Eight days out from the July 4, 2026 begin-construction deadline, the conversation in the solar-finance market has rotated again. Through May the question was whether the safe-harbored pool would clear 200 GWdc (it did, by Wood Mackenzie’s 216 to 240 GWdc range). Through early June the question was whether the IRS Notice 2025-42 vacatur would reopen the 5 percent cost safe-harbor as a working pathway (the June 6 ruling did, and Treasury has not appealed). The question now, with the credit-eligibility piece settled at the headline level, is what the audit lookback actually does to a 240 GW pool of project-level claims over the next 5 to 7 years. The answer is not abstract. It is showing up in tax-equity flip yields, in tax-insurance premium quotes, and in the discount the secondary market is applying to projects whose begin-of-construction documentation rests on the physical-work pathway versus the cost safe-harbor.
What the audit window actually is
Start with the statutory baseline. Section 6501(a) of the Internal Revenue Code sets the general assessment period at 3 years from the date the return is filed. For an investment tax credit claimed under Section 48E, the relevant return is the placed-in-service-year corporate return, not the begin-construction-year return. A project that placed in service in 2028 against a 2026 safe-harbor would have a default 3-year IRS audit window running through roughly 2032 (assuming a 2029 return filing for tax year 2028).
The 3-year baseline is the soft floor, not the realistic exposure. Two extensions matter for the OBBBA-safe-harbored cohort. First, Section 6501(e)(1)(A) extends the assessment period to 6 years where the taxpayer omits gross income exceeding 25 percent of the amount reported. Tax credits are not gross income, but the practical Service position (consistent with the audit posture taken on IRA-era 1603 grant claims and on Section 1603 disallowance litigation) is that ITC overstatement at the 25 percent threshold triggers the extended period through a related-element analysis. The Tax Court has not foreclosed this. Second, Section 6501(c)(1) suspends the limitations period entirely in cases of fraud, and the IRS has historically read aggressive begin-of-construction documentation as adjacent to that line where physical-work evidence is reconstructed after the fact.
The practical read for a tax-equity counsel sitting on a $400 million ITC structure: 3-year exposure is the optimistic scenario, 6-year exposure is the working assumption, and indefinite exposure is the tail risk that has to be insured. For a project that places in service in 2029 under the four-year continuity window, that pushes meaningful audit risk into 2036. The credit clock locks at July 4, 2026. The audit clock runs through the end of the decade.
What the documentation pile-up actually looks like
The 216 to 240 GWdc of safe-harbored capacity sits on two qualifying pathways, and the audit risk profile differs sharply between them.
Physical-work pathway projects (the dominant approach through 2024 and into early 2025) rely on demonstrating that significant physical work of a significant nature began before July 4, 2026. The Treasury continuity-of-construction guidance and the underlying 2013-29 and 2018-59 notices set the conceptual framework, and post-IRA Treasury practice on Section 45 and Section 48 has applied it to manufacturing of components, site mobilization, racking installation, and trenching. The audit question is whether the physical work was both significant in nature and continuous from the begin-construction date through commercial operation. A project that mobilized in June 2026 and then paused for 18 months of interconnection-study delays has continuity exposure that the cost safe-harbor pathway does not carry.
The 5 percent cost safe-harbor (revived by the June 6 District Court ruling vacating IRS Notice 2025-42) requires the taxpayer to have paid or incurred at least 5 percent of total project cost by the deadline, plus continuous efforts toward project completion. The audit question is dual: did the 5 percent threshold actually clear, and were efforts continuous. The first question is arithmetic. The second is judgmental. Practitioners are converging on a position that documents board-approval activity, financing milestones, EPC scope progression, and interconnection-application advancement as the evidentiary spine for continuity, but the IRS has not issued post-vacatur guidance and the audit posture remains to-be-determined.
A third documentation pile that matters: the cell-and-module safe-harbor stockpile concentrated at the developer level (where modules were purchased and warehoused but project assignment came later). The IRS position on inventory-to-project assignment has tightened since 2024, and the audit question on these inventories is whether the begin-of-construction date attaches to the purchase or to the assignment. The conservative position assigns the date at project allocation, which compresses the safe-harbor benefit for inventory that sits in a warehouse for 18 to 24 months before being assigned to a specific project.
Across the 240 GWdc pool, the distribution is roughly: 55 to 65 percent on physical-work pathway, 20 to 30 percent on cost safe-harbor (a number that is climbing post-vacatur), and 10 to 20 percent on warehoused inventory awaiting assignment. The audit risk concentrations track that mix.
The 1603 precedent
The Section 1603 grant program (the IRA’s cash-grant predecessor under the American Recovery and Reinvestment Act) is the closest historical analog to the post-OBBBA audit cycle. Between 2009 and 2012, Treasury awarded approximately $26 billion in cash grants to renewable projects under the begin-construction framework. The audit and recapture cycle that followed has played out over 10 to 12 years, with the Federal Claims Court and Federal Circuit handling the disputed cases, most notably in solar cost-basis disallowance litigation (the Alta Wind and California Ridge lines).
Two lessons from 1603 inform the OBBBA pool. First, the audit cycle is granular. Treasury and IRS reviewed individual project files, requested cost-basis substantiation, and disallowed portions of claims at the project level rather than across portfolios. A 240 GW pool will produce a multi-thousand audit caseload across the 2029 to 2036 window. Second, the disputed cases concentrate on cost-basis stacking (developer fees, related-party transactions, marketing intangibles) rather than on the begin-construction date itself. The OBBBA cohort, where begin-construction is itself the gating element, will see a different distribution of disputes weighted toward the date-establishment question.
The 1603 disallowance rate is informative. Of the approximately $26 billion awarded, contested disallowances ran in the 2 to 5 percent range by dollar value, with a long tail of settlements and partial concessions. Applied to the OBBBA pool, where ITC value on 240 GWdc at conservative 30 percent ITC and approximate $1,000/kWdc installed cost is roughly $72 billion of credit value, a comparable disallowance rate translates to $1.4 to $3.6 billion of contested credit. The tax-equity market is pricing for the upper half of that range.
How tax-equity yield is responding
Tax-equity flip yields on utility-scale solar have moved 40 to 70 basis points wider on safe-harbored project structures since the June 6 vacatur ruling, and the spread between physical-work and cost-safe-harbor project structures has compressed by roughly 20 basis points (the cost-safe-harbor pathway is now priced closer to physical-work, where pre-vacatur it carried a 50-basis-point premium because the pathway was unavailable). The widening is not a credit-event repricing. It is the audit-risk premium catching up to the size of the safe-harbored pool.
Three signals from term sheets seen in May and June 2026:
Tax-indemnity insurance market. Premium pricing on tax credit insurance policies covering ITC recapture and disallowance has firmed by 80 to 150 basis points of policy limit since April. The major carriers writing this risk (the AIG, Liberty, and Euclid books, plus the specialty MGAs) are quoting policies on safe-harbored projects with explicit carve-outs for begin-construction-date disputes, which leaves the audit risk uninsured on the most contested element. Brokers are placing the residual exposure into syndicated layers at materially higher premiums than 2024 comparables. The total cost of tax insurance on a typical $400 million ITC structure has risen from roughly 1.5 to 2.0 percent of policy limit in 2024 to 2.8 to 3.5 percent post-vacatur, and the policy exclusions have widened.
Tax-equity investor diligence. The diligence packages requested by tax-equity investors on safe-harbored projects in May and June 2026 are running 60 to 90 days longer than 2024 timelines and are heavier on begin-of-construction documentation. Investors are requiring physical-work evidence (dated site photographs, contractor sign-offs, drone surveys, time-stamped equipment delivery logs) and cost-safe-harbor evidence (invoices, wire confirmations, board resolutions, EPC change-order documentation) in parallel where the developer claims both pathways. The diligence headcount required has roughly doubled per deal, and the timelines have stretched into commitment lapses on weaker documentation packages.
Secondary-market discount. Projects sold into the secondary market in June 2026 with safe-harbored credit positions but thin physical-work documentation are trading at 4 to 8 percent discounts to comparable projects with strong documentation packages. The discount is not credit-eligibility uncertainty (the projects are eligible). It is the present value of expected audit exposure plus insurance premium drag.
The Notice 2025-42 vacatur ripple
The June 6 District Court ruling vacating IRS Notice 2025-42 restored the 5 percent cost safe-harbor as an available pathway, but it also created an interim guidance vacuum. Notice 2025-42 had narrowed the cost safe-harbor and tightened the inventory-assignment rules. Vacatur returned the pre-Notice framework, which is the 2013-29 and 2018-59 baseline, but Treasury has not issued replacement guidance and has not committed to a timeline for doing so.
The practical effect for developers safe-harboring in the next 8 days is asymmetric. Projects that already established physical-work documentation are unaffected. Projects that pivoted to cost safe-harbor after the late-2025 Notice 2025-42 publication now have a pathway available again, but they are documenting against a framework that may be replaced retroactively by post-deadline Treasury guidance. The legal positions being taken assume that any replacement framework will not apply retroactively to begin-construction dates already established, which is the better reading of administrative law precedent, but it is not airtight.
The audit-risk consequence: projects safe-harbored under post-vacatur cost-safe-harbor documentation are sitting on a residual interpretive risk that physical-work projects are not. Tax counsel is pricing it. Lenders are pricing it. The flip-yield differential captures part of it.
What to watch into 2027
Four datapoints over the next 12 months will calibrate the audit-cycle exposure.
The Treasury and IRS replacement guidance posture. If Treasury issues replacement guidance for the vacated Notice 2025-42 in Q3 or Q4 2026, the substance and the effective-date language will determine whether post-vacatur safe-harbored projects retain their pathway. A non-retroactive replacement is the consensus expectation. A retroactive replacement would be litigated to the Federal Circuit and would compress the pool of secure cost-safe-harbor projects by 15 to 30 percent until resolution.
The Q3 2026 tax-equity term sheet trend. Yield levels and insurance-premium quotes through the third quarter will indicate whether the current widening is a vacatur-event repricing that compresses back or a structural shift to a higher post-OBBBA yield regime. The market consensus from major tax-equity arrangers (the JPM, Bank of America, and US Bank books, plus the specialist tax-equity funds) is that 30 to 40 basis points of the widening is structural and that 30 to 40 basis points compresses back over 6 to 12 months as documentation diligence standardizes. A tighter-than-expected compression argues that the market underweighted audit risk. A wider-than-expected persistence argues the opposite.
The first wave of IRS Letter 525 audit notices on 2026 placed-in-service safe-harbored projects. The earliest projects in the safe-harbored pool that will place in service are the late-stage developments where construction was already well advanced before the safe-harbor deadline. Those projects file 2026 returns in 2027 and become audit-eligible in 2027 to 2030. The opening of the audit cycle on this cohort will set the precedent for the 2028 to 2032 placed-in-service wave, which is the bulk of the pool. Audit selection rates, disallowance theories, and settlement postures from the first 50 to 100 audits will calibrate the rest.
The tax-credit-transfer market repricing. The OBBBA preserved tax-credit transferability under Section 6418, and the transfer market has been the primary clearing mechanism for safe-harbored ITC. Transfer prices were running 92 to 95 cents on the dollar through early 2026. Post-vacatur transfer prices for safe-harbored credits have widened to 88 to 92 cents on the dollar with tighter buyer indemnities and shorter transfer windows. The next 12 months of transfer pricing will indicate whether the audit-risk repricing flows fully through to the cash market for credits or whether the tax-equity structure absorbs more of it.
The shape of the question
The headline reading on July 4 has been that the credit cliff is, for safe-harbored solar, already discounted. That reading is correct as far as it goes. The follow-on reading, less discussed because it is operationally invisible until 2029 to 2032, is that the audit lookback creates a structural exposure that the credit-eligibility question does not. The audit cycle will run for the full decade after placed-in-service, and the disallowance distribution will produce a $1.4 to $3.6 billion contested-credit pool, concentrated in physical-work documentation gaps and inventory-assignment disputes.
The H2 2026 tax-equity market is the first surface where that exposure is showing up in prices. The next surfaces, in order, are the tax-insurance market (already firming), the secondary project-sale market (already discounting), and the credit-transfer market (in the early stages of repricing). By the time the first wave of audit notices opens in 2027, the pricing will have moved another 30 to 50 basis points on average, and the developers and investors who sized the audit-risk premium correctly through H2 2026 will be the ones with the cleanest 2028 to 2032 cash flows.
Credit is locked. The audit clock starts now.