Solar ITC Cliff: Thesis Test — What Actually Happened
Nine months ago I predicted a 50-75% residential solar surge followed by a 30-40% drop. Here's how that forecast held up against actual 2025 data and what 2026 looks like now.
In June 2025 — about a week before the One Big Beautiful Bill Act was signed into law — I published an analysis predicting that the expiration of the 30% residential solar Investment Tax Credit would produce two distinct moves: a 50-75% installation surge in 2025 as homeowners rushed to lock in the credit, followed by a 30-40% demand drop in 2026 once it was gone.
Nine months on, there’s enough data to grade the forecast. The short version: the policy setup landed exactly right, but the behavioral prediction did not. Here’s the full scorecard and what I’d do differently.
Scorecard
| Prediction | Actual outcome |
|---|---|
| ITC expires end of 2025, no phase-down | Correct — OBBBA signed July 4, 2025; Section 25D eliminated December 31, 2025 |
| 50-75% residential surge in 2025 | Wrong — residential was down ~7% year-to-date through Q3 2025 |
| 43% cost jump for cash/loan buyers post-ITC | Correct mechanically for customers who own their systems |
| 30-40% demand drop in 2026 | Directionally right, but magnitude likely overstated; SEIA models ~19% |
| NEM 3.0 as the best demand analog | Partially — elasticity framework was reasonable, but the analogs diverged on key structural differences |
What the analysis got right
The policy call was clean. The OBBBA eliminated Section 25D with no phase-down, exactly as the Senate and House bill texts indicated at the time. The cost math — a 43% out-of-pocket increase for a cash or loan buyer on a 7 kW system — holds for anyone who owns their system outright.
The framing that “a cliff reshapes behavior” also proved directionally correct, just not in the way I expected. Behavior did shift dramatically — toward safe-harboring equipment and toward third-party ownership structures — rather than toward a pull-forward installation rush.
What the analysis missed
The supply chain couldn’t have absorbed a frenzy. The analysis acknowledged installer capacity as a lower-bound constraint. What it didn’t anticipate was module sell-outs. By Q4 2025, manufacturers and distributors were sold out of both domestic and imported modules through year-end 2026. Even if every homeowner who wanted solar had tried to rush installations, the supply chain couldn’t have delivered them. Any panic-buying got throttled before it showed up in installation numbers.
The lease loophole turned a cliff into a channel shift. The original analysis treated the ITC as a binary on/off. In reality, Section 48E — the commercial clean energy credit — lets third-party owners of residential solar (leases and PPAs) continue claiming the ITC through 2027. Nearly half of national residential solar installations were customer-owned in H1 2025; the other half, operating under third-party ownership, kept access to the credit. That escape hatch removed the urgency for a large portion of potential buyers: if you can lease instead of buy, there’s no cliff to race against.
The baseline was already soft. The 2016 and NEM 3.0 analogies assumed a healthy baseline to surge from. In 2025, high interest rates had already compressed residential solar demand — SEIA reported a 13% year-over-year drop in Q1 2025. Applying a 1.5-1.8x surge multiplier to a declining baseline produces a very different number than applying it to a growing one.
The 2026 outlook
As of the SEIA/Wood Mackenzie 2025 Year in Review (March 2026), residential solar installed 4,647 MWdc in 2025 — essentially flat, down 2% from 2024. The frenzy that would have made 2026’s cliff dramatic simply didn’t happen.
For 2026:
- Residential — Wood Mackenzie forecasts a ~19% contraction following Section 25D expiration. The remaining TPO eligibility for the ITC and bonus adders cushions the decline and supports recovery beginning in 2027. Safe-harboring activity at the end of 2025 will support TPO ITC qualification through mid-2030.
- Commercial — a ~13% contraction expected as developers complete the last NEM 2.0 projects in California and shift to smaller builds under the Net Billing Tariff.
- Community solar — the bright spot, with 12% growth expected, driven by markets outside New York and emerging states like New Mexico, Virginia, and Delaware.
- Utility-scale — nearly 44 GWdc expected in 2026, roughly flat with 2025’s 43 GW, partly because projects that slipped out of Q4 2025 are now landing in 2026.
The structural demand drivers — rising retail electricity rates, falling equipment costs, and grid resiliency concerns — remain intact. SEIA forecasts more than 60 GWdc of residential additions between 2026 and 2036 even without the credit.
The real test for the 30-40% drop thesis is 2027, when even the Section 48E credit begins winding down for projects that didn’t safe-harbor in time.
What I’d weight differently next time
The elasticity framework — price and payback elasticity applied to historical analogs — was reasonable. The problem was analog selection. The 2015 rush and NEM 3.0 both happened in markets with functioning supply chains and no structural escape hatch. This situation had both.
A more accurate model would have:
- Capped the demand surge estimate at installer and module supply, not just installer capacity
- Segmented the market by ownership structure (owned vs. TPO) before applying any behavioral multiplier, since TPO buyers face a completely different incentive structure
- Used a softer baseline given rate-driven demand compression already in progress
The cost math and policy read held. The behavioral prediction required more market structure than the framework carried.
All data sourced from SEIA/Wood Mackenzie 2025 Year in Review (March 2026), SEIA Q3 2025 Solar Market Insight, and POWER Magazine.