Solar after subsidies: the panel is solved, the paperwork isn't

The short version

Solar won on cost years ago, and the module glut made the panel close to a rounding error. The federal credits are now a closing window: a safe-harbored book that runs to 2030, then a cliff. The margin now sits in three places: proving compliance to the credit buyer, delivering power in the hours an offtaker will pay for, and clearing permitting and interconnection faster than the next project. If you sell into solar, you are selling into one of those three problems, not a cheaper watt.

The mechanism

Two forces pushed in opposite directions, and the gap between them is where the money now gets decided.

The first is oversupply. Global factory capacity runs well ahead of demand, so module prices collapsed and the hardware lost its margin. The US doesn't get that price, because tariffs and origin rules keep domestic modules at a multiple of global spot. Either way, hardware no longer decides whether a project pencils.

The second is the subsidy unwind. The 2025 tax law (OBBBA) put a hard start-of-construction deadline on wind and solar under §45Y and §48E, ended the homeowner credit, and added foreign-entity (FEOC) rules with a ten-year recapture tail. A credit used to work like a coupon. Now it is an underwriting problem. The buyer pays face value minus what it thinks clawback risk is worth, so a developer who documents compliance from day one keeps more of the credit.

Put those together and solar's economics turn on things the panel doesn't control: the offtake contract, the hours the power arrives, the queue position and the evidence file.

How it developed

• April 10 post on the 87% cliff. My first framing: the drop in planned buildout after 2028 reflects the credit deadline, not falling demand.
• May 1 post on the $0.09 module. The global module price against a US price roughly three times higher. The thread pushed hard on what "a watt" means, and that shaped everything after it.
• June 20 and July 3 posts on pull-forward. In both the US and China, deployment numbers that looked like transition signals were partly deadlines pulling projects forward.
• July 14 post on Spain. Spain also lives in the storage thread, so only the solar-economics point here: the grid worked and consumers benefited, while merchant solar capital didn't get paid. Those are separate questions.
• July 22 post on ITC underwriting. This is where the thread turned. The buyer prices the credit net of FEOC and recapture risk, and deals that build in proof from the start keep moving.
• July 30 post on the utility-by-utility sort. With the residential credit gone everywhere at once, demand re-sorted around rate trajectory, outage exposure and permitting speed.
• August 6–26 posts. Arctic solar: the price of power, not latitude, decides where solar makes money. Utah: contracts outran federal policy. Transshipment: customs origin rules move more trade than solar policy does. India: capital wants storage or a firm offtaker before it writes the check.
• October 1 and 5 posts. Panels are a small share of a US rooftop's installed cost. Annual installs fell while the cumulative chart climbed, and Texas, not California, led additions.
• Smaller items. California's plug-in solar bill (September 12) and Cuba's solar imports (October 15) pointed the same way: where the grid is strained, resilience gets bought behind the meter.

Where it stands now

• The start-of-construction deadline has passed. Projects had to begin construction by July 4, 2026. A project that missed it has to be placed in service by December 31, 2027 to stay eligible (One Big Beautiful Bill Act; McGuireWoods). Safe-harbored projects can run to the end of 2030 (SEIA/Wood Mackenzie via pv magazine USA, September 10, 2026).
• The 5% safe harbor came back late. On June 6, 2026, the D.C. district court vacated IRS Notice 2025-42 (Oregon Environmental Council v. IRS). That restored the 5% cost safe harbor for large solar shortly before the deadline (Stoel Rives, June 10, 2026). One update reported no government appeal as of early September. I could not confirm the current appeal or remand status.
• FEOC rules apply, but only in interim form. IRS Notice 2026-15 (February 12, 2026) set interim material-assistance rules for projects beginning construction after December 31, 2025. Treasury still intends to propose regulations and new safe-harbor tables, and I found no record that they have been issued yet.
• The deadline rush is still landing. The US installed 11.4 GWdc in Q2 2026, up 45% year over year, driven by utility-scale while residential fell. The report names permitting and interconnection as the binding constraints (SEIA/Wood Mackenzie, Q3 2026 Solar Market Insight, September 10, 2026).
• Import prices get a floor on December 4. A Section 232 proclamation (August 6, 2026) sets minimum import prices, including $0.38/W for modules and $0.22/W for cells, plus tariffs capped at 15% for some partners (BDO; Mercom, August 2026). I did not read the proclamation text.

The pushback worth taking seriously

• "Define the watt." Commenters kept pressing on the gap between module price, installed cost and delivered energy, and between capacity and energy. They were right, and it moved my read. That gap is now the thesis: the panel is solved, and everything else is where cost and margin sit.
• Firming cost belongs on solar's bill. A recurring objection was that storage and backup should be charged to solar, and another pushed back that an annual-share figure holds only in certain hours. I accept the precision point. Separately, capital now prices the shape of output, which is why storage attach is becoming part of bankability.
• It's offtake, not finance. On the India post, commenters argued that money is available but utility buyers won't sign for solar without storage. That's sharper than my framing. Bankability is a contract question first.
• Plug-in solar has real limits. Circuit safety, insurance and meters that miscount exports came up repeatedly. Fair: it lowers the cost of entry, it isn't backup power.
• The cliff is specific to wind and solar. Other technologies keep a slower phase-down, so "clean energy loses its credits" overstates it.

What it means if you sell into this

For solar developer and solar-finance founders, these are the levers I'd look at:

1. Sell proof, not savings. A credit buyer discounts for FEOC tracing, construction-start evidence and recapture exposure. An audit-ready file from day one shrinks that discount.
2. Sell the hour, not the megawatt. If you improve the shape of output (storage, dispatch, offtake structure), lead with the offtaker's question: can you deliver when it's worth something?
3. Go after soft costs. In residential and small commercial, permitting, interconnection and customer acquisition are most of the bill, and volume has moved to leases and PPAs that still reach §48E.
4. Sort accounts by utility territory, not state. Rate trajectory, outage history and permitting speed predict demand better than the incentive table.
5. Plan for the 2030 edge. The safe-harbored book runs to 2030. Buyers already modeling unsubsidized economics are the ones to talk to now.

The question I'd want answered first: If your book still has to close after 2030, is your compliance file one a credit buyer would sign off on today?

What would change my read

• Treasury issues replacement beginning-of-construction guidance, or an appeal reinstates Notice 2025-42, and it narrows what counts as safe-harbored.
• Final FEOC regulations turn out simple enough that compliance stops differentiating anyone.
• Solar-only PPAs without storage keep signing at scale in the big merchant markets after 2027.

If you're already seeing one of these, I'd rather hear it now than find it in the next filing.

What to watch

• December 4, 2026: Section 232 price floors take effect. Do US module prices in SEIA/Wood Mackenzie's Q1 2027 Solar Market Insight (expected around March 2027) move toward the $0.38/W floor?
• SEIA/Wood Mackenzie Q4 2026 Solar Market Insight (expected around December 2026): Does contracted utility-scale volume hold up in the first full quarter after the July 4 deadline?