Shanthi Muthiah is managing director of energy strategy and solutions at ICF. Himali Parmar is vice president of energy markets at ICF.
As data center developers confront interconnection queues stretching years, and utilities confront surging demand growth and lagging supply, traditional models of load service are proving inadequate. In the next few years, in markets where supply and demand are fundamentally out of balance, grid connection increasingly comes with a choice: either bring the needed power yourself, or bring flexibility.
FERC's December 2025 order to the PJM Interconnection is particularly significant, directing PJM to create new transmission service options, including non-firm service and pathways for dedicated on-site generation. Additionally, as long as supply lags demand, large loads that do not bring their own new generation will be curtailed first in an emergency. Against this backdrop, load flexibility can indeed help address reliability concerns and enable load growth — but with important tradeoffs and cautions to be understood and addressed.
PJM constraints, and how flexibility can close the gap …
PJM’s own load growth forecast calls for approximately 30 GW of new demand by 2030, a figure the RTO is straining to meet. The latest 2027/2028 base residual capacity auction cleared with a shortage, and the North American Electric Reliability Corp.’s outlook for PJM from 2029-2030 places it in the “high risk” category for reliability concerns. PJM expedited interconnection with the Reliability Resource Initiative, or RRI, program in 2025 and has recently drafted the framework for an emergency auction to try to source further new supply, with uncertain outcomes. Simply put, utilities and grid operators must plan around a demand trajectory they cannot serve with the current generation pipeline.
ICF modeled what PJM can realistically add in generation capacity through 2030: 16.2 GW of net new unforced capacity, with the majority coming from RRI resources. Running the numbers forward with this assumption, the grid can support only about 13.5 GW of additional firm load, or less than half of PJM’s current load forecast. However, when load flexibility is layered as a tool on top of that 13.5 GW firm load ceiling, PJM can, in theory, reliably absorb an additional 30 GW without further new generation, meeting and even exceeding its current load forecast.
… but not without risks.
The anchoring point in the analysis is curtailment among flexible large load consumers averaging ~80 hours annually. However, this risk is not evenly distributed. In most scenarios, minimal curtailment is required, but in extreme conditions, curtailment could be several times higher, exceeding 300 to 400 hours a year. This is before accounting for any additional curtailment triggered by local transmission issues, which are particularly likely before interconnection-related transmission upgrades are completed.
Distributing this risk is a point of contention. Data center companies have called for limiting parameters or caps as part of flexibility frameworks to help bound their risks. Utilities and consumer groups, on the other hand, argue that caps would result in spillover and lower reliability for residential and commercial consumers.
Additionally, there are cost concerns. As capacity is increasingly utilized even in currently unconstrained hours, less efficient generators are forced to run more frequently, pushing wholesale energy prices higher. Price impacts will vary by region, and wholesale increases do not pass through uniformly to retail bills — transmission, distribution and regulatory cost recovery mechanisms all shape what consumers ultimately pay. That said, the directional pressure is real and sustained load growth without commensurate supply additions will put upward pressure on prices.
Every stakeholder has a role
Stakeholders today face operationalizing load flexibility to solve near-term constraints and setting the preconditions today that move the grid away from a curtailment-reliant system over time.
Utilities have several avenues to structure solutions that spread benefits across stakeholders.
Load flexibility does not necessarily have to come from the data centers themselves. Existing demand response programs can be expanded to create more headroom. Retail DR can provide lower-cost flexibility: for example, Public Service Co. of Oklahoma's program procured capacity at $32/kW-year (versus $100+/kW-year for BTM generation at a data center).
Proactive engagement with data centers on siting, sizing and operational decisions matters. Providing public data on transmission availability, risk assessments and defined conditions under which curtailment occurs can enable better decisions. The most valuable shift would be moving beyond generic DR toward specific contracts structured around engagement at the workload level, distinguishing disruption-tolerant operations from mission-critical ones.
Data centers must engage as genuine grid partners: bringing their unique visibility into workloads to the table, co-investing in or directly funding utility DR programs (which can simultaneously help address affordability concerns) and participating in regulatory reform (an area that FERC Chairman Laura Swett recently highlighted as a gap). This applies across the spectrum of large loads. Large players with the sophistication and balance sheet to lead should be doing so more proactively.
ISO/RTOs must design now for the current reality that supply and transmission lags load growth. Creating services to operationalize, and potentially compensate, loads for curtailment can provide transparency and ensure fairness across wider footprints than single-utility areas.
Removing hurdles to interconnection can help resolve the supply/demand imbalance more quickly. For example, the Midcontinent Independent System Operator and PJM have fast-tracked shovel-ready generation projects (via ERAS and RRI initiatives, respectively). SPP has gone further — its CHILL service delivers conditional grid access in 90 days, and HILLGA allows parallel study of load and supporting generation.
State and federal policymakers can ensure that the ISOs and utilities have legal frameworks and appropriate mandates to implement flexibility and data center-specific tariffs to protect ratepayers.
At the federal level, promoting reform and standardization of interconnection standards is critical. While generation queues have been refined through decades of rulemaking, the process' maturity for load interconnection is a decade behind. Requirements and processing vary widely across transmission owners. FERC’s order to PJM is constructive and should be emulated in other regions.
Getting this right serves everyone who depends on a reliable, affordable grid, and builds the foundation for a decade of economic growth.
