Thematic Exposure -- 9/10

Oligopoly Hard-Gate: PASS -- 6 Merchant Nuclear Operators Nationally, 2-Player PJM Duopoly

9 / 10. Behind-the-meter co-located nuclear in PJM that a hyperscaler can buy today: Susquehanna (TLN) and Three Mile Island restart (CEG-Microsoft, 2028+). That is the entire universe -- two plants, two operators, ~3.3 GW combined. By any reasonable definition this is a 2-player local oligopoly, and arguably a 1-player monopoly during the 2026-2028 window before TMI restarts. Replacement supply is structurally throttled for 10+ years.

TLN cleared 6,820 MW in the 2025/26 PJM BRA at $269.92/MW-day -- a ~9x jump from $28.92/MW-day in the prior auction. The 2026/27 and 2027/28 BRAs cleared near the administrative price cap (~$329-333/MW-day in RTO), locking in the higher capacity revenue trajectory through the entire near-term forecast period.

Forward capacity revenue locked: $963M (2026), $1,053M (2027), $443M (2028 -- partially auctioned). This is the most contractually visible piece of the AI-power thesis.

Within PJM: Susquehanna + TMI Restart = 2-Player PJM Duopoly Through 2028

PJM contains roughly 33 GW of nuclear. Of merchant-contractable nuclear in PJM, Constellation dominates with ~14 GW. Talen's Susquehanna is the only large merchant nuclear plant in the PPL/MAAC eastern zone -- physically adjacent to the highest-LMP, most data-center-saturated corridor in PJM (NoVA/Loudoun/DOM zone is the demand sink; PPL/MAAC is the closest deliverable nuclear supply).

Behind-the-meter co-located nuclear in PJM that a hyperscaler can buy today: Susquehanna (TLN) and Three Mile Island restart (CEG-Microsoft, 2028+). That is it. Two plants, two operators, ~3.3 GW combined. By any reasonable definition this is a 2-player local oligopoly, and arguably a 1-player monopoly during the 2026-2028 window before TMI restarts.

BTM Rejected Twice -- Restructured as Front-of-the-Meter to Sidestep the Question

The original 480 MW behind-the-meter expansion ISA was rejected by FERC in a 2-1 vote in November 2024 (Commissioners Christie, See in majority), and rejected again on rehearing in April 2025, on grounds that the co-located load structure unduly shifts transmission costs to PJM ratepayers (AEP/Exelon estimated ~$140M/yr cost shift). TLN appealed to the Fifth Circuit (Jan 2025).

Management response: the June 2025 expanded AWS PPA was restructured as a "front-of-the-meter" 1,920 MW PPA through 2042 -- sidestepping the behind-the-meter regulatory question entirely. The 300 MW legacy BTM block will also transition to front-of-the-meter once transmission upgrades complete (~Spring 2026). McFarland on the Q4 2025 call: "We signed a revamped and doubled front-of-the-meter PPA with Amazon at Susquehanna."

Implication: the regulatory risk that capped Susquehanna's BTM monetization has been structurally re-routed, not eliminated -- TLN gives up some economics (transmission charges, network upgrade costs) in exchange for regulatory certainty. The thesis is intact; the path is just slightly less profitable on a per-MWh basis than a pure BTM deal would have been. PJM-wide rulemaking on co-located load remains a sector overhang.

Headline math: $18B / (1,920 MW × ~8,000 hrs × 17 years) ≈ $70/MWh blended -- but the contract ramps over 7 years, so steady-state pricing is materially higher. Market analysts triangulate run-rate at $80-100/MWh for the gross PPA, well above current PJM West Hub ATC (~$45-55/MWh).

Downside protection: Inflation Reduction Act nuclear PTC floor at $43.75/MWh protects realized revenue even in a soft-power-price scenario. Combined with the locked $963M / $1,053M / $443M of forward capacity revenue, the cash flow visibility through 2028 is unusually high for an IPP.

Why this matters: Talen has already monetized the local-monopoly option. The AWS deal is the empirical proof that a hyperscaler will pay an >1.5x premium to ATC pricing for 24/7 carbon-free baseload at a deliverable PJM-East location. That premium is the score.