Great article and insights! My one quibble is with the characterization of capacity markets as having previously been a reliable planning tool. These constructs were never functional; the scale of data center growth just exposed their inadequacy in ways no one could ignore.
This piece nails the paradigm shift: by late 2025, U.S. renewable PPAs morphed from voluntary ESG checkboxes into hard infrastructure rationing tools, driven by AI/data center demand outrunning grid expansion.
Buyer concentration (tech ~75% of U.S. PPAs), deliverability mandates (locational/congestion basis), and hyperscaler scale preferences are reshaping everything—developers chase repeatable big contracts, marginal renewables get pre-allocated to AI loads.
Pricing confirms it: LevelTen Q4 2025 shows P25 solar up 3.2% QoQ (~$61-62/MWh, +9% YoY), wind slightly down but elevated overall. PJM's 2026/2027 BRA cleared at the cap (~$329/MW-day), with billions in incremental costs tied to data center forecasts.
Policy catching up fast—Virginia's GS-5 class (effective Jan 2027 for >25 MW loads) enforces 85%/60% minimum demand charges to curb cross-subsidies, mirroring Texas SB 6 logic.
Investor pivot to "power-first" (upstream co-location like Blackstone's $25B+ PA play, portfolio hedging, geographic selectivity) is the logical response—control over energized capacity beats capital alone.
Clean power scarcity is now the binding constraint on AI scaling. Nuclear/SMRs or behind-the-meter gas/renewables hybrids look like the escape valves—how soon do you see them materially easing the squeeze? Great analysis—spot on for 2026 realities
The PPA squeeze feels real, it’s less a sudden shock and more the result of demand finally running into physical limits. Data centre growth has simply moved faster than grid upgrades, new generation, and planning approvals can keep up with. What’s changing is the assumption that clean power can always be locked in early, cheaply, and at scale. In reality, feels like PPAs are becoming more conditional, more location-dependent, and more exposed to grid constraints and timing gaps. From a delivery point of view, this pulls projects back to basics where the power actually is, how fast it can arrive, and what interim steps are needed along the way. It’s more about reconciling them with how infrastructure really gets built.
Great article and insights! My one quibble is with the characterization of capacity markets as having previously been a reliable planning tool. These constructs were never functional; the scale of data center growth just exposed their inadequacy in ways no one could ignore.
This piece nails the paradigm shift: by late 2025, U.S. renewable PPAs morphed from voluntary ESG checkboxes into hard infrastructure rationing tools, driven by AI/data center demand outrunning grid expansion.
Buyer concentration (tech ~75% of U.S. PPAs), deliverability mandates (locational/congestion basis), and hyperscaler scale preferences are reshaping everything—developers chase repeatable big contracts, marginal renewables get pre-allocated to AI loads.
Pricing confirms it: LevelTen Q4 2025 shows P25 solar up 3.2% QoQ (~$61-62/MWh, +9% YoY), wind slightly down but elevated overall. PJM's 2026/2027 BRA cleared at the cap (~$329/MW-day), with billions in incremental costs tied to data center forecasts.
Policy catching up fast—Virginia's GS-5 class (effective Jan 2027 for >25 MW loads) enforces 85%/60% minimum demand charges to curb cross-subsidies, mirroring Texas SB 6 logic.
Investor pivot to "power-first" (upstream co-location like Blackstone's $25B+ PA play, portfolio hedging, geographic selectivity) is the logical response—control over energized capacity beats capital alone.
Clean power scarcity is now the binding constraint on AI scaling. Nuclear/SMRs or behind-the-meter gas/renewables hybrids look like the escape valves—how soon do you see them materially easing the squeeze? Great analysis—spot on for 2026 realities
The PPA squeeze feels real, it’s less a sudden shock and more the result of demand finally running into physical limits. Data centre growth has simply moved faster than grid upgrades, new generation, and planning approvals can keep up with. What’s changing is the assumption that clean power can always be locked in early, cheaply, and at scale. In reality, feels like PPAs are becoming more conditional, more location-dependent, and more exposed to grid constraints and timing gaps. From a delivery point of view, this pulls projects back to basics where the power actually is, how fast it can arrive, and what interim steps are needed along the way. It’s more about reconciling them with how infrastructure really gets built.