The Texas battery fleet earned $2.03 per kilowatt-month in October 2025. Twelve months earlier, the same fleet was paid 40% more for the same work. The fleet that took that revenue away is the fleet itself.
That growth is compressing its own economics. Electric Reliability Council of Texas (ERCOT) battery revenue fell 40% in a year, and ancillary payouts dropped from $1.52 to $0.50 per kilowatt-month.
The trade is rotating toward contracted capacity and multi-day assets. The market-design reforms landing through 2027 decide who profits.
Storage was sold to investors as a simple trade: buy power when it is cheap, sell when it is scarce, collect the spread. That trade works until enough batteries are doing it at once. Every new megawatt flattens the same spike the next megawatt was counting on.
Modo Energy's settlement data shows the mechanism in plain numbers. Batteries in Texas charge so hard into the midday solar glut that they lift their own buying price, and they discharge so hard into the evening that they set the selling price themselves. This map covers what is growing, what is falling, what is new, and where the money actually lands.
Battery earnings are sliding
ERCOT battery energy storage settled at $2.03 per kilowatt-month in October 2025, 40% below the same month a year earlier. · Modo Energy, 2025
The build-out is accelerating
The world added 108 GW of new battery storage in 2025, 40% more than in 2024; installed capacity is now eleven times the 2021 level. · IEA Global Energy Review 2026
Merchant dependence is broad
41% of US utility-scale battery capacity lists price arbitrage as its primary use; 66% uses it in some form. Frequency regulation is primary for 24%. · EIA, 2025
Growing: a fleet that doubled in a year
The build-out has no precedent. The International Energy Agency (IEA) counts 108 GW of new battery storage deployed worldwide in 2025 in its Global Energy Review 2026, 40% more than in 2024. Installed capacity now stands eleven times higher than in 2021. Roughly 80% of what was added last year is utility-scale, and about 90% of it uses lithium iron phosphate (LFP) chemistry, a share that sat below half as recently as 2021. China took about 60% of global additions.
The US numbers follow the same line. The Solar Energy Industries Association (SEIA) recorded 57.6 GWh of new storage in 2025, the largest single year on record, and 9.7 GWh in Q1 2026, up 32% year on year. The US Energy Information Administration (EIA) expects utility-scale storage to roughly double between the end of Q1 2025 and the end of 2026, to about 65 GW.
Data centers have become the demand anchor of this build-out. Battery-based UPS capacity grew 30% last year to 45 GW, and SEIA's forecast of 613 GWh of US deployment by 2030 is built on data center demand. Capital is following: Q1 2026 battery deals reached 17 transactions at 2.4 GW with disclosed debt above $2.9 billion, and 60% of that volume sat in ERCOT.
That is also where the revenue problem is sharpest.
Falling: the revenue stack loses its legs
Modo Energy's ERCOT settlement data for October 2025 is the cleanest illustration of the problem: $2.03 per kilowatt-month, a 40% drop from October 2024. August 2025 was worse: $2.27 per kilowatt-month, 69% below August 2024. Installed capacity in the region grew by more than 4 GW over the same stretch.
Two forces do the damage. Charging costs more: average daily peak charging in ERCOT rose from 1.7 GW to 3.3 GW, so batteries now lift the prices they buy at. Selling pays less: batteries increasingly set the peak price themselves, and each one bids the next one down. Modo's own phrasing is blunt: batteries "reduced peak prices and cannibalized their own opportunity."
The ancillary stack is draining faster. Frequency regulation and reserve payouts in ERCOT fell from $1.52 per kilowatt-month in October 2024 to $0.50 a year later. Day-ahead spreads tightened 38% and real-time spreads 14%, even as peak demand rose and gas prices climbed 41%. The compression came from the supply side: more batteries chasing the same few hours.
California's Independent System Operator (CAISO) runs the same playbook. The growing battery fleet flattens midday prices and crowds the evening ramp, where a four-hour box once had the whole peak to itself. Average merchant revenues have retreated from their peak years, and the spread between average and top-quartile operators has widened into a chasm. Optimizers that skip the low-value hours and chase scarcity keep far more of their revenue; the set-and-forget arbitrage schedule no longer works.
The pattern is structural. Storage build-outs eat their own markets everywhere they scale.
New: long-duration money and market redesign
Merchant arbitrage is a shrinking slice of the stack. The growth is in contracts and in assets that do not depend on the same four-hour window.
Form Energy is the clearest case. Its iron-air battery discharges for 100 hours, long enough to cross a multi-day wind drought, at a cost target near $20 per kilowatt-hour. As we wrote in August, the market once wrote the concept off; the $750 million raise and the factory in Weirton, West Virginia changed the story. The plant is ramping toward 500 MW a year, more than 75 GWh of commercial projects are under agreement, and Crusoe has signed for 12 GWh to back AI data centers starting in 2027. Xcel, Georgia Power, Great River Energy and Dominion all have pilot or construction contracts in place.
Iron-air pays for duration in efficiency: it returns only 40-50% of the energy it takes in, versus 85-95% for lithium. That trade becomes acceptable when the input is curtailed solar that would otherwise be discarded at near-zero cost, and when the payout is a $1,000-per-megawatt-hour scarcity hour a lithium box could never reach.
The other shift is market design. Six of seven US grid operators are mid-reform: RTC+B in ERCOT, Slice of Day in CAISO, the PJM regulation redesign, MISO's ERAS process, NYISO's Index Storage Credit. Each one changes what a battery is paid for, and each is a different bet on where value survives.
Operators that hold contracted capacity and multi-day assets keep their revenue. Merchant-only four-hour boxes absorb the compression.
The revenue stack, compared
Storage revenue splits into four streams with very different trajectories.
| Revenue stream | What pays for it | Trend, 2025-26 |
|---|---|---|
| Energy arbitrage | Price spread between surplus and scarcity hours | Dominant. ERCOT arbitrage share of revenue roughly tripled in a year, but spreads are compressing (day-ahead −38%) |
| Frequency regulation | Keeping the grid at 60 cycles per second | Primary use for 24% of US capacity; payouts fall as more batteries compete |
| Ancillary and reserve | Emergency response and headroom | ERCOT payouts fell from $1.52 to $0.50 per kilowatt-month |
| Capacity and Resource Adequacy | Contracted availability | Rising share. CAISO leans on RA contracts; reforms through 2027 define the terms |
What to watch
Four signals tell you whether the compression is bottoming or continuing.
ERCOT ancillary payouts. If they hold near $0.50 per kilowatt-month, merchant-only projects retrade lower.
Who sets the peak price. Batteries, or gas peakers back on their feet.
The contracted share of new deals. Capacity and Resource Adequacy volume versus merchant exposure.
The interconnection queue. 749 GW of storage was waiting to connect at the end of 2025. The queue is the brake pedal on the cannibalization curve: clear it fast and spreads compress faster; keep it clogged and existing operators hold their peak hours a little longer.
The trap for new capital is reading the record deployment as proof of the business model. The deployment is proof of the problem. The revenue that justified the first gigawatt-hours is being consumed by the gigawatt-hours that followed. What survives is contracted capacity, multi-day assets, and operators disciplined enough to skip the hours everyone else is chasing.