Will data center clean energy PPAs signed in H1 2026 remain concentrated among 4 or fewer providers?

The structural argument for concentration is real but oversold. Hyperscaler power demand IS selecting a narrow technology profile (firm, dispatchable, carbon-free), which structurally excludes hydrogen and resi solar. However, the qualifying technology corridor — utility-scale renewables + storage + nuclear + gas-peaker hybrid — is populated by FAR more than 4 credible providers when you look beyond the in-universe set. NextEra alone signs multiple GW of PPAs per year; Constellation (Microsoft Three Mile Island), Vistra, Talen, Brookfield, Engie, Pattern, Invenergy, Ørsted, EDP Renováveis are all active hyperscaler counterparties. The 80%/<=4 threshold is genuinely hard when 8-10 providers are each capable of >=10% share of new announcements in a 6-month window. Q1-Q2 2026 will almost certainly include nuclear restart announcements, multiple NextEra solar+storage deals, and at least 2-3 IPP gas-to-clean conversions. Lean NO.

Two opposing forces dominate. Concentration force: hyperscaler procurement preferences proven counterparties for speed-to-power, IRA domestic content + FEOC narrow the qualifying bidder set, capital cycle compression eliminates new entrants. Diffusion force: the universe of qualifying providers in H1 2026 includes 8+ firms with credible track records, and hyperscaler-driven diversification is itself a procurement principle. The critical scope question is whether 'clean energy PPAs' includes nuclear. If YES (Constellation, Vistra count), then 5+ providers each at >=10% is plausible. If NO (nuclear excluded), then the qualifying set shrinks to BE/NRG/AES/FLNC + NextEra + Brookfield + Pattern + a few others — and concentration among top 4 becomes more achievable. Sector dossier explicitly separates 'non-nuclear clean energy' from nuclear, suggesting the analyst intent is non-nuclear scope. That tightens the qualifying set and pushes probability higher. But the broader market interpretation will likely include nuclear, given hyperscalers don't draw that distinction. Lean slightly under coin flip.

The market premise is interesting but the >=10% threshold for 5+ providers is low. In H1 2026, plausible major announcements include: a NextEra renewables-plus-storage hyperscaler deal (likely 1+ GW); Constellation expanding Microsoft/Three Mile Island contract or signing AWS; Vistra signing additional gas-to-clean conversions for hyperscalers; Brookfield Renewable announcing follow-on deals to its 2024 Microsoft PPA; Talen Energy Amazon-related announcements; AES adding to existing 8.2 GW; NRG announcing more BYOP wins; BE announcing new orders. That's already 8 providers. Even if BE and AES dominate the in-universe announcements with 30-40% combined, the rest distributes across 5-7 outsiders — easily clearing the 5x10% threshold. The structural-concentration thesis from the dossier is correct about the technology corridor but underestimates the headcount of qualifying providers within that corridor.

Three considerations push around equal probability. (1) The dossier's Top Finding #2 is unambiguous: 'Data Center Power Demand Is the Dominant Selection Mechanism — and It Excludes Half the Sector.' This is a strong concentration prior. (2) However, the dossier also notes its own selection bias — only 7 of hundreds of sector participants — and explicitly calls out NextEra/Constellation/Vistra as material competitors in scope. (3) The resolution criteria specifies 'newly announced' PPAs, not cumulative backlog. H1 2026 announcement flow may include 6-9 distinct providers because hyperscalers spread procurement risk across counterparties. The single most important factor is whether nuclear PPAs count as 'clean energy PPAs' for resolution — which is ambiguous. Conservative midpoint: 42%.

Hyperscaler PPAs are concentrated structurally because the technology profile (firm, dispatchable, carbon-free) is selective. But the universe of qualifying providers is larger than the 4 in the source trigger. NextEra, Constellation, Vistra, Brookfield, Talen are all active hyperscaler counterparties and will likely each capture material share in H1 2026 announcements. The 80%/<=4 threshold is genuinely demanding given how diversified hyperscaler procurement is. Lean below coin flip but with meaningful tail probability if nuclear is excluded from scope.