01The Setup
Through 2023, most utility integrated resource plans assumed roughly flat to modestly declining commercial and industrial load growth. The narrative was: efficiency gains roughly offset electrification gains; data centers were a growing segment but absorbed by existing generation. That narrative did not survive 2024.
Hyperscaler announcements in Q3-Q4 2024 added more than 50 GW of planned data center capacity to interconnection queues across PJM, MISO, ERCOT, and SPP. Some of those announcements were aspirational; many were real — signed leases, financed buildouts, announced operational dates inside 36 months.
Utility planners spent 2025 rewriting load forecasts. Generators spent 2025 reading those forecasts and updating their forward curve assumptions. The forward curve, finally, is the audit trail of what the consensus actually believes.
02The Data
- PJM 2027-2030: Backwardation flattened
- Back-of-curve power: Firmer than front
- Henry Hub 2027-2030: Similar firming
- Gas-fired CCs: Still marginal price-setter
- Capacity factor revisions: Downward
- New build queue wait: 5-7 years in major ISOs
The PJM 2027-2030 power forward strip used to be in backwardation — nearer-dated prices higher than later-dated. That shape has flattened materially. The back of the curve is now firmer than the front, indicating the market expects demand to materialize and supply not to keep pace.
Henry Hub forward gas for the same period shows similar firming, with the steepest moves in years where new gas-fired generation interconnects. Gas demand for power generation is the connecting tissue — combined-cycle plants remain the marginal price-setting unit during peak hours in most regions.
Interconnection queue analysis tells the same story from the supply side. New build solar, wind, storage, and gas projects sit in 5-7 year queue waits across major ISOs. The supply response to surging demand is structurally lagged.
03The Implication
Power and gas prices are more correlated than they were five years ago, because gas generation is the marginal price-setter in tight reserve conditions, and tight reserve conditions are becoming more common. Procurement decisions hedging power risk without considering gas risk are missing half the exposure.
Strategies that assumed flat forward curves through 2028 are pricing risk incorrectly. Customers with hedge ladders rolling off in 2027-28 are about to relock at materially higher prices. Sustainability strategies built on assumptions of expanding clean energy supply are running into the reality that clean energy capacity is now competing with surging baseload demand — and demand is winning.
04The Recommendation
- For procurement. Lengthen contract tenor for customers with operational flexibility. The longer-dated forward prices are still below where the curve is heading once interconnection actually delivers the announced capacity. There's a window — probably 2026 — where forward prices haven't fully priced the demand growth.
- For DR-capable customers. The capacity stack is improving meaningfully. Customers with curtailable load should evaluate revenue stacking across capacity market participation, ancillary services, and demand response programs.
- For sustainability roadmaps. Recalibrate. PPAs for new build clean energy are pricing higher; existing clean capacity is being claimed faster. Net-zero commitments built on 2022 assumptions about clean supply abundance need updated assumptions.
- For ESPs and large customers. Operational flexibility — load shifting, behind-the-meter generation, demand response — is worth more now than it was. Investments in that flexibility have shorter payback periods than the underlying capex would have suggested in a flat-curve world.
AI demand was a thesis in 2023, a debate in 2024, and a forward curve in 2026. The market believes the load is coming. The procurement decisions that don't update accordingly are pricing yesterday's market into tomorrow's contracts.