Case study · California
California: a record battery VPP on an annual budget
Rolling outages in 2020, a capacity payment that built a gigawatt of home batteries in three years, record dispatches in 2026, and no money for 2027.
VPP Economics has no affiliation with, sponsorship from, or endorsement by any utility, aggregator, platform, or device maker named here.
The short answer
- The problem: California runs short on the hottest evenings. Rotating outages hit in August 2020, and the state set aside a $3.4 billion Strategic Reliability Reserve in 2022 to avoid a repeat (California Energy Commission, CEERT).
- The solution: pay home batteries for capacity they prove. The Demand Side Grid Support (DSGS) program pays aggregators per kW of battery discharge demonstrated on hot evenings. Option 3 grew from 10 MW enrolled in 2023 to 768 MW in 2025 (CEC).
- The outcome: record dispatches. More than 140,000 batteries delivered 580 MW on September 9, 2026 (Sunrun), and 745 MW on October 6 (Electrek).
- The verdict: cost-effective, but unfunded. A Brattle study commissioned by Sunrun and Tesla puts net savings at $28–206 million over 2025–2028 (Brattle). The 2026 state budget gave DSGS no money for 2027 (Canary Media).
The problem: short on the hottest evenings
On August 14 and 15, 2020, the California Independent System Operator (CAISO) ordered rotating outages during a West-wide heat wave. In 2021, the Bootleg Fire cut about 4,000 MW of imports (CEC Joint Agency Reliability Assessment). The risk falls in the early evening: solar output fades while air conditioning load is still high.
In June 2022, Assembly Bill 205 created a Strategic Reliability Reserve to cover up to 5,000 MW in emergencies (CEC). During the heat wave of August 31 to September 9, 2022, the reserve provided between 554 and 1,416 MW (CEERT comments).
What the shortfall costs
The reserve cost about $3.4 billion: $2.37 billion for the Department of Water Resources’ Electricity Supply Strategic Reliability Reserve, mostly for gas and temporary generation, $700 million for the Distributed Electricity Backup Assets (DEBA) program and $295 million for DSGS (CEERT, CEC).
Capacity itself got expensive. Utilities and community choice aggregators must buy resource adequacy (RA) contracts for every kW of expected peak. The weighted average price of system RA nearly doubled, from $14.37 to $26.26 per kW-month between 2023 and 2024 (CPUC Energy Division). The DEBA recommendation for new grid-scale storage was about $410 per kW, one time (CEC).
The solution: pay for capacity that batteries prove
DSGS launched in August 2022. Its first season paid 44 participants $7.88 million for 3,100 MWh, mostly from backup generators (CEC). In 2023, the CEC added Option 3, an incentive for aggregated batteries (CEC guidelines). Three design choices made it work:
Per kW, not per event
Aggregators earn a monthly capacity rate for each kW their fleet proves in tests and events. In 2025, the season totaled $62.10 per kW for a 2-hour commitment, $74.52 for 3 hours and $82.80 for 4 hours, plus a 30% bonus (EnergyToolBase).
Price, not only emergencies
Events trigger when the CAISO day-ahead price exceeds $200 per MWh, or the grid declares an emergency. Brattle expects about 35 events a year (Brattle).
One contract, many brands
The CEC pays the aggregator, which sets the customer’s share (Olivine DSGS FAQ). Sunrun and Tesla brought fleets that were already installed and connected.
Discharge beyond typical behavior
Performance is the discharge above a baseline of what the battery would normally do, so owners are paid for extra grid support, not for routine self-consumption (Tesla).
What happens during an event
- A price signal. The CAISO day-ahead price for an evening hour tops $200 per MWh, or the grid declares an Energy Emergency Alert (Brattle).
- The aggregator schedules the event. Tesla’s events last no more than two hours between 4 and 9 p.m., at least once a month and at most 35 times a season (Tesla).
- Batteries charge, then discharge. Batteries charge ahead of the event, then discharge down to the owner’s backup reserve. Energy powers the home first and is exported after that.
- Owners can opt out. Skipping an event lowers the average performance and the pay (Tesla).
- Settlement after the season. The CEC pays aggregators, and aggregators pay owners. Tesla pays $64.55 per kW in 2026, about $284 for one Powerwall in its example (Tesla). Sunrun pays a fixed reward, such as $150 per battery in CalReady+ (Sunrun).
Demand response or VPP?
DSGS Option 3 sits between the two. The fleet is a virtual power plant (VPP) in its hardware: software-dispatched batteries that export to the grid and are paid on measured capacity. Its dispatch is event-based: a few dozen evenings a year, not daily. Unlike Puerto Rico’s emergency program, it is triggered by market prices; see our LUMA case study.
VPP-grade
Two-way aggregator control, battery telemetry and grid export from more than 140,000 homes.
Price-triggered capacity
Up to about 35 events a year when wholesale prices spike, paid per kW of demonstrated capacity.
Orchestrated
No daily dispatch and no direct bidding into CAISO markets. On our spectrum, it is an event-based VPP with a capacity payment.
Money trail for one kW in DSGS Option 3
From the state budget to battery owners, per kW of demonstrated capacity, per Powerwall and by season.
Swipe to see the full chart →
The outcomes
Option 3 enrollment
Megawatts of battery capacity enrolled each season.
Swipe to see the full chart →
For program designers, the lesson is that a capacity payment, not the number of events, built the fleet. In 2024, Option 3 paid $8,342,880 to 37,650 participants and demonstrated 117.8 MW in September (CEC). Participation more than tripled the next year. Yet the 2025 season had no price-triggered or emergency events at all, only monthly tests. Owners were paid to stand ready.
Dispatch records
Megawatts. Company peaks are not the same as averages over an event.
Swipe to see the full chart →
On September 9, 2026, Tesla Powerwalls supplied 517 MW from about 110,000 units, 55% of them owned or operated by Sunrun. More than 30,000 other batteries added 63 MW (Sunrun). The next evening, Southern California Edison requested a dispatch and got more than 140 MW. The companies said the fleets could have delivered more than 720 MW (ESS News). The 580 MW is company-reported peak output, with no hourly data published (Nature World News).
Then October 6 set another record: 745 MW, with more than 600 MW from over 85,000 Powerwalls, about 7 kW per home (Electrek). October 5 was the first time DSGS and the Emergency Load Reduction Program (ELRP) were called together across every participating utility territory.
The economic verdict
Brattle’s cost-benefit estimate
Millions of dollars, 2025 to 2028. Shaded bars are ranges.
Swipe to see the full chart →
On Brattle’s numbers, DSGS is a success. It projects enrolled capacity rising from about 700 MW to 1.3 GW by 2028, at a cost of about $196 million against $223–402 million in benefits. About 90% of the cost is payments to participants (Brattle). Brattle’s test result, 539 MW over two hours, was about 1.9% of CAISO’s net peak (ESS News). Because the study was paid for by the two largest aggregators, the range deserves an independent check.
Cost per kilowatt
Dollars per kW. Not like-for-like: DSGS and RA are per season or year; DEBA is a one-time incentive.
Swipe to see the full chart →
The cost comparison points the same way. DSGS paid about $71 per demonstrated kW in 2024. Its 2-hour rate is $81 per kW a season. System RA cost about $315 per kW a year at 2024 benchmark prices, although a 2-hour battery does not count as a full RA resource and the comparison is indicative only.
The funding problem
Millions of dollars.
Swipe to see the full chart →
The weakness is the funding source. DSGS is paid from the state budget, not utility rates, so it competes with every other line item each year. Funding was cut in 2024 and 2025 (Canary Media). For 2026, Option 3 got $19.5 million, raised to about $42.7 million by the legislature, and new entrants were closed except EV chargers (CalSSA). The governor proposed moving DSGS customers into the utilities’ ELRP (Utility Dive). The final 2026 budget left nothing for 2027, after a $70 million proposal failed (Canary Media).
Likely net savings
Lower RA and peak energy costs, by Brattle’s estimate, without paying through bills.
A modest, uncertain payment
About $284 a season for one Powerwall at Tesla’s 2026 rate, if the budget covers it.
A gigawatt with no contract
Fleets built in three years face a gap in 2027 with no long-term commitment.
Reliability on an annual budget
A cheap reserve that must be refunded every year.
Lessons for program designers
Capacity pay builds fleets
Paying per kW demonstrated, even in a year without events, grew enrollment from 10 MW to 768 MW in two seasons.
A price trigger beats waiting for emergencies
The $200 per MWh trigger dispatches batteries when energy is expensive, not only when the grid is failing.
Budget money is not durable
A proven, cost-effective program paid from the general fund lost its 2027 money. Rate-funded or market-based pay lasts longer.
Use the fleets already installed
Sunrun and Tesla brought tens of thousands of connected batteries; no new hardware was needed.
Baselines matter
Paying only for discharge above typical behavior keeps the state from buying what owners already do.
Publish the hourly data
Record dispatches are company-reported peaks. Public interval data would make the case stronger.
Where else the model could work
Price spikes without a capacity market
ERCOT has no capacity market, so a state-funded capacity pay would be new. See the ERCOT money trail.
Already paid per kW
ConnectedSolutions pays a seasonal rate per kW from energy efficiency budgets; see our case study.
Evening peaks, growing fleets
Arizona, Nevada and Colorado share the solar-driven evening peak. A capacity payment through utility rates would avoid the budget risk.
Related: how PG&E’s $2 per kWh ELRP pays batteries in the per-kWh money trail.
Method and caveats
An independent case study for program designers, regulators, aggregators, households and other readers. Data checked October 27, 2026.
Enrollment, payments and demonstrated MW come from the CEC. Dispatch records in 2026 are company-reported peaks. The cost-benefit estimate is Brattle’s, in a study commissioned by Sunrun and Tesla. Cost-per-kW comparisons mix seasonal, annual and one-time costs and are indicative only. Household figures are gross payments, not net profit.
VPP Economics has no affiliation with, sponsorship from, or endorsement by the California Energy Commission, CAISO, Olivine, Tesla, Sunrun, Brattle or any utility named here.