September’s numbers are in. My system made 2,457.5 kWh, about 8% above Tesla’s estimate for the month and 5% below last September. By my bill-with-vs.-without-solar estimate it saved me about $353 on FPL, and the solar is now 33.0% of the way to paying for itself.
What the panels made
The best day was September 26 at 110 kWh and the worst was September 4 at 38 kWh. The average was 82 kWh a day.
Daily production, September 2026
Source: my Tesla app daily data (exported). Dashed line: monthly average.
Solar vs. Tesla’s estimate and last September
Tesla estimated 6,803 kWh for August–October. I split that evenly, about 2,268 kWh a month; Tesla didn’t give a September number.
Why less than last September? I don’t fully know. Sunlight on the ground was almost the same: about 152.7 kWh/m² this September vs. 153.2 last year, from a weather service. It was a bit cloudier and wetter (average cloud cover 63% vs. 60%, 5.6 vs. 4.2 inches of rain) and hotter (average high 85.6°F vs. 82.0°F), and hot panels make a little less. My weather model for the roof expected only about 1.4% less, though, and the panels came in 5.3% lower. I haven’t pinned down the rest.
Where the energy went
| Sep 2025 | Sep 2026 | |
|---|---|---|
| Solar made | 2,593.8 | 2,457.5 |
| Home use, including the car | 3,742.6 | 3,908.9 |
| Bought from FPL (grid import) | 1,742.1 | 2,440.8 |
| Sent to FPL (grid export) | 526.2 | 971.4 |
| Net from FPL (import − export) | 1,215.9 | 1,469.4 |
| Powerwalls charged / discharged | 697 / 630 | 18 / 0 |
The biggest change from last year is the Powerwalls. This September they sat at a 100% backup reserve, so they stayed full and barely moved (18 kWh in, nothing out, all month). Last September they charged and discharged every day. With the batteries idle, the solar I didn’t use during the day went to FPL instead of the batteries, and at night the house drew from FPL instead. That’s why both imports and exports are much higher this year. On FPL’s 1:1 net metering, that shuffle costs nothing; what counts is the net, which rose by 254 kWh because home use (including the car) went up by 166 kWh and the panels made 136 kWh less.
The app’s Energy Flow panel shows where the solar itself went: Home 1,442.9 kWh (58%), Vehicle 74.9 kWh (3%), Powerwall 10.2 kWh (0%) and Grid 929.5 kWh (39%). The “Vehicle” line is only the solar that went straight into the car; most of its charging is overnight from FPL on its own meter. The app’s grid figure is a little below the 971.4 kWh total export in my data download; I haven’t worked out the 42 kWh difference.
What it saved
My savings estimate compares my FPL bill without solar to my bill with it, both at October 2026 rates (how FPL prices a kWh). The car is left out because it’s on its own EV meter and plan.
| September (house only, car excluded) | Sep 2025 | Sep 2026 |
|---|---|---|
| House use with no solar (kWh) | 2,900 | 2,900 |
| Net from FPL after solar (kWh) | 300 | 400 |
| Banked kWh used from earlier months | 0 | 0 |
| Bill without solar (with ~2.5% tax) | $414.10 | $414.10 |
| Bill with solar (with ~2.5% tax) | $48.48 | $61.04 |
| Saved (estimate) | $365.63 | $353.06 |
Both years are valued at today’s rates so they compare fairly. These house figures come from the Tesla app’s monthly breakdown, rounded to two significant figures, so they won’t match the exact totals above.
Savings ticker
Savings ticker · snapshot as of September 30, 2026 (estimate)
32.96% paid back as of September 30, 2026
This is a frozen snapshot for this update. The current savings ticker is here.
* Includes the 30% federal tax credit I received for my 2023 install. That credit isn’t available for systems installed after 2025 — see break-even without it.
Related: production vs. Tesla’s estimate, what my Powerwalls actually do and all monthly updates.