PJM Grid Risks 2030 Reliability Crisis From AI Data Center Load
Pennsylvania PUC report warns modeled electricity shortages could exceed PJM criteria by 100 times under high load growth.
A study commissioned by the Pennsylvania Public Utility Commission warns that rapid expansion of AI data centers could push the 13-state PJM Interconnection power grid into severe reliability deficits by 2030. In a worst-case scenario with constrained generation buildout, modeled loss-of-load expectations reach more than 13 days per year with potential electricity shortages. The findings underscore how surging electricity demand from hyperscale facilities threatens regional grid stability and long-term supply planning.
Key details
The report, released by Pennsylvania Public Utility Commission (PUC) Chairman Steve DeFrank, assesses PJM's resource adequacy across multiple demand trajectories through 2040. Under the baseline reference scenario using PJM's 2026 forecast, the modeled 2030 loss of load expectation (LOLE) reaches 0.59 days per year—nearly six times worse than PJM's official planning criterion of one shortage event every ten years (0.10 days per year). Under a worst-case scenario of accelerated data center additions and delayed generation deployment, the LOLE rises to 13.20 days per year, representing shortage risks over 100 times higher than the target threshold. Pennsylvania, historically a major net energy exporter in PJM, is projected to see its net annual exports shrink from 91 TWh in 2025 to 69 TWh by 2035 and 38 TWh by 2040 in the reference case. Under high-load conditions, the state could become a net energy importer of 5 TWh annually by 2040, creating unmet regional load requirements if new power plants fail to materialize.
Why this matters
The study provides quantitative evidence that the rapid expansion of AI compute clusters is outpacing regional power generation and transmission capacity. As hyperscalers reserve gigawatt-scale grid connections, the resulting supply-demand imbalance threatens systemic grid reliability and exposes residential utility customers to heightened risk of emergency power curtailments or elevated capacity charges.
Context
This report follows a series of regulatory interventions across PJM's footprint, including Pennsylvania Governor Josh Shapiro's recent executive order mandating that large data centers supply their own new power generation. PJM capacity auction prices recently hit historical caps due to rising demand forecasts, while federal regulators at FERC and state commissions have increasingly scrutinized how large-load additions impact regional energy markets and ratepayer equity.
What happens next
The Pennsylvania PUC voted unanimously to direct commission staff to draft updated emergency curtailment rules and convene a technical conference on data center cost allocation. PJM and state officials will also continue evaluating load forecasting methodologies and reliability backstop procurements, while the U.S. Department of Energy considers extending operational lifespans for several regional coal-fired power plants to maintain near-term reserve margins.
Source: Utility Dive Published on AI Usage Global, author: AUG Bot
