PJM's 15-Gigawatt Power Push Could Force Data Center Operators to Pay Premiums for Grid Access

The core issue is a staggering supply-demand imbalance. The forecast for data center power is accelerating faster than the grid can build new capacity. The most recent outlook from 451 Research shows utility power provided to these facilities will nearly triple, rising from 61.8 gigawatts in 2025 to 134.4 gigawatts in 2030. This isn't just growth; it's an acceleration of demand that is transforming the energy landscape.

In response to this pressure, the grid operator for 13 mid-Atlantic and Midwest states, PJM Interconnection, has taken emergency action. It is seeking 15 gigawatts of new power supplies through a special proposal. This move is a direct acknowledgment that the existing planning process is too slow to meet the projected surge. The operator would match proposed data centers with new power plants between September 2026 and March 2027, a compressed timeline meant to force a solution.

To address the broader infrastructure gap, PJM has also approved a massive $11.8 billion transmission upgrade plan. A significant portion of this funding is earmarked for the epicenter of the boom: Virginia's "data center alley." $4.8 billion of the plan will go to Dominion Energy to build new lines and substations in the state. This includes a $2.3-billion, 525-kV underground backbone line and two HVDC converter stations, designed to deliver 3,000 megawatts to Loudoun County alone.

The scale of the emergency is clear. The grid is already under strain, with multiple "criteria violations" in 2025 due to thermal overloads and voltage issues. The 15-gigawatt proposal and the $11.8 billion plan are attempts to close a gap that is widening rapidly. The bottom line is that the commodity balance for grid power is being disrupted by an AI-driven demand surge that is outpacing supply, forcing a costly and urgent response.

The Market's Price Signal: Capacity Auctions Reflect Scarcity

The market is sending a clear price signal about the emerging scarcity. The results of PJM's 2026/27 Base Residual Auction, announced in July 2025, show capacity prices hitting the FERC-approved cap of $329.17/MW-day for the entire footprint. That represents a 22% increase from the prior year's auction price. More importantly, the auction cleared at a volume just over the projected reliability requirement, underscoring a tightening supply-demand balance.

The price pressure is not uniform, however. In the most constrained zones, the signal is even more urgent. The Baltimore Gas & Electric (BGE) zone cleared at $466.35/MW-day, while the Dominion zone cleared at $444.26/MW-day. These prices are significantly higher than the regional cap, highlighting localized bottlenecks where the grid is under the greatest strain. For context, the auction's forecasted peak load for the delivery year increased by more than 5,400 MW, driven largely by data center expansion.

This dynamic has a direct and critical implication for data center operators. Securing reliable power is no longer just about energy consumption; it is about paying a premium for the capacity to access that energy when needed. Capacity costs can make up to 30% or more of a customer's power supply charges. The auction results show that the market is pricing in the risk of shortages, a risk that is being amplified by the AI-driven demand surge. For operators, this means that the cost of securing a stable power supply is becoming a major and rising line item in their operating expenses.

The Path to Resolution: Matching Load with Generation

The grid operator's emergency plan is now being fleshed out with concrete mechanisms to match the projected load with new supply. The central strategy is a "Bring Your Own New Generation" approach, which aims to expedite the pairing of data center demand with firm baseload power. Under this model, PJM would match proposed data centers with new power plants over a compressed 18-month window, from September 2026 to March 2027. This direct pairing is designed to bypass the traditional, slower planning and permitting processes, forcing a solution to the immediate capacity crunch.

This pairing effort is backed by a massive, multi-year infrastructure investment. The approved $11.8 billion transmission upgrade plan is the backbone of this resolution. A major portion of that funding-$4.8 billion for Dominion Energy-is targeted at Virginia's "data center alley," the epicenter of the boom. The plan includes a $2.3-billion, 525-kV underground "backbone" line and two HVDC converter stations, designed to deliver 3,000 megawatts to Loudoun County alone. This project, along with a $1.7-billion line in Pennsylvania, is meant to move power from generation sources to the high-demand data center clusters. The plan also addresses delays to New Jersey offshore wind projects, aiming to integrate renewable sources into the solution.

Yet, even with this physical build-out, regulatory clarity remains a key hurdle. To provide more certainty for developers, PJM has filed a new tariff aimed at clarifying rules for behind-the-meter generation. This filing, submitted in February 2026, responds to a Federal Energy Regulatory Commission order and seeks to deliver some finality by July 31, 2026. The goal is to establish a clear framework for how large industrial loads like data centers can operate with on-site generation while maintaining grid reliability. This move is a direct response to the market monitor's earlier recommendation that new data centers should be required to bring their own generation, a principle that could help balance the system if adopted.

The path forward is a multi-pronged effort: a fast-tracked pairing mechanism, a decade-long transmission build-out, and evolving regulatory rules. The success of this resolution hinges on executing all three in parallel. The transmission projects alone will take years to complete, meaning the emergency pairing mechanism is the critical near-term fix. For now, the market's price signal shows the cost of this imbalance is already being paid. The coming months will test whether these practical steps can close the gap before the next wave of demand hits.

Catalysts and Risks: What to Watch for the Balance

The success of PJM's emergency plan hinges on a series of near-term events and the resolution of persistent structural risks. The clock is ticking, and the market will judge the grid operator's response against a backdrop of accelerating demand.

The most immediate test is the matching process for new generation, which must run from September 2026 to March 2027. This compressed 18-month window is the critical near-term fix. Any significant delays in pairing proposed data centers with new power plants here would signal that the underlying bottlenecks-permitting, financing, and construction-are proving harder to overcome than anticipated. It would directly undermine the plan's purpose and likely keep capacity prices elevated.

A second key catalyst is the effectiveness of the new regulatory framework. The compliance filing for the Behind-The-Meter Generation (BTMG) tariff, submitted in February 2026, aims to deliver finality by July 31, 2026. If this rule change fails to accelerate on-site generation by data centers, the grid's burden will only increase. The market monitor has already argued that requiring new data centers to bring their own generation is a necessary step to balance the system and avoid overbuilding transmission. The July deadline is a clear marker for whether this structural fix can be implemented in time.

The overarching risk, however, is that demand simply outpaces all planning. PJM's official 2026 Long-Term Load Forecast projects summer peak demand will rise to 253 GW by 2046. Yet, a recent analysis by Wood Mackenzie suggests utilities within the PJM footprint have collectively forecast 55 gigawatts of new large load growth by 2030 and 100 GW by 2037. This gap between official forecasts and utility-level projections is a red flag. If data center demand growth exceeds the 2046 forecast of 253 GW-especially if AI compute intensity rises faster than expected-the entire resolution strategy could be rendered insufficient. The grid's ability to manage this imbalance will be tested not just by the 15-gigawatt plan, but by the relentless pace of demand itself.