Will a new nuclear power purchase agreement exceeding 1 GW total capacity be announced by a hyperscaler by September 30, 2026?

The asset-repricing channel is demonstrably active — 10 GW+ in binding nuclear PPAs signed in 18 months establishes that 1 GW per hyperscaler is well within demonstrated cadence. CEG's CEO publicly characterizing additional Microsoft capacity discussions as 'past the seventh-inning stretch' is a leading indicator that one specific deal is in late-stage negotiation. With ~6 months in the window and five active hyperscalers (Microsoft, Amazon, Google, Meta, Oracle) each with non-trivial individual probabilities, the union probability of any one reaching 1 GW cumulative is structurally above 50%. Resolution allows cumulative aggregation, lowering the bar (multiple 300-500 MW deals to one hyperscaler clears it). PJM capacity at 77% probability >$300/MW-day reinforces deal economics. Counterweight: 'signed or binding' eliminates partnership announcements that historically outpace binding contracts by 6-18 months (Meta/Oklo binding PPA only 14% by YE2026).

Two opposing forces dominate. Toward YES: VST signed 2,600+ MW cumulatively across 2024-2025 — the cadence alone supports 1+ GW per hyperscaler in 6 months as base case; resolution accepts cumulative aggregation per hyperscaler, which is the most likely realization path; demand signal is unambiguous (PJM >$300/MW-day at 77%). Toward NO: the dossier puts CEG additional PPA at 48% and TLN new 500+ MW at 35% — these are firm-level not hyperscaler-level probabilities, and 1 GW threshold per hyperscaler is meaningfully harder than 'a new PPA exists.' Also, available merchant nuclear capacity is finite and increasingly tied up; FERC has obstructed behind-the-meter expansions; saturation thesis is explicitly raised in market description. Pipeline depth is real but conversion rate from announced partnership to binding contract is slow (Meta/Oklo at 14% by YE2026 illustrates the gap). Settling slightly above coin-flip on union-probability logic but discounting for correlation across deals.

Reading the resolution criteria carefully: 'cumulative new commitments from one hyperscaler' is the operative phrase. This is significantly easier than a single >1 GW deal. Microsoft already has TMI 835 MW; only ~165 MW of new commitment to Microsoft from anyone clears the threshold. Amazon has Susquehanna 960 MW; needs ~40 MW of incremental commitment. But these are cumulative *new* commitments in the window — historical capacity doesn't count. Even so, Microsoft's known active discussions for additional CEG capacity, plus the precedent that single-reactor PPAs are ~700-1,200 MW, makes one Microsoft deal likely sufficient. Probability favors YES via Microsoft path alone (35-45%), and additional probability mass from Amazon, Google, Meta, Oracle. Apply ~20% correlation discount and the result is ~58%.

The optimistic case relies heavily on continuation of an 18-month cadence, but cadence is not destiny. Three discrete drags: (1) the dossier explicitly notes that addressable fleet capacity is becoming saturated — most merchant nuclear has already been signed or is constrained; (2) FERC has been obstructive on behind-the-meter (Talen/Susquehanna precedent), narrowing the BTM channel that produced the largest single deal (Amazon 960 MW); (3) 'signed or binding' is a strict bar — many deals progress as MOUs/term sheets first, with binding execution slipping into following quarters. The CEG 48% / TLN 35% firm-level rates, when translated to hyperscaler-aggregated >1 GW, do not obviously sum to >50% once you exclude the partnership announcements that lack capacity figures. Rate at coin-flip with mild upside bias from raw demand signal.

Cumulative-per-hyperscaler resolution allows aggregation of smaller deals. With CEG at 48% probability of additional PPA and TLN at 35% for 500+ MW contract, plus active Microsoft-CEG late-stage talks, plus four other hyperscalers in market, union probability of any one crossing 1 GW is structurally above 50%. Counterweight: 6-month window is short and 'signed or binding' criterion strict. Settle above coin-flip, below high-conviction YES.