NEM 3.0 didn't kill solar in California. It killed solar without batteries.

For the second half of 2023, every California solar quote I saw assumed the customer would export half their generation back to the grid at retail rates. That assumption was already wrong by the time it was being typed. The CPUC's NBT decision had taken effect in April. The retail-rate export days were over.

Two years later, we have data. The Net Billing Tariff replaced retail-rate net metering with hourly export credits derived from the Avoided Cost Calculator — a methodology that aims to pay you what your kilowatt-hour is actually worth to the grid at the moment you produce it. Sunny April afternoon? About four cents. Hot September evening? Seventy.

The decay schedule is the headline.

Export credits aren't fixed. They follow a decay schedule — each new vintage of customers gets a lower set of avoided-cost values than the previous one, because the marginal value of midday solar to the grid keeps falling as more solar comes online. A system interconnected in 2023 locks in 2023 values for nine years. A system interconnected in 2027 locks in 2027 values, which are a fraction of the 2023 set.

For a panel-only system in PG&E territory, the difference between a 2023 vintage and a 2027 vintage is about a six-year extension to payback. For systems with batteries, the difference is closer to two.

"The policy didn't penalize solar. It penalized exporting electrons at the moment of lowest grid value, which is exactly what an unbatteried system does."

What the math says.

We ran the numbers on 84 California quotes interconnected after April 2023. Median first-year export value, panel-only: $412. Median first-year self-consumption value (solar offsetting your own load): $1,840. The grid is paying about a quarter of what your meter is offsetting. Without storage, the system can only deliver value during your daytime load hours.

A battery moves that calculus. Charge during midday surplus, discharge during the 4–9pm peak when import rates exceed 50¢/kWh, and the battery captures 80% of the value of what NEM 2.0 used to deliver for free. The capital cost is real — $11,000–$14,000 for a typical Powerwall-class install — but the IRR on that incremental capital is in the 9–12% range under our 2025 vintage assumptions. That's a separate investment decision from the panels, and it almost always wins on a post-NBT roof.

Headlines vs. structure.

The "NEM 3.0 killed solar" framing makes a good chart but doesn't survive contact with the numbers. Total California residential interconnections fell about 60% in the first year of NBT. They're climbing back, with a much higher attached-storage rate — the latest CALSSA data has 78% of new residential applications including storage, up from 14% pre-NBT. The market didn't disappear. It re-shaped, and the re-shaping is what the policy was designed to do.

The honest read: if you're in California, you should still be modeling solar — but you should be modeling solar-plus-storage, you should be using current-vintage avoided cost values rather than historical net-metering credits, and you should be very skeptical of any installer quote that uses a single round-trip annual export value as a placeholder. The decay schedule means the next homeowner to interconnect will see worse numbers than this one. Stale assumptions get more stale.