PJM ties a 6.8GW capacity backstop to the cost of large-load growth

PJM Interconnection has proposed allocating the cost of a future reliability backstop according to the growth of large electricity loads, as demand from data centres adds pressure to its capacity market. The October 29 amendment addresses a 6,831-megawatt shortfall for the 2028–29 delivery year and proposes opening bids on March 1, 2027. The decision connects two questions facing AI infrastructure developers: how additional reliable supply will be secured, and which customers will finance it.[1][3]

PJM serves parts or all of 13 US states and the District of Columbia. Its capacity market pays resources to be available during a future delivery period, separately from the day-to-day sale of electricity. The 6,831MW figure is expressed in unforced capacity, or UCAP: capacity credited under PJM’s reliability rules. It measures the gap against a future procurement requirement, rather than a shortage at a particular hour today or the nominal electrical size of a new data-centre campus.[1][2]

How the proposed backstop would work

The amended Reliability Backstop Procurement would seek resources that existing arrangements did not secure, while allowing qualifying bilateral supply contracts to reduce the amount bought through the backstop. PJM proposes a maximum willingness to pay of an MW-weighted average $555 per megawatt-day and commitments lasting up to 15 years. Those terms concern payment for capacity availability, rather than a retail price per unit of electricity consumed.[1][2]

The schedule is specific. PJM says the revised process would start accepting bids March 1, close the bid window March 28, determine the final procurement target and supply amounts on April 28 and announce results May 10. Each date belongs to a proposed future sequence. On September 29, the Federal Energy Regulatory Commission largely accepted the earlier filing but suspended execution for five months while particular issues were resolved. The October 29 amendment addresses that pause. Readers should therefore distinguish a formally filed plan from an already completed acquisition of new generating capacity.[1]

Who would carry the cost?

PJM proposes basing allocation on large-load demand growth between the actual summer peak in 2026 and the forecast for the 2028–29 delivery year. The obligations would run through the relevant load-serving entities under the proposed rules. This matters to both developers and other customers: spreading the cost broadly could make residents and unrelated businesses finance capacity associated with new large users, while allocating it to that growth could change the economics of locating new computing facilities.[1]

The amendment also revises credit and collateral requirements, the responsibilities of load-serving entities that exit PJM, and rules for demand reduction and distributed resources to offset a need for new capacity. These are not administrative footnotes. A long-term capacity contract creates payment obligations even if a buyer changes its business plan or withdraws from a market. Collateral is intended to protect against default. Demand management could reduce the amount of generation procured if it reliably shifts or cuts peaks, but its value must be measured against actual operating obligations, not an untested promise that AI centres can always turn off machines.[1]

Connecting site plans with capacity obligations

An AI-campus announcement may quote connection capacity, IT load or the eventual size of a multi-phase site. Those figures cannot be compared directly with an accredited capacity-market shortfall without aligning the delivery year, load forecast and reliability assumptions. PJM plans to update the procurement target as new commitments become firm. Its approximately 6,800MW description is the rounded version of the 6,831MW shortfall, not an additional deficit.[1][2]

The broader issue is the speed of construction. A data centre can sometimes commit capital and bring computing equipment online more quickly than new generating resources, transmission upgrades and permits can be completed. Large-load customers may respond by financing dedicated generation, accepting demand limits or negotiating firm supply contracts. Each approach shifts risk among developers, utilities and consumers. PJM's amended plan is an attempt to make that risk allocation explicit before shortages become operational emergencies. Its final design will depend on regulatory decisions and actual commitments, not on the power needs announced in corporate presentations.[1][3]

The next decisions concern FERC’s response to the amendment, firm large-load forecasts and the bilateral supplies credited against the target. Bids and eventual contracts would then show which resources the backstop actually secures. For developers, the proposed schedule creates a route from announced power demand to a defined funding obligation before the 2028–29 delivery year.[1]

Report an error or suggest a correction