The residential solar business changed its product this year. At pv magazine USA Week 2026, Tesla, FranklinWH, and Ohm Analytics described a market in which the panel on the roof is no longer the sale. The battery is. Residential solar installation volume fell 19 percent year over year, according to data presented by Ohm Analytics partner Ryan Feeney, while residential storage held steady at nearly 1 GWh installed in the second quarter of 2026. That divergence pushed the national battery attachment rate to 49 percent. California’s NEM 3.0 projects reached 80 percent. In Texas, storage installations rose 132 percent year over year, roughly three-quarters of it standalone retrofits.
The context is the end of the federal residential tax credit: Section 25D expired December 31, 2025, leaving a smaller, pricier, far more storage-dependent market. Feeney’s framing, as reported by pv magazine USA on November 3, was blunt: solar down 19 percent, storage “close to a gigawatt-hour” and “roughly flat.” One caveat: other trackers count the segment differently — SEIA and Benchmark Mineral Intelligence put residential storage at 657 MWh in the second quarter, down 27 percent year over year. The direction is contested; the attachment rate is not: roughly half of new residential solar now comes with a battery, four out of five in California.
The battery is now the product
Ronnie Pettersson, FranklinWH’s vice president of product and application engineering, described the shift as a change in the kitchen-table conversation: from “how much solar do you want to have on the roof” to “how many batteries do you really need.” The reasons were consumer-facing — outages, rising retail prices, grid stress from data-center load — but the economics are structural: as net-metering export rates decline, exporting surplus solar pays less and storing it for the evening pays more. Storage has moved from optional add-on to the central driver of project economics.
Texas shows the retrofit market doing the work: storage installations up 132 percent year over year, three-quarters of it standalone retrofits from homeowners who already own panels. Feeney’s projection was equally direct: five years out, storage is the standard model, and “successful contractors are going to be the ones that sell batteries, enroll them, and can explain it to homeowners.” Ohm Analytics estimates 30 to 35 percent of residential batteries are already enrolled in a virtual power plant program — the grid-services revenue is increasingly part of the financing, not an afterthought.
A fleet that already behaves like generation
Doug Middleton, Tesla’s utility and VPP business development lead, told the panel Tesla now manages more than 1.5 million Powerwalls globally, with 2.5 GW — nearly 5 GWh — enrolled in active VPP programs. In California, Tesla aggregated more than 600 MW across multi-hour summer dispatches, and the Powerwall VPP fleet grew 50 percent year over year. In Puerto Rico, the aggregated network acts as a continuous marginal generation resource, preventing rolling blackouts. Middleton’s line — “this is not demand response, right? It’s demand-sited generation” — is the claim the rest of this column tests against the market rules.
The performance data supports the claim: Connecticut Green Bank verification puts FranklinWH and Tesla participation above 90 percent during requested utility dispatches, against below 70 percent for competing manufacturers. That gap matters — a fleet that answers nine calls in ten is a resource; one that answers seven is a forecast error. It also matters for Middleton’s financing argument: if VPP revenues can be written into customer financing contracts, a homeowner’s monthly battery lease payment falls from roughly $200 to roughly $100, doubling the addressable market. Whether that halving survives contact with actual program payments is the next section’s question.
Paid like a load, not a generator
Here is the market-design problem, as Middleton stated it: legacy rules in CAISO and PJM still treat battery aggregations solely as load reduction, so home batteries generally cannot be paid for exporting power into wholesale markets. The rules were written for demand response — large loads agreeing to consume less — and a discharging battery is forced through mechanisms built for a load that stops consuming. “There’s a lot of policies that were set before battery storage was really a technology, and the way we participate is through mechanisms designed for large loads,” Middleton said. “The fact that a distributed battery can’t get paid for exports in wholesale markets is like such an obvious win.”
Texas illustrates the point from the other side: ERCOT has gone further than any other operator and still stops short. Under its Aggregated Distributed Energy Resource pilot, home batteries participate as Aggregate Load Resources — settling at the load-zone price rather than a nodal price, exports treated as negative load, and excluded from ancillary services. The pilot has grown fast — 8 MW of discharge capacity in June 2025 to 171 MW in September 2026, with 341.2 MW of total qualified capacity, after ERCOT raised the cap to 500 MW in March — but the settlement treatment is the tell: a resource dispatched like a generator and settled like a load is paid for half of what it does. ERCOT’s board votes December 7 on Phase 4, which would add a storage-only option dispatched by node with no ERCOT-wide cap; even that proposal still pays the load-zone price and excludes the batteries from ancillary services.
The federal fix exists on paper and arrives slowly. FERC Order 2222, issued in September 2020, requires every FERC-jurisdictional operator to let aggregated distributed resources bid into wholesale energy, capacity, and ancillary-service markets. Six years later, ISO New England went live November 1, 2026; CAISO finished its framework in November 2024 and NYISO launched its DER model in April 2024. PJM’s full model takes effect February 1, 2028. That is the timeline when the fix travels through RTO rulemaking — measured in years, with the largest U.S. wholesale market last in line.
The money that does exist
What exists instead is a patchwork of utility and state programs that pay handsomely for the same flexibility the wholesale markets will not buy. In Massachusetts, ConnectedSolutions offers up to $400 per kW-year for batteries in congested distribution circuits — a base incentive plus a $125 per kW-year distribution-service adder; Eversource’s ConnectedSolutions+ pilot pays that rate near the Alewife, Hyde Park, and Dewar substations in parts of Cambridge, Milton, and South Boston, against a standard $275 per kW. In Texas, utility VPP programs pay up to $325 a year per household — Entergy Texas’s Battery Storage Solutions program pays FranklinWH customers up to that amount. In New Jersey, PSE&G’s GridSmart Battery Program, launched August 26, 2026, offers a $5,000 incentive toward a qualifying home battery plus up to $25,000 in interest-free on-bill repayment, in exchange for VPP enrollment. In California, Ava Community Energy offers up to $6,000 in upfront battery rebates for qualifying low-income customers.
None of these payments is a wholesale market price. They are retail programs, administered by utilities or community choice aggregators and sized by administrators rather than discovered in an auction. A battery in a congested Massachusetts circuit earns $400 per kW-year because a utility decided the local wires are worth that much; the same battery on an uncongested circuit earns the $275 base rate; one in a state with no such program earns its backup value and bill savings, and little else. Compensation follows the program map, not the physics. And the programs are capped, piloted, and revocable in ways wholesale participation is not: GridSmart and ConnectedSolutions+ are pilots, and the ADER cap is one ERCOT has had to raise three times.
The workaround arrives through law and contract
The most consequential development in the panel’s account may be the one that bypasses the RTOs entirely. A recent New Jersey law — P.L. 2026, Chapter 32, approved July 7, 2026 — directs the state Board of Public Utilities to set standards within 12 months for electricity provision to large data center customers, including parameters for demand-reduction measures those customers may take directly or procure from third parties, such as demand response enrollment and behind-the-meter storage. As pv magazine reported it, the law lets commercial data centers contract directly with distributed energy resource aggregators. The statute is more careful — it orders the regulator to write the framework, and the contracting right will be defined there — but the direction is unmistakable.
The deal that shows what the framework enables predates the law. On June 2, 2026, Voltus announced a three-year Bring Your Own Capacity agreement with Google: Voltus aggregates up to 100 MW of distributed resources a year — batteries, smart thermostats, and other flexible assets from homes and businesses — into a Google-funded virtual power plant in PJM, with Voltus responsible for Google’s capacity obligation. On October 22, the companies expanded the deal by 1,070 MW, ramping over three to four years under a contract running through 2040; the first 100 MW phase was oversubscribed, Voltus said. The structure is the point: a large load buyer paying an aggregator directly for accredited distributed capacity, outside the capacity auction, because the auction’s rules for distributed resources are still being written. Voltus chief executive Dana Guernsey has compared these contracts to the power purchase agreements that financed utility-scale solar and wind — a structure other buyers can copy.
The irony should not be smoothed over. The Google–Voltus arrangement is, in substance, a capacity deal: Google pays for accredited capacity against its obligation, and Voltus delivers it by aggregating resources that reduce net load at the hours that matter. It is the load-reduction model, commercialized at scale by a private buyer rather than reformed by a market operator. It works, it is financed, and it grew tenfold in four months — but it does not resolve the design question. It routes around it, for buyers large enough to contract directly.
What this means for market design
The design question is straightforward. A home battery can do two things the system values: reduce the household’s draw from the grid, and inject stored energy back into it. The first is compensated, through demand response, capacity accreditation, and bilateral deals like Google’s. The second is, in the largest wholesale markets, not compensated at all, or compensated only as negative load at a zonal price, without access to the ancillary products the battery can technically provide. The fleet this column described on October 30 as flattening real-time prices in California and Texas is, in the East, still paid like a thermostat.
That asymmetry has a distributional consequence. Bilateral deals and utility programs reward the resources easiest to aggregate and the buyers large enough to contract: data centers with capacity obligations, utilities with VPP pilots, aggregators with portfolios in the hundreds of megawatts. A homeowner with one battery in a state without either program gets backup value and bill savings, and little else — though her battery discharges the same evening power as the one in the Massachusetts pilot. Wholesale access is the mechanism that would pay the resource rather than the program; Order 2222 was written on that premise. That its PJM implementation lands in February 2028, while the New Jersey legislature and a hyperscaler wrote a working alternative in 2026, is the clearest measure of how far rulemaking lags the resource it governs.
There is also a revenue question the storage-first model cannot dodge, raised in this column in October: batteries earn from the spread between cheap charging hours and expensive discharging hours, which a large fleet compresses. In ERCOT, the average peak-to-trough value for a two-hour project has fallen 25 to 30 percent below its 2021 starting point as the fleet grew past 13 GW of discharge capacity. The residential fleet faces the same arithmetic: every battery in a VPP makes the evening hours it discharges into slightly less valuable. Program payments like ConnectedSolutions’ $400 per kW-year are, in that light, a hedge against the saturation of the energy spread, paid by utilities that value local reliability more than the wholesale price ever will. The programs are the revenue bridge — the question is whether they lead to wholesale participation or substitute for it.
What to watch
Four things, in the order they will become visible. First, ERCOT’s December 7 vote on ADER Phase 4: whether the storage-only option advances, and whether load-zone settlement and the ancillary-services exclusion survive — if they do, Texas will have scaled a pilot that still pays batteries like loads. Second, ISO New England’s first weeks under Order 2222, live since November 1: whether aggregators register residential fleets in volume, and at what prices. Third, the New Jersey Board of Public Utilities’ standards, due within 12 months of the July law: whether third-party procurement parameters cover residential aggregations, or serve only the data centers that lobbied for them. Fourth, the attachment rate — whether 49 percent holds as the post-credit market finds its level, and whether the retrofit-driven Texas pattern spreads as export rates keep falling.
The panel’s verdict on the installer business is probably right: five years out, storage is the standard model. The open question is whether the market that buys what those batteries do catches up in the same five years — or whether the fastest-growing flexible resource in the power system spends them paid like a load, by programs and private buyers, while rules written for another era finish their slow revision.