Learn AI compute, then follow the market
Morning BriefContact
Compute CollegeComputeTape

Power Watch

Power Watch: learn the grid constraints behind AI compute

Track the grid and power constraints shaping AI compute capacity.

Power Watch tracks the regional power constraints that can delay data-center buildout, limit usable GPU capacity, or raise compute costs.

View MarketWhy power matters

Regions

Regional Heat Map

Constraint scores for AI data-center and GPU cloud capacity, as of Jul 22, 2026. Higher means more constrained.

RegionStatusScoreTrendConfidenceMain ConstraintAI Compute ImpactSource
PJMSevere85 / 100TighteningHighThe July 2026 auction again cleared at its price cap ($325/MW-day, $16.4B total) and procured 138,318 MW, but the 14.7% reserve margin remained below the reliability requirement as large data-center loads continued entering the forecast.Large AI campuses face capped but still elevated capacity costs and tighter reserves; PJM is pursuing backstop procurement and flexible connect-and-manage rules while energization risk remains unresolved.PJM Inside Lines
ERCOTSevere90 / 100TighteningHighLarge-load queue above 233 GW (about 70%+ data centers) against roughly 85 GW peak; in early June 2026 ERCOT moved to a 'Batch Zero' process that triages large loads against what the transmission system can support today.Fastest-growing buildout region, but Batch Zero deliverability screening and tight reserve margins mean projects without grid readiness face rejection or long delays.ERCOT — Large Load Integration
SoutheastElevated65 / 100TighteningMediumUtilities planning 10+ GW of new capacity for data centers (Georgia Power ~10 GW; Southern 50 GW pipeline; Dominion 47 GW contracted), though independent analysts call the forecasts aggressive.Heavy utility commitments and new gas-plus-storage, but contested load forecasts and rate/tariff decisions add cost and approval risk.Georgia PSC — Data Center Fact Sheet
MISOModerate55 / 100EasingHigh2026/27 capacity prices eased (summer $424/MW-day, down from a $666 record); adequacy met across all zones but flagged at risk as large loads accelerate.More near-term headroom than PJM or ERCOT, but interconnection timelines lengthen as large-load requests grow.Utility Dive
CAISOModerate50 / 100StableHighModest data-center forecast (+1.8 GW by 2030); summer 2026 assessed well-positioned with ~8 GW of new resources, but large-load cost allocation and 'pulsating' loads remain unresolved.Lower AI-campus pressure than PJM or ERCOT; high prices and interconnection-policy uncertainty are the watch items.California Energy Commission
SPPWatch45 / 100TighteningMediumPeak load projected to nearly double (56 to 105 GW over 10 years); 26+ GW of >100 MW load requests (9 GW data centers), served by new but curtailable large-load tariffs (HILL/CHILL).Current headroom and a fast 90-day interconnection path, but curtailable terms and future growth temper long-term firmness.SPP — High Impact Large Load Integration
Pacific NorthwestModerate50 / 100StableMediumNorthwest planners put new data-center and chip-fab growth at roughly 1.5–5.0 average GW by 2030; BPA's 2026 study still flags hydro, transmission, and peak-capacity uncertainty as Provider of Choice obligations are finalized.Low-cost hydro remains attractive, but transmission availability, hydro variability, and the need for flexible large-load designs limit how quickly firm new compute load can energize.BPA — 2026 Loads & Resources Study (White Book)
NationalElevated70 / 100TighteningMediumAI data-center load growth is outpacing grid planning in the highest-demand regions (PJM, ERCOT, Southeast).Power remains a binding medium-term constraint on GPU cloud capacity, concentrated in the Eastern Interconnection.Load-weighted composite

Deals

Power Deal Tracker

Large power agreements and utility arrangements tied to AI data-center capacity.

Buyer / OperatorPower TypeRegionSizeStatusExpected TimingImpactConfidence
Hyperscaler / AI cloudNuclear / PPANot disclosedNot disclosedAnnouncedNot disclosedLong-term capacity supportMedium
Data-center operatorGrid interconnectionPJMNot disclosedPendingNot disclosedEnergization riskMedium
AI infrastructure providerGas / behind-the-meterERCOTNot disclosedProposedNot disclosedPotential capacity unlockMedium
Utility / data-center customerTariff / rate caseSoutheastNot disclosedWatchNot disclosedCost and approval signalMedium

Queue

Data-Center Power Queue

Major AI data-center projects and their power-access risk.

Project / AreaOperatorRegionEstimated MWUtility / GridStatusPower RiskTimelineConfidence
Northern Virginia clusterMultiplePJMNot disclosedDominion / PJMActiveHighOngoingMedium
Texas AI campus pipelineMultipleERCOTNot disclosedERCOT utilitiesExpandingMedium / HighOngoingMedium
Southeast data-center growthMultipleSoutheastNot disclosedRegional utilitiesRisingMediumOngoingMedium
Pacific Northwest AI sitesMultiplePNWNot disclosedRegional utilitiesWatchLow / MediumOngoingMedium

Signals

Top Power Signals

The latest source-backed power items most relevant to compute capacity.

Aug 2, 2026

PJM Capacity Auction Procures 138,318 MW of Generation Resources as Work Continues To Address Growing Electricity Demand

Signal: PJM Capacity Auction Procures 138,318 MW of Generation Resources as Work Continues To Address Growing Electricity Demand. Impact: Upward pressure on compute costs. The grid-market change described in “PJM Capacity Auction Procures 138,318 MW of Generation Resources as Work Continues To Address Growing Electricity Demand” can raise the power component of compute costs before it changes physical GPU supply. Affected market area: Capacity timing.

Source: PJM Interconnection

Jul 31, 2026

Exelon ‘high probability’ data center load falls 40%

Signal: The drop to 11 GW reflects Exelon's ability to weed out speculative projects through "transmission service agreements," said CFO Jeanne Jones.</ Impact: Positive for future capacity. The project described in “Exelon ‘high probability’ data center load falls 40%” adds prospective compute capacity, but permitting, power procurement, construction, and deployment timing prevent it from increasing immediately available GPU supply. Affected market area: Capacity timing.

Source: Utility Dive

Jun 9, 2026

FirstEnergy asks FERC to require data centers to pay for transmission interconnection costs

Signal: FirstEnergy's proposal adopts a cost allocation practice from the gas pipeline sector. It comes ahead of the Federal Energy Regulatory Commission's expected large load… Impact: Upward pressure on compute costs. The grid-market change described in “FirstEnergy asks FERC to require data centers to pay for transmission interconnection costs” can raise the power component of compute costs before it changes physical GPU supply. Affected market area: Cost and Capacity timing.

Source: Utility Dive