The AI energy premium is now measurable in a primary grid operator’s pricing data. That’s new.
Per Monitoring Analytics’ PJM market monitoring data, wholesale power prices in the PJM Interconnection, the grid serving 65 million people across 13 states and the District of Columbia, averaged $136.53 per megawatt-hour in Q1 2026. A year earlier, that average was $77.78. The arithmetic is a 75.5% increase, consistent with the 76% figure reported across sources. Monitoring Analytics is the independent market monitor for PJM, this is T1 authority for PJM pricing data.
Alongside the pricing data, NERC, the North American Electric Reliability Corporation, issued a reliability alert related to computational load and AI data center demand growth. The alert’s existence is credible given the source authority. The specific designation level (“Level 3”) reported in some outlets has not been confirmed against NERC’s formal classification system, and this brief won’t use it. Call it what it is: a rare grid reliability alert from the primary U.S. grid oversight body, focused on AI infrastructure demand.
Warning
NERC classification: The specific 'Level 3' designation reported in some outlets has not been confirmed against NERC's official alert classification system. This brief uses 'reliability alert' only. Operators should check NERC's official publication for the formal designation and required response.
The projections frame where this goes. EPRI’s 2026 analysis projects that AI data centers could consume between 9% and 17% of U.S. electricity by 2030. That’s a wide range, the spread reflects genuine uncertainty in AI infrastructure buildout pacing, but even the low end would represent a structural shift in national power demand. S&P Global projects AI data center demand reaching 134.4 GW by 2030. Both are analyst projections with inherent uncertainty. Both come from T1 authorities.
The Utah “Stratos Project” adds a local dimension. A large-scale data center development in Utah, reportedly called the Stratos Project, is facing community and regulatory opposition over a reported power requirement of approximately 9 GW, according to Quartz and Newsweek. That figure has not been confirmed against project filings. For reference, 9 GW would represent a single project consuming roughly 0.8% of current U.S. total installed generating capacity, an extraordinary concentration.
This is the third major energy-grid story connected to AI data center demand to land in this hub in the current cycle, following Oregon’s Schedule 96 tariff development and the Hitachi Power-as-a-Service announcement earlier this month. The Q1 2026 PJM pricing data is the first time AI infrastructure demand has been reflected in primary-source grid operator pricing numbers rather than analyst forecasts. That’s the distinction that makes this item structurally different from the adjacent coverage.
What to Watch
For data center operators and hyperscaler infrastructure teams, the $136.53/MWh figure is a planning anchor. Energy costs that were modeled at $70-80/MWh for 2025-2026 infrastructure planning cycles are running roughly double that. Contracts structured on those assumptions need review. Enterprise AI buyers one or two layers removed from direct power procurement, those paying through cloud providers, should track whether Q1 2026 energy cost increases are appearing in hyperscaler earnings commentary.
Watch the NERC alert documentation. The formal NERC publication will specify exactly what the alert requires from grid operators and large load customers. That detail, not the headline, is what data center operators need to track.