On July 14, 2026, PJM Interconnection — the grid operator for 65 million people across 13 mid-Atlantic and Midwest states — closed its Base Residual Auction for the 2028/29 delivery year at $325 per megawatt-day, the maximum price its own price collar allows. It was the third consecutive auction to clear at or near the ceiling, and the second straight year the region came up short of its own reliability target, missing by 6,831 megawatts. The total bill for that capacity: $16.4 billion. If you are negotiating a colocation contract, a build-to-suit lease, or a power purchase agreement anywhere inside PJM territory, this auction result is not background noise — it is the clearest signal yet of what your power costs will look like in 2028, and it tells you exactly who regulators think is driving the shortfall.
What actually happened in the July auction
PJM’s capacity market works like an insurance policy: generators bid to be available during the following year’s highest-demand hours, and the clearing price becomes the capacity charge baked into everyone’s power bill. The official PJM results release shows 138,318 MW of generation capacity procured through the main auction, with committed capacity of 152,083.4 MW UCAP against a reliability requirement of 156,013 MW UCAP — a 6,831 MW gap, wider than the roughly 6,500 MW shortfall in the prior auction for 2027/28, which Utility Dive reported was PJM’s first-ever reliability miss.
The price itself is capped by a FERC-approved collar, and that cap did real work this time. Utility Dive’s reporting shows the auction would have cleared around $555/MW-day without it — and as high as $777/MW-day in the ComEd zone covering northern Illinois. With the collar, PJM’s total capacity cost came to $16.4 billion; without it, the same capacity would have cost an estimated $29.7 billion. That collar is the only reason your 2028 bill isn’t already 80% higher than it is.
How much of this is actually data centers
PJM COO Stu Bresler has been direct about the cause: the 2028/29 peak load forecast came in roughly 2,000 MW higher than the prior auction’s forecast, and that increase is attributed mainly to new large loads — overwhelmingly data centers. Of the 3,613 MW year-over-year increase in the reliability requirement, PJM itself attributes 1,374.5 MW UCAP directly to Large Loads.
Monitoring Analytics, PJM’s independent market monitor, put a dollar figure on it. In reporting from Industrial Info Resources, Monitoring Analytics president Joe Bowring estimated that roughly $6.3 billion of the $16.4 billion in total 2028/29 capacity payments is attributable to data center load. That’s not a rounding error on a regional power bill — it’s the single largest identifiable driver of this auction’s cost, and it’s a number regulators, state consumer advocates, and competing load-serving entities are now quoting back at PJM and at hyperscalers directly. This builds on the broader cost-allocation fight already underway at FERC over who pays for the transmission upgrades data centers require — the capacity market is simply a second front in the same argument.
It’s worth noting the counterargument PJM and generation owners make: demand growth of any kind, from reshoring manufacturing to electrification of heating and transport, tightens a capacity market that has also lost firm generation to retirements. Data centers are the fastest-growing and most visible piece of that demand, but Aurora Energy Research’s Julia Hoos argued to Utility Dive that the deeper problem is structural — the market isn’t bringing new supply or demand response online fast enough, regardless of which load category is growing. She warned regulators risk triggering what she called an “intervention doom loop” if they respond to this auction with ad hoc fixes rather than market redesign.
How the cost actually lands on a buyer’s bill
This is the part most procurement teams underestimate: capacity costs for a facility inside PJM typically aren’t a flat per-megawatt-hour charge. They’re based on your site’s Peak Load Contribution (PLC) — your facility’s average demand during the five highest-demand hours across the five highest summer peak days, known as the 5 Coincident Peak or 5CP methodology, as Vistra’s commercial energy team explains. A data center that runs flat around the clock, which is most of them, has little ability to shed load during those five hours the way a manufacturing plant can — which means its PLC, and therefore its capacity bill, tracks close to its full nameplate draw.
Need Expert Guidance?
Talk to a Data Center Expert
21+ years of hands-on experience in data center design, operations & infrastructure. Book a quick discovery call to discuss your project.
📞 Book a Discovery CallThe dollar impact is already showing up in real contracts. Peter Cavan of Unison Energy told Utility Dive that a 10-megawatt industrial customer’s monthly capacity charge is on track to rise from roughly $6,000 in 2024 to approximately $70,000 by 2028 — a more than tenfold increase driven almost entirely by these consecutive price-capped auctions. Scale that ratio to a 50 MW or 100 MW campus and the capacity-charge line item alone moves from a rounding error to a material budget line that needs its own forecasting process, not an afterthought buried in a power purchase agreement.
The reliability math behind that bill is also getting harder, not easier. POWER Magazine’s analysis of the auction notes that new generation and uprates cleared at only about 525 MW this round — down from 774 MW in the December 2025 auction — while cleared demand response actually fell by 277 MW to 7,365 MW. In other words, the supply side of the market is shrinking its contribution at the same moment large loads are expanding theirs. PJM’s Stu Bresler has also flagged that always-on data center demand is pushing peak grid risk later into the winter months, a seasonal shift that changes how utilities plan maintenance outages and how buyers should think about curtailment exposure outside the traditional summer peak window.
What PJM is trying to do about it — and why buyers shouldn’t wait
PJM’s board isn’t sitting still. Per Utility Dive’s reporting, PJM was preparing FERC filings for a backstop capacity auction intended to run in September 2026, alongside a “connect and manage” framework specifically aimed at bringing data center load online with more predictable timelines and cost responsibility. Cavan’s blunt assessment of the 2029/30 outlook: “Do not expect this to materially change.” Advocacy groups like NRDC have pushed a sharper version of the same point, arguing PJM should require large loads — specifically data centers — that haven’t brought their own new generation to the table to bear more of the cost they’re creating, rather than spreading it across the general ratepayer base through the capacity market and transmission charges alike.
None of this is reason to panic, but it is reason to model capacity exposure explicitly rather than treating it as a line item that will sort itself out. For a buyer evaluating a PJM-territory site today, that means asking for your PLC methodology and 5CP exposure in writing before signing, stress-testing a 2029/30 capacity line that could repeat the pattern Cavan described, and factoring the real possibility of a backstop auction or new large-load tariff class changing your cost structure again within 24 months. The auction result is a known quantity now — the open question is how fast regulators act on the gap between what data centers are drawing and what they’re paying for directly.
Buyers who want a structured way to run these numbers before signing a PJM-territory contract can start with TechInfraHub’s data center buyer’s toolkit.
Sources
- PJM Interconnection — “PJM Capacity Auction Procures 138,318 MW of Generation Resources” (July 14, 2026 press release)
- Utility Dive — PJM 2028/29 capacity auction price cap and reserve shortfall coverage
- POWER Magazine — “PJM’s 2028/2029 Capacity Auction Falls Short of Reliability Standard for Second Straight Year”
- Industrial Info Resources — Monitoring Analytics estimate on data center share of PJM capacity costs
- Vistra Commercial Energy — Peak Load Contribution and 5CP methodology explainer
Written by
Raajeev Ratra
Data Center Infrastructure Expert | 21+ Years in DC Design, Operations & Project Management
Raajeev is a seasoned data center professional with hands-on experience in hyperscale facilities, colocation design, power & cooling infrastructure, and global DC operations. He shares practical insights to help engineers and IT leaders build better infrastructure.