Data Centers Now Drive 40 Percent of PJM’s Capacity Costs

PJM Interconnection’s December capacity auction for the 2026/2027 delivery year produced a price of $329.17 per megawatt-day, the highest in the market operator’s history. The total cost was $16.4 billion. Of that, $6.5 billion — 40 percent — is attributable to data center load.

Roughly $6.2 billion of those costs relate to data centers that have not been built yet but could come online by 2027 or 2028. PJM’s capacity market is forward-looking by design: it procures generation resources to meet projected demand three years out. The auction result means that PJM expects enough data center construction to fundamentally reshape the demand side of a 13-state grid within three years, and that the generation side of the equation has not kept up.

What the price signal means. A capacity price of $329/MW-day is roughly five times the historical average for PJM auctions. Capacity costs flow through to retail electricity rates. When 40 percent of those costs are driven by a single customer class that barely existed a decade ago, the price signal carries implications well beyond data center economics.

For utilities, $329/MW-day changes the math on which generation resources are economic to build or retain. Natural gas peakers that were marginal at $100/MW-day are suddenly profitable. Battery storage systems that provide 4-hour capacity earn substantially more. Every demand response program becomes more valuable. The high clearing price is not a malfunction — it is the market telling participants that PJM needs more capacity, urgently, and is willing to pay for it.

For ratepayers, the question is simpler and more uncomfortable. Consumer advocates in PJM territory have already argued that residential and small commercial customers should not subsidize power infrastructure for hyperscale data centers. If the next auction clears higher, the political pressure to make data centers pay their own infrastructure costs will intensify.

Contrast with Texas. While PJM’s auction signaled scarcity, ERCOT data tells a different story. Modo Energy reported that battery interconnection applications in ERCOT dropped 50 percent in the second half of 2025. Only 85 percent of projects with signed interconnection agreements are expected to be built as queue timelines stretch beyond four years. Texas, which led the nation in storage deployment through 2025, is entering a phase where market saturation and development risk are tempering new investment.

The divergence between PJM and ERCOT illustrates a fundamental point about energy storage markets: they are local. The capacity shortage driving record prices in the Mid-Atlantic does not exist in Texas. The market signals that make a battery project compelling in northern Virginia may not apply in West Texas. National deployment statistics obscure regional dynamics that determine which projects actually get built and which earn returns.

The deployment record. Against this backdrop, Tesla reported a Q4 2025 storage deployment record of 14.2 GWh, bringing its full-year total to 46.7 GWh — up 49 percent year-over-year. This is notable partly because Tesla’s EV deliveries fell 16 percent over the same period. Energy storage has become the growth business inside Tesla, with record profit margins for the fifth consecutive quarter.

In Australia, Synergy completed the 500 MW / 2,400 MWh Collie Battery, the country’s largest operational BESS, powered by CATL cells. The project followed Australia’s aggressive 12 GWh long-duration storage tender results. LG Energy Solution now targets 90 GWh in storage orders and over 60 GWh in production capacity — 80 percent of it in North America — by the end of 2026. The supply side is scaling to meet the demand signal from markets like PJM.

The domestic content ratchet. The IRS raised the domestic content threshold for BESS projects beginning construction in 2026 to 50 percent of manufactured product costs, up from 40 percent. Direct payment recipients that begin construction in 2026 or later will receive no refundable ITC credit without meeting the threshold. This is not a bonus — it is a gate.

The 50 percent requirement creates immediate pressure to source cells, enclosures, and inverters from domestic manufacturers. The Qcells-LG Energy Solution 5 GWh deal announced this week, using LFP batteries from LG’s US factory, is a direct response to this policy signal.

Wholesale market access. FERC has set a December 31, 2026 deadline for NYISO to achieve full compliance with Order 2222, enabling distributed energy resource aggregations to participate in energy, ancillary services, and capacity markets. ISO-NE’s energy and ancillary services implementation is set for November 1, 2026, with capacity market participation beginning February 2027.

These deadlines matter because they will allow aggregated commercial storage systems to earn revenue in wholesale markets for the first time in the Northeast. When combined with behind-the-meter demand charge savings and utility incentive programs, wholesale market access creates a third revenue layer that changes the investment thesis for commercial-scale batteries.

The data center premium. PJM’s $329/MW-day auction result is a price assigned to a future that the market expects but cannot yet serve. Data centers consumed roughly 2 percent of US electricity in 2022. By 2030, that share is projected to reach 6 to 9 percent. The grid infrastructure required to serve that growth does not exist, and the interconnection queue to build it stretches years.

Battery storage fills the gap — not by replacing generation, but by managing when power is consumed. A 4-hour battery at a data center shifts load from peak periods that drive $329/MW-day capacity prices to off-peak periods that cost a fraction as much. The economics are circular: data centers drive up grid costs, which makes storage more valuable, which makes storage more economic to deploy at data centers.

PJM’s auction made that circular logic visible in a single number.

Sources