Behind-the-meter optimization
30 GW of installed solar+storage runs on commissioning-day schedules
Behind-the-meter solar and storage systems degrade, shift production curves, and encounter shading changes that commissioning-day models never anticipated. Most systems still run the dispatch schedules programmed on day one. The gap between actual production and optimized production widens every month as conditions drift from assumptions.
The asset was optimized once. The conditions change daily. The schedule does not.
Behind the meter is behind the curve
US behind-the-meter solar+storage capacity reached 30 GW in 2024, yet most residential and C&I systems run on static schedules designed at commissioning. Time-of-use rate changes, demand charge windows, and utility program availability shift quarterly — the installed system does not. The gap between installed capability and captured value widens every year these assets run on fixed rules.
The hardware is deployed. The intelligence is not.
How AI optimizes behind-the-meter solar+storage
Monitor production vs expected output
Compare actual inverter output against weather-adjusted theoretical production. Degradation, soiling, shading, and equipment issues each leave distinct signatures in the production gap.
Adaptive dispatch scheduling
Recompute charge/discharge schedules daily based on updated production forecasts, rate structures, and demand charge windows. What worked in January does not work in July.
Detect and diagnose underperformance
Separate normal degradation from actionable issues. String-level analytics identify which panels need cleaning, which inverters are clipping, and which batteries are aging faster than expected.
Optimize economic life decisions
Model remaining useful life against replacement economics. Know when to repower, when to add capacity, and when to ride the degradation curve.
Portfolio-level monitoring vs panel-level AI management
| Metric | Manual Process | AI-Optimized |
|---|---|---|
| Forecasting accuracy (MAPE) | 8-10% | 3.21% |
| Decision cycle time | 4-8 hours | 15 minutes |
| Billing query resolution | 2-3 days | < 5 minutes |
| Residual value model refresh | Quarterly | Daily |
| Operational data utilization | < 30% | 98%+ |
| Margin capture potential | Baseline | 5-12% uplift |
BTM value capture hierarchy
Installers that pair hardware with ongoing AI optimization capture recurring revenue (8-12% of savings share) beyond the installation margin. Those that install-and-forget cede the optimization layer to third-party platforms like Stem and Enphase. The value migrates from the panel to the control layer.
The installer that walks away after commissioning is handing margin to the platform that stays.
Key players
Sunrun
Largest US residential solar+storage installer; 900K+ customers, remote fleet optimization.
Enphase Energy
Microinverter leader with IQ battery; cloud-based monitoring for 3M+ systems.
Tesla (Powerwall)
Residential battery leader; Virtual Power Plant enrollment in CA, TX.
Stem Inc
AI-driven commercial storage optimization (Athena platform); 6 GWh under management.
What we have shipped in this space
Residuals — operational telemetry to financial instruments
Battery degradation curves, solar performance decay, and generation asset condition converted from operational telemetry into residual instruments that reflect actual state.
Our residual value system converts operational telemetry from solar and storage assets into financial instruments that reflect actual asset condition rather than book depreciation schedules.
Residual pricing from real degradation data, not accounting assumptions.
Ready to instrument your operations?
Get a solar asset valuation and degradation analysis. We'll show you the gap between assumed and actual performance, and the exact revenue optimization available.
Schedule an auditExplore more
Related activities
Microgrid operations→
Microgrids operate in island mode where generation must match load in real time without utility back...
Renewable generation→
Renewable generation has zero marginal cost but uncertain output. When forecasts overpredict, curtai...
Congestion revenue rights→
Congestion revenue rights in ISO markets are a $7B annual profit pool where returns accrue to partic...
Common questions about AI in solar asset management
What is the annual degradation rate for silicon PV panels in high-dust environments?
Silicon panels in high-dust regions (Middle East, India, Southwest U.S.) show 0.8–1.5% annual degradation; standard environments show 0.5–0.8%. Cumulative degradation over 25 years can reach 12–25% in high-dust areas versus 8–15% in clean environments.
How much performance penalty does panel soiling impose on utility-scale solar farms?
Unmanaged soiling reduces annual output by 2–6% in moderate climates and 8–12% in arid/dusty regions. Monthly cleaning cycles recover 4–8% lost output; weekly cleaning approaches diminishing returns and exceeds economic threshold except in extreme soiling environments.
What is the optimal cleaning frequency given water costs and degradation rates?
Economic optimization of cleaning typically yields 6–12 week intervals in moderate environments and 2–4 week intervals in high-dust regions. At $0.50–$1.50 per megawatt-hour water cost, cleaning economics decline significantly in water-stressed regions.
How much can predictive maintenance reduce unexpected downtime on a 100-megawatt solar asset?
Predictive maintenance programs combining thermal imaging and electrical monitoring reduce unplanned downtime by 40–60% compared to reactive maintenance. Such programs require 2–3% capex investment but typically deliver 15–25% improved asset utilization.