentrated bet on one hyperscaler, one nuclear plant, and PJM market dynamics — but the fleet is diversifying and contracted cash flows are growing.

If They Win

If the AWS PPA ramp accelerates (all 1,920 MW delivering by 2028 vs. the base schedule), Cornerstone closes summer 2026 adding ~$500M EBITDA, additional hyperscaler/co-locator PPAs materialize from the active pipeline across PJM, PJM capacity markets clear at uncapped levels (>$500/MW-day), and the RBP framework enables new build with 15-year contracts — Talen becomes a $2.5B+ EBITDA power generation platform with the most valuable combination of contracted nuclear cash flows and strategically positioned fossil generation in the tightest wholesale power market in America. The flywheel model — generate contracted cash flows, acquire assets, develop powered land, sign more PPAs — becomes self-reinforcing. Stock re-rates to $500–600 as the market prices in the platform value rather than just the single AWS PPA.

Not financial advice. All scores generated via AI algorithms using public data.