California solar under NEM 3.0: all we have to decide is what to do with the utility that is given to us

Every figure below comes from the same data that runs redo.energy calculators: our per-utility tariff database, the solar calculator's own energy-balance and export-credit code, and the Tesla Powerwall 3's installed price from our battery database. One house, one array, run through six California utilities.

Net Energy Metering 3.0 (NEM 3.0, officially the Net Billing Tariff) arrived in April 2023 and the shorthand has been the same ever since: California killed rooftop solar. The rule is real and it did change the math. But it also only applies to three utilities.

California has more than forty electric utilities. NEM 3.0 binds the three the state regulates: Pacific Gas and Electric (PG&E), Southern California Edison (SCE), and San Diego Gas and Electric (SDG&E). The city-owned utility in Los Angeles is exempt. Sacramento wrote its own rule. Santa Clara sells power for eighteen and a half cents. If you've read that solar no longer pays in California, it isn't even true of those three. A Central Valley house pays its panels off at each of them in under nine years. At some of the other forty, it pays them off faster.

So the first question isn't whether to buy solar. It's who sends you a bill.

What NEM 3.0 actually changed

Under the old rule, a kilowatt-hour you exported to the grid was worth the same as a kilowatt-hour you bought back. Your meter ran backwards and the rate didn't matter much.

Under NEM 3.0 the export is paid at avoided cost, which in our tariff data lands between four and six cents. Buying the same kilowatt-hour back costs 39 to 49 cents. That gap is the whole story, and everything else follows from it.

Solar used to be a generation product. Now it's a self-consumption product. The value of your array is no longer how much it makes. It's how much of what it makes you use yourself, in the hour it's made.

Same house, six utilities

Here's a 7 kilowatt (kW) array on a house in the Central Valley using 12,000 kilowatt-hours (kWh) a year, which is a normal number when the air conditioning runs all summer. The system costs $18,908 installed in our cost model, and there's no federal credit anymore. More on that below.

Read the first two rows together. The Los Angeles Department of Water and Power (LADWP) charges 31.6 cents, far under SDG&E's 49.0. Its customers get the best solar payback of the six anyway, by two and a half years.

LADWP is municipal, so NEM 3.0 doesn't apply to it. Full retail net metering is still in place, credits never expire, and there's no time-of-use rate to navigate. Every kilowatt-hour that array exports is worth 31.6 cents instead of four to six. Being exempt from the rule is worth more than seventeen cents of retail rate.

Silicon Valley Power, the municipal utility in Santa Clara, shows the same thing at a lower price. Its power costs 18.5 cents, and it still runs the original net metering, because it hasn't reached the 5 percent cap that would let it close the program. Every exported kilowatt-hour is worth the full 18.5 cents, so its payback lands level with PG&E's at less than half the price.

At the bottom, the Sacramento Municipal Utility District (SMUD) isn't a victim of NEM 3.0 either. SMUD has run its own Solar and Storage Rate since March 2022, which since June 2026 pays a flat 9.6 cents at any hour for exports, up from 7.4. Its problem is simpler: when power costs fifteen cents, each kilowatt-hour you avoid buying saves you fifteen cents. Cheap electricity is a good problem to have, and it makes solar slow.

"I live in LA" does not answer this question

LADWP serves the city of Los Angeles. It does not serve the region, and the difference matters more than almost anything else here.

Step outside the city limits and you are usually on SCE, fully under NEM 3.0. Santa Monica is SCE. Long Beach is SCE. So is most of the South Bay and most of the San Gabriel Valley. Inside Los Angeles County there are also municipal islands running their own rules and their own prices: Burbank at 27.8 cents, Glendale at 41.4, Pasadena at 37.6, Vernon at 14.6.

Glendale is the one to sit with. It charges 41.4 cents, less than two cents above SCE's 39.8, and plays by entirely different rules, because it is municipal and SCE is not. Same metro, same weather, same roof, nearly the same price per kilowatt-hour, and a different answer on whether a battery earns anything.

So the question is never which city you live in. It is which utility sends the bill, and that can change across a street.

Why a battery works in one city and not another

The usual advice under NEM 3.0 is that you need a battery. In regulated territory that's mostly right. Statewide it's wrong in both directions, and the reason is simple enough to hold in your head.

A battery doesn't make electricity. It moves electricity through time.

That's the whole thing. Your panels make power at noon. Your house wants power at 7 PM. A battery carries some of the noon power to the evening. So a battery earns exactly one thing: the difference between what a kilowatt-hour is worth in the evening and what it's worth at noon.

At noon, a kilowatt-hour you don't use gets exported, and you're paid the export price. In the evening, a kilowatt-hour you don't have stored gets bought, and you pay the retail price for that hour. The gap between those two numbers is the battery's entire paycheck.

So judge the battery on its own, separately from the panels. Same house, same 7 kW array, one Powerwall 3 at $12,500 installed:

Here's the part worth noticing: in every single row, the battery moves the same 3,991 kilowatt-hours a year. Same hardware, same sun, same house, same amount of energy carried from noon to evening. Nothing about the battery changes between these rows. The only thing that changes is what that energy is worth when it arrives, and that alone swings the answer from six years to never.

Notice too that the "earned per kWh" column lands below the gap in every row, for two reasons. The first is round-trip loss: put ten kilowatt-hours into a battery and about nine come back out. The rest is heat. The second is that a house only uses so much power in the evening peak. Here that's about 2,500 kilowatt-hours a year at the three regulated utilities, so the other 1,500 the battery delivers cover the late evening at the off-peak price. The gap is the most a stored kilowatt-hour can earn. Round-trip loss is a small tax when the gap is big, and it's the whole story when the gap is zero.

San Diego: the gap is enormous, so the battery is easy

SDG&E's evening power costs 64.7 cents. Its export credit is 4.4 cents. Every kilowatt-hour the battery carries into the evening peak is worth about 60 cents. About 2,500 of them land there each year, and the other 1,500 still earn about 40 cents, because even SDG&E's off-peak power costs 45 cents. Together they earn $2,093 a year, and $12,500 of hardware pays for itself in 6.0 years.

The irony is that San Diego is the one territory the state decided needed no help. When the California Public Utilities Commission wrote NEM 3.0 it gave PG&E and SCE customers a temporary bonus on top of the export price, the ACC Plus adder, and gave SDG&E customers nothing, on the reasoning that San Diego rates were already high enough to reach a payback under nine years unaided. That reasoning was right. It also means SDG&E has the most expensive evenings in California, and expensive evenings are exactly what a battery is paid to avoid.

PG&E: a narrower gap, and a slower payback

PG&E's evenings cost 43.9 cents, not 65. Across the year its time-of-use rate doesn't charge much more in the evening: our normalized shape puts its off-peak at 0.97 of average and its peak at 1.12. Twelve percent above average is not an expensive evening.

So the same battery, moving the same 4,000 kilowatt-hours, earns $1,431 instead of $2,093, and takes 8.7 years to pay for itself. The hardware is warranted for ten years. You would spend most of that decade paying the battery off and finish about a year before the warranty ends. That isn't a disaster. It's a thin margin, so if a quote shows a faster payback, ask what evening price it assumes.

Southern California Edison sits in between. Its evenings cost 53.3 cents, 34 percent above its average, so the same battery earns $1,611 a year and pays for itself in 7.8 years.

Los Angeles: the gap is zero, so the battery loses money

LADWP is exempt from NEM 3.0, so it still runs full retail net metering. A kilowatt-hour you export earns 31.6 cents. A kilowatt-hour you buy back costs 31.6 cents. The gap is zero.

With full retail net metering, the grid is already your battery. It accepts everything you send, gives it back at the same price, holds it indefinitely, charges nothing for the service, and loses nothing along the way. It is a perfect battery, and you already own it.

A $12,500 Powerwall does the same job slightly worse. It moves the same 4,000 kilowatt-hours, earns nothing for moving them because both ends are priced the same, and loses about ten percent of them to round-trip inefficiency. So it earns negative 3.3 cents per kilowatt-hour moved, which comes to about $130 a year of pure loss. Payback goes from 5.0 years to 8.7, and the annual saving falls from $3,756 to $3,624.

There are good reasons to buy a battery in Los Angeles. Keeping the lights on in an outage is a real one, and if that's why you want it, buy it. Just don't let anyone tell you it pays for itself, because in LADWP territory it actively doesn't.

Sacramento: a small gap, and a smaller rebate

SMUD pays a flat 9.6 cents for exports at any hour. Its 5 to 8 PM peak is cheap for most of the year: 17.8 cents from October through May, and 37.7 cents only in summer. Our model counts both, which puts SMUD's evening at about 24 cents across the year, so the gap is 14.8 cents. But the peak runs only three hours on weekdays, so about 1,100 of the 4,000 kilowatt-hours the battery moves land in it. The battery takes 48.9 years on its own. Every summer evening is counted at full price, and it's still a no.

SMUD also pays a battery rebate through its My Energy Optimizer Partner+ program: $300 per kilowatt-hour, capped at $6,000, since September 23, 2026. It was $500 and $10,000 before that, and a project submitted for interconnection by September 22 and enrolled by December 31, 2026 keeps those amounts. On a 13.5 kilowatt-hour Powerwall that's $4,050, which cuts the cost to $8,450 and the payback to about 33 years. The same program pays a Tesla battery that answers SMUD's events $110 a quarter, or $440 a year. Count that and payback falls to about 12 years, but it's paid only for events the battery joins, so treat it as the optimistic figure rather than the plan.

SMUD isn't alone, and it's no longer the biggest. Burbank pays $530 per kilowatt-hour to customers who can't claim a federal tax credit. Since the homeowner credit ended with 2025, that's every homeowner. On a Powerwall that's $7,155, which brings the battery's own payback from 14 years to about six. Customers who can claim a credit get $275. Pasadena pays $350 per kilowatt-hour ($550 for income-qualified households), Silicon Valley Power 15 cents per watt-hour of capacity, about $2,000 on a Powerwall, and Anaheim up to $1,500. PG&E pays $300 toward a portable generator or portable battery in high fire-risk areas. The Self-Generation Incentive Program (SGIP), which used to do this statewide, is waitlisted.

The rule you can carry away

Find two numbers: what your utility pays you for an exported kilowatt-hour, and what it charges you for one at 7 PM. Subtract. If the answer is over 45 cents, a battery pays. If it's around 40, it pays back just inside its ten-year warranty. If it's 30 or less, it won't pay back inside the warranty unless a rebate carries it. If it's near zero, you already have a battery and it's called the grid.

Your climate zone changes it as much as your utility

The same array on the same rate, moved from the Central Valley to the coast:

Inland houses run air conditioning, and air conditioning is the best load solar has ever had: it peaks in the afternoon, in summer, which is exactly when the panels are working. Forty-one percent of production gets used on site without any equipment to help it.

On the coast the panels do fine and the house barely uses anything during the day. Self-consumption falls to 24 percent, three quarters of production goes to the grid at five cents, and payback stretches by about four and a half years.

Which produces the sentence that surprises people: under NEM 3.0, the sunniest, mildest roofs aren't automatically the best investments. Santa Monica has better weather than Fresno and worse solar economics.

Notice the last row. LADWP moves by six months between the two climates, because when exports are worth full retail it doesn't much matter when you use the power. The exemption doesn't only raise the return. It makes the return robust to how you live.

The electric vehicle answer is about timing, not ownership

An electric vehicle (EV) is the biggest flexible load most households will ever plug in, so it should be the perfect partner for solar. Whether it is depends entirely on when it charges.

Our model runs EV charging on the standard overnight schedule, 11 PM to 6 AM, the window most utilities push you toward and most people use. On PG&E inland, that takes self-consumption from 41 percent down to 34, and payback from 8.6 years to 9.8.

Charging overnight makes your solar economics worse. The car adds load in the hours the sun isn't up, which changes nothing about how much production you eat, while adding to the bill the panels can't offset.

The fix isn't to skip the EV. It's to charge it in the afternoon when you can, on a weekend, working from home, at a workplace charger. The same applies to anything else you can move: the pool pump, the electric water heater, the dishwasher, the laundry. Those cost nothing and they're the cheapest self-consumption you'll ever buy.

A heat pump helps too, and less than you'd hope. California heating loads are modest and winter is when production is lowest, so a heat pump raises self-consumption in the months the array has least to give. In the northern inland part of the state it's worth more than on the coast.

The federal credit is gone, and it isn't coming back this year

The residential clean energy credit (25D) terminated at the end of 2025. A California homeowner buying solar in 2026 gets no federal credit, which is why the costs in this piece have no subsidy in them. A lot of the internet still says 30 percent. It's out of date.

State money doesn't fill the gap. The California Solar Initiative finished years ago, LADWP's rebate ended in 2018, and the Self-Generation Incentive Program (SGIP) for batteries is waitlisted. The few utility battery rebates left, listed above, aren't included in the tables. The property tax exemption for solar still applies.

One consequence worth understanding: leases and power purchase agreements became relatively more attractive, because the business-side credit didn't end on the same schedule and a third-party owner can still claim it. You give up ownership of the asset to get the benefit of a credit you can no longer claim yourself. Whether that trade is good depends on the contract, and those contracts are where the solar industry does its worst work.

If you're in regulated territory, sooner is worth something

The ACC Plus adder is set on the day your system gets permission to operate, and it's locked for nine years from that date. It's also declining about 20 percent a year. PG&E's is around 0.88 cents for a 2026 connection, down from 2.2 cents in 2023. SCE's is about 1.6 cents, down from 4.0.

That's a small number attached to a real question. Waiting a year doesn't cost you the difference for a year. It costs you the difference for nine.

Some honest caveats

These are simple paybacks: system cost divided by first-year savings, with no rate escalation, no panel degradation, and no financing. California rates have risen faster than inflation for a decade, which makes every payback here conservative, but treat these as comparisons between utilities rather than as a promise about your bill.

The battery model dispatches greedily into the evening and doesn't bid into demand response programs, so a battery enrolled in a utility program earns more than shown. Our load shapes are typical rather than yours. SMUD raised its export credit from 7.4 to 9.6 cents in June 2026 and reviews it every four years, so check the current figure before you plan around it. And if your utility isn't one of the six here, the first thing to find out is whether it still runs full retail net metering, because that one fact decides whether a battery earns anything.

The takeaway

Solar still pays in California. What changed is that it now pays through your meter rather than through the grid, so the answer depends on facts about you rather than facts about the state.

Find out who your utility is. If it's municipal, you may still be on full retail net metering, in which case buy the panels and skip the battery. If it's PG&E, SCE, or SDG&E, you're on net billing, and the array is worth what you can consume: a battery pays in San Diego, pays more slowly at SCE, and takes most of its warranty to pay back at PG&E. If you're inland with air conditioning you're most of the way there without buying anything. If you're on the coast, a battery lifts self-consumption from about 24 percent to 55 and cuts the whole system's payback by one to two years. Outside San Diego, though, it takes longer than its ten-year warranty to pay for itself.

Our solar calculator runs your own address against your own utility's export rules, time-of-use windows, and battery economics. Run it before you sign a quote.