Two October days, two very different markets. In California, an unseasonable heat wave pushed load to just above 46 GW on October 5 — within 10 percent of the year’s summer peak — and real-time prices barely stirred, topping out at $209.31/MWh the next day and failing to crack $100/MWh on most days of the week-long event. Eight days later, in the East, a mild autumn Tuesday with a peak load of only 108 GW sent real-time prices at PJM’s Western Hub above $1,500/MWh and pushed the U.S. Department of Energy to issue an emergency order. The contrast is the clearest illustration yet of what battery storage does to scarcity pricing — and of what happens in markets that have not yet reached the inflection point.
The mechanism is worth stating plainly, because it is often described as if it were magic. Batteries do not create energy; they move it across hours. When solar floods the midday market, batteries charge, absorbing surplus that would otherwise depress prices or be curtailed. When solar fades and evening demand peaks, they discharge, adding supply in precisely the hours when the system is tightest. In real-time markets, where prices are set every five minutes by the marginal unit, a fleet of batteries discharging into the evening ramp is, in effect, a fleet of fast, zero-fuel peakers that bid against the gas turbines that used to set those prices. Add enough of them, and the scarcity hours that once produced four-figure prices start to flatten.
California’s October heat wave that wasn’t a price event
Start with the California numbers, because they are startling on their own terms. Load peaked just above 46 GW on October 5, driven by unseasonably warm weather. For context, the 90th percentile of historical weather conditions in early October would translate to an expected load peak of about 35 GW, with a median expected peak of about 31 GW. The year’s overall summer peak was 49.9 GW on September 9 — less than 10 percent higher than the October reading. By any historical standard, this was a summer-grade stress event arriving in the shoulder season.
And yet real-time energy peaked at $209.31/MWh on October 6, with most days during the week-long heat wave failing to crack even $100/MWh. To appreciate how unusual that is, look at the recent history. From 2019 through 2023, CAISO saw 24 days with real-time energy prices above $800/MWh, including six days above $1,400/MWh, in a market where energy offers are capped at $2,000/MWh. Since 2025, there have been no days with real-time prices above $700/MWh and only three days with peaks above $250/MWh — against 109 such days in the 2019–2023 period.
Correlation is not causation, and other factors plainly contributed: hydro conditions vary year to year, and coordination with neighboring balancing authorities through the Western Energy Imbalance Market and the Extended Day-Ahead Market has improved. But the storage overlay is stark. Since 2018, the maximum battery discharge in CAISO has risen from roughly 100 MWh to over 13 GWh. The fleet that barely existed eight years ago now discharges, at its peak, an amount comparable to a meaningful share of the entire evening ramp. The price spikes did not disappear because California stopped having hot days. They disappeared, in large part, because the system now has something to throw at the hours when prices used to spike.
Texas: the same pattern, on fast-forward
ERCOT tells the same story with a compressed timeline, which is what makes it instructive. Battery growth in Texas started much later — peak discharge was only 1.4 GWh in early 2024 — but once available battery assets crossed the 7.5 GW threshold, the pattern repeated. From March 2021 through September 2025, the system-average hub saw 110 days with a peak real-time price above $1,000/MWh, including 46 days above $3,000/MWh. Since September 2025, there have been no spikes above $1,500/MWh and only 13 days above $500/MWh.
The timeline matters because it answers the obvious objection: that CAISO’s calm October was a weather or hydro fluke. Two different markets, with different designs — CAISO has a capacity mechanism and a $2,000/MWh offer cap; ERCOT runs the energy-only model described earlier in this series, relying on scarcity pricing through its operating reserve demand curve — both saw extreme real-time prices fade as storage crossed a threshold of installed capacity. The common variable is the fleet, not the market design.
It took CAISO approximately three years to grow from 1.5 GW to 7.5 GW of battery power. ERCOT did the same in about 18 months. That acceleration is the reason the eastern markets deserve attention now rather than later: the build-out, once it starts, may not take as long as the western experience suggests.
October 13 in the East: what scarcity looks like without storage
Now the other October day. On Tuesday, October 13, PJM’s Western Hub saw hourly real-time prices above $1,500/MWh. The instinct is to assume a heat wave or a cold snap, but the weather was unremarkable. Peak load was 108 GW, well below the summer peak of 162 GW set on July 2. Net load — load minus wind and solar output — was about 104 GW, against roughly 154 GW at a typical summer peak. By the demand measures, this was a mild day.
The driver was the calendar, not the thermometer. October is generator maintenance season: the shoulder months are when plant owners take units offline for the work that cannot be done during summer or winter peaks. PJM reported 46 GW of planned outages and 60 GW of total outages on October 13. When outages are added back to demand, the picture changes completely. The effective load forecast — demand plus unavailable capacity, the measure that actually describes how tight the system is — reached 167 GW, the 12th highest in PJM since 2016. PJM had originally forecast 54 GW of planned outages six days ahead; by Saturday that had dropped to 48 GW, and finally to 46 GW, as the operator called on generators to return to service where they could. It was not enough.
MISO was running tight at the same time. Its peak effective load of 135 GW on October 13 blew past the recent record of 133 GW set in August 2023, triggering a Conservative Operations declaration, and real-time prices at the Indiana Hub reached nearly $1,000/MWh — less extreme than PJM, but high enough to limit the imports that might have relieved PJM’s shortage. The Department of Energy, acting on a PJM application filed October 11, issued an emergency order on October 12 authorizing PJM to dispatch specified units and direct backup generation through October 16, with PJM expecting to come within 900 MW of its reserve margin on October 14.
The point is the shape of the event. This was a scarcity event manufactured by the maintenance calendar, not by weather — and it produced exactly the kind of extreme, sustained real-time prices that California and Texas have largely stopped seeing. The eastern markets have not yet built the shock absorber.
How far the East is from the inflection point
The storage gap is large. In 2026, MISO’s peak battery discharge reached about 922 MWh. SPP and PJM both stayed below 500 MWh. Set those figures against CAISO’s peak discharge above 13 GWh and ERCOT’s above 13 GW of installed capacity, and the scale of the difference is plain: the eastern fleets are, for now, roughly an order of magnitude smaller relative to the systems they serve.
Market size cuts both ways here. PJM and MISO are much larger systems than CAISO or ERCOT — PJM’s summer peak alone is more than three times CAISO’s — so they may be able to absorb more storage capacity before reaching a saturation point. A gigawatt of batteries means something different in a 160 GW system than in a 50 GW system. But the October 13 event shows that the relevant comparison is not system size; it is the size of the tight hours. The hours that set extreme prices are the hours when tens of gigawatts of capacity are offline for maintenance and the remaining fleet is stretched. In those hours, even a modest storage fleet discharging into the evening would have clipped the peak.
The paradox: storage eats its own revenue
Here is the uncomfortable part of the story, and the reason this column treats the saturation point as a genuine economic problem rather than a victory lap. Batteries earn much of their energy-market revenue from the spread between low-priced charging hours and high-priced discharging hours. When a large fleet charges at midday and discharges in the evening, it raises midday prices and lowers evening prices — compressing the very spread it lives on. Build enough storage to eliminate scarcity pricing, and you may eliminate the financial conditions that made the storage economic in the first place.
The ERCOT numbers trace this arc precisely. Using a peak-to-trough calculation — charging in the lowest-priced hours and discharging in the highest, assuming perfect round-trip efficiency — one-hour and two-hour storage resources saw potential energy-market revenues more than double between March 2021 and the end of 2023. Since then, as maximum discharge grew past 13 GW, peak-to-trough values have fallen to levels 25 to 30 percent below their 2021 starting point. That calculation excludes ancillary services revenue, but those prices have fallen sharply as well, as the earlier discussion of regulation and reserve markets would predict: the same fleet that flattens energy prices also floods the fast ancillary products it once dominated.
In the first nine months of 2026, the average peak-to-trough delta for a two-hour storage project in ERCOT was just $34,632 per MW of installed capacity. That is the system average, and it hides the dispersion that matters. At specific locations, 101 price nodes were on pace for annual revenues of at least $62,064 per installed MW — roughly in line with the system-wide values of 2021 — with 20 nodes potentially exceeding $73,781 per MW. Looking ahead to 2030–2034, a proprietary nodal forecast projects eight ERCOT nodes with potential annual revenue above $78,073 per MW for a two-hour project in an equilibrium scenario. The golden era of 2022–2023 may be gone, but the opportunity has not vanished; it has moved from the system level to the nodal level.
What this means for the markets that haven’t saturated
For PJM and MISO, the lesson of October 13 runs in both directions. In the near term, the economics of storage in those markets look unusually attractive: the scarcity hours are real, they recur every maintenance season, and the storage fleet is still small enough that new projects would be discharging into genuinely high prices rather than into a market they themselves have flattened. The October 13 event is, in that sense, an advertisement for building storage in the East.
But the California and Texas experience is a warning about what comes second. A developer underwriting a project on today’s scarcity spreads is implicitly betting that the spreads persist — and the western evidence says they compress as the fleet grows, and compress faster than many early models assumed. The projects that survive saturation are the ones sited where congestion, not system-wide scarcity, creates durable price separation: nodes where transmission constraints isolate local prices from the system average. That is a much more granular underwriting exercise than buying into a market-wide trend, and it rewards operators who understand nodal pricing dynamics rather than those who simply chase the cheapest land or the largest queue position.
For market designers, the saturation story raises a question that goes back to the missing-money problem discussed earlier in this series. If storage flattens real-time prices, the energy market alone will pay less for the flexibility the system needs — even as the system’s need for flexibility grows with renewable penetration. The revenue has to come from somewhere: capacity accreditation that properly values storage’s contribution during risk hours, ancillary products that pay for ramping and fast response, or scarcity pricing mechanisms that preserve a signal even when the fleet is large. None of these is settled, and the eastern markets will have to answer them while their storage fleets are still being built, not after.
What to watch
Three indicators will tell you where each market sits on this curve. First, peak battery discharge relative to system size — the figure to track in PJM and MISO as their fleets grow toward the levels CAISO and ERCOT have already passed. Second, the frequency and magnitude of extreme real-time prices: the count of days above $500 or $1,000/MWh is a cleaner signal than any annual average, because averages hide exactly the hours that matter. Third, the nodal dispersion of storage revenues — whether the money is concentrating at congested nodes even as system averages fall, which is the pattern ERCOT now shows.
October gave us both halves of the story in a single month: a heat wave that failed to move prices in California, and a mild Tuesday that spiked them past $1,500/MWh in PJM. The difference between those two days is, increasingly, a measurable quantity of batteries. The markets that have built them are learning what saturation costs. The markets that haven’t are about to learn what it pays.