Beginning of Construction: The Physical Work Test, the 5% Safe Harbour, and a Vacated Notice

Beginning of construction is a date that decides whether a project has a tax credit. It is established by satisfying one of two tests, and for the last twelve months the question of how many tests exist has itself been contested — eliminated by notice in August 2025, restored by a federal court in June 2026, and now sitting under an appeal whose outcome could apply retroactively.

A developer today has to document both routes simultaneously, because the law under which its position will eventually be judged is not yet settled. That is an unusual thing to have to model, and it is the operating reality heading into the 4 July 2026 deadline.

This is the first post in a new series on policy and regulation, and the first of three on beginning of construction. You will get what the two tests require, what counts as physical work of a significant nature and what is explicitly excluded, exactly what Notice 2025-42 did and what the court did to it, and how to size a safe harbour cushion against the cost overrun that can retroactively defeat it.

ℹ️ Note: This describes how these rules work in practice. It is not tax or legal advice, and the position described is unsettled — confirm the current status of the appeal before relying on any of it.

Why Does Beginning of Construction Matter?

Because it fixes eligibility. Establishing that construction began before a statutory deadline is what preserves a project's credit, and the consequence of failing is not a smaller credit — it is no credit at the rate the financing assumed.

It also starts the continuity clock that determines how long the project has to reach placed-in-service, which is the subject of the next post in this series. So the date does two things at once: it qualifies the project, and it begins the countdown on that qualification.

For financing purposes the date is a condition precedent in substance even when it is not one in form. Tax equity prices off credit certainty, and a beginning-of-construction position that cannot be evidenced is a position that gets discounted or indemnified rather than funded.

What Are the Two Routes?

Historically, either demonstrating physical work of a significant nature or paying or incurring at least 5% of the total cost of the facility before the deadline.

	Physical work test	5% safe harbour
What it requires	Physical work of a significant nature, on site or off site	Paying or incurring ≥5% of total facility cost
The standard	Nature of the work, not amount or cost	Purely quantitative
Main risk	Whether the work qualifies	Cost overrun retroactively failing the 5%
Evidence	Contracts, work records, manufacturing progress	Invoices, payments, delivery and title records
Status today	Always available	Eliminated Aug 2025, restored June 2026, under appeal

They are alternatives, not cumulative. Satisfying either one establishes the date — which is precisely why removing one of them mattered so much.

What Counts as Physical Work of a Significant Nature?

Work on the facility itself or on components integral to it, judged by what the work is rather than by how much of it has been done. The governing standard is that the test of significance focuses on the nature of the work performed, not the amount or the cost.

That distinction is the one developers most often get wrong in both directions. A small amount of genuinely integral work qualifies. A large amount of preparatory expenditure does not.

The canonical example is a transformer. Physical work on a custom-designed transformer that steps up the voltage of electricity produced at the facility to the voltage needed for transmission is physical work of a significant nature, because power conditioning equipment is an integral part of the activity the facility performs. The transformer is not incidental to generating electricity — it is part of how the electricity gets delivered.

Off-site work counts on the same logic. The manufacture of components, mounting equipment, support structures such as racks and rails, inverters, and transformers used in electrical generation that step up voltage to less than 69 kilovolts, together with other power conditioning equipment, can all constitute physical work of a significant nature.

Two practical features follow from "off-site counts."

The work can be somebody else's. Manufacturing performed by a supplier under a binding written contract can establish the date for the project, which is why safe harbour equipment orders are a standard development tool.

Custom matters. The transformer example turns on the equipment being custom-designed for the facility. Work producing inventory items held for general sale does not have the same connection to a specific project, and the distinction between bespoke and off-the-shelf is where diligence concentrates.

What Is Explicitly Excluded?

A defined list of preliminary activities, excluded even where their cost is properly capitalised into the facility's depreciable basis. That last qualifier is the trap — spending money that appears in the project's basis does not mean the spending counted.

The excluded activities are: planning, designing, financing, exploring, researching, permitting, environmental and engineering studies, site clearing, test drilling unrelated to foundations, and removal of old equipment.

Read that list against a typical development budget and the problem is obvious. Almost everything a developer spends money on in the two years before construction falls inside it. Permitting, interconnection studies, environmental work, engineering, financing costs — all real, all capitalised, none of it physical work of a significant nature.

The exception hiding in the list is worth noting: test drilling unrelated to foundations is excluded, which implies drilling that is related to foundations may qualify. Foundation work is physical work on the facility; geotechnical investigation for its own sake is not.

ℹ️ Note: Site clearing is excluded but excavation for foundations is not the same activity. The boundary between preparing a site and beginning to build on it is where the physical work test is actually contested, and contemporaneous photographs and work records are the evidence that settles it.

What Did Notice 2025-42 Do?

Removed the 5% safe harbour for most wind and solar. Issued in August 2025 in response to Executive Order 14315, the notice generally eliminated the 5% safe harbour for purposes of the section 45Y clean electricity production credit and the section 48E clean electricity investment credit for wind and solar projects.

It did not remove it entirely. As McGuireWoods described at the time, the notice left the guidance mostly unchanged but limited the 5% safe harbour to low-output solar facilities — those of 1.5 MW or less. Everything larger, and all wind, was left with the physical work test as the sole route.

The notice applied prospectively from 2 September 2025, so positions established under prior guidance before that date were not disturbed.

The practical effect was to convert a quantitative, documentable test into a qualitative one for the bulk of the market. A 5% safe harbour position is an invoice trail. A physical work position is a judgement about whether what was done was significant in nature — harder to evidence, harder to insure, and considerably harder to close a tax equity investment against.

What Did the Court Do?

Vacated it. On 6 June 2026 the US District Court for the District of Columbia decided Oregon Environmental Council v. Internal Revenue Service, Case No. 25-4400 (CKK), and threw the notice out.

McGuireWoods sets out three independent grounds on which the court found the notice arbitrary and capricious:

• "inadequate explanation for a major policy change" — insufficient justification for eliminating a method that had been available for over twelve years
• failure to address serious reliance interests, despite explicit industry comments raising them
• "no explanation for why the 5% Safe Harbor was eliminated for wind and large-scale solar projects while remaining available for other technologies"

Orrick describes the same reasoning: Treasury "failed to provide a reasoned explanation for treating wind and large-scale solar projects differently from other technologies eligible for technology-neutral tax credits" and for "departing from longstanding IRS guidance."

The scope is broad. The court granted universal vacatur rather than limiting relief to the plaintiffs, holding that this was necessary to address injury flowing from the notice's effects on the broader market. The decision restores the 5% safe harbour as an available method of establishing beginning of construction for wind and large-scale solar under sections 45Y and 48E.

Where Does That Leave a Developer Today?

Documenting both routes, and assuming neither is settled.

Two cautions from the same sources define the position. McGuireWoods notes that "the government will almost certainly seek a stay of the vacatur pending appeal," and — the part that matters for risk — that "a reversal of the court's holding could have retroactive effect," so the decision should not be treated as final pending appellate resolution. Orrick observes that any appellate ruling, and any revised IRS guidance, is "unlikely to occur before July 4."

That combination is unusual and uncomfortable. The deadline arrives before the legal question is resolved, and the resolution may reach backwards.

The practical response both firms recommend is the same: maintain thorough recordkeeping, and continue "preserving flexibility and documenting compliance under both beginning-of-construction pathways" until clarity emerges. In other words, treat the 5% safe harbour as available and simultaneously build a physical work position as though it were not.

That is more expensive than either route alone. It is also considerably cheaper than discovering after an appellate reversal that the only position you documented was the one that got vacated.

ℹ️ Note: For a project relying on the restored safe harbour, the diligence question a tax equity investor will ask is not "did you spend 5%?" but "what is your position if the vacatur is reversed retroactively?" Have an answer before the question arrives.

Why Does a Cost Overrun Defeat a 5% Position?

Because the test is five percent of the total cost of the facility, and the total cost is not known on the date the test is applied. A developer spends against an estimate and is judged against an outturn.

This is the structural weakness of a test that looks purely arithmetic. Spend 5.2% of a projected cost, then watch the project come in 15% over budget, and the position that looked comfortable is retrospectively short. Nothing about the spending changed; the denominator moved.

Which is why safe harbour spending is sized with a cushion rather than at the threshold, and why the size of that cushion should be derived from the project's own cost risk rather than from a convention.

How Does the Single Project Rule Change the Calculation?

It changes what "the facility" means, and therefore what the denominator is. Beginning of construction is determined at the level of a single project, and a single project can comprise multiple facilities — so one qualifying act can establish the date across a portfolio.

Aggregation is a facts-and-circumstances test rather than an election. Troutman Pepper Locke sets out the factors weighed in deciding whether multiple facilities form one project, including whether they are:

• owned by a single legal entity
• constructed on contiguous land
• described in a common power purchase agreement
• served by a common intertie
• sharing a common substation
• described in common environmental permits

Multiple facilities operated as part of a single project may be aggregated and treated as a single facility solely for the purpose of determining the beginning-of-construction date. That word does real work — aggregation for this purpose does not merge the facilities for every other purpose.

Two consequences follow, and they pull in opposite directions.

Aggregation is efficient. Safe harbour spending or physical work on one part of a portfolio can establish the date for the whole, which is why developers order equipment against a project rather than against an individual array. It also means the 5% denominator is the whole project's cost, not one phase's — a larger number, but reached once.

Disaggregation is the escape hatch. Facilities treated as a single project may later be disaggregated and treated as separate energy properties for continuity purposes. Those placed in service before the continuity deadline qualify under the safe harbour; the remainder must satisfy the continuity requirement on the facts. That flexibility is the direct mitigation for a portfolio where some phases run late — a subject the next post takes in full.

The planning implication is to decide the project boundary deliberately rather than discover it in diligence. A boundary drawn widely reduces the number of qualifying acts required and raises the cost denominator. Drawn narrowly, it does the reverse and creates more positions to evidence. Neither is right in the abstract; what is wrong is drawing it one way in the development budget and another way in the tax memorandum.

ℹ️ Note: The aggregation factors are evidentiary as much as legal. Common interties, shared substations and a single PPA are facts recorded in documents that already exist — which means the project boundary a developer intends should be visible in the interconnection agreements and offtake contracts, not asserted for the first time in a tax opinion.

How Do You Size a Safe Harbour Cushion in Excel?

Solve for the spend that survives your own downside cost case, not the spend that satisfies the base case.

The inputs

Assumptions, labelled as such:

Base case total facility cost        $250,000,000
P90 cost outcome (downside)          $287,500,000    (15% overrun)
Statutory threshold                  5.0%
Planned safe harbour spend           $13,500,000
Deadline                             2026-07-04

The base case test

Required_Base   = 250,000,000 × 5.0%              = $12,500,000
Planned_Spend   = $13,500,000
Ratio_Base      = 13,500,000 / 250,000,000        = 5.40%
Headroom_Base   = 13,500,000 − 12,500,000         = $1,000,000   → passes

The downside test

Required_P90    = 287,500,000 × 5.0%              = $14,375,000
Shortfall       = 14,375,000 − 13,500,000         = $875,000     → FAILS
Ratio_At_P90    = 13,500,000 / 287,500,000        = 4.70%

The same spend that gave a million dollars of headroom against the budget is $875,000 short against a fifteen percent overrun. No decision was made badly; the denominator simply grew.

Solving for the cushion

Spend_Required  = P90_Cost × 5.0%                 = $14,375,000
Cushion_%       = Spend_Required / Base_Cost      = 5.75% of base cost
Additional      = 14,375,000 − 13,500,000         = $875,000

Expressed the useful way: to survive a 15% overrun you must spend 5.75% of the base case, not 5%. The general form is 5% × (1 + overrun%), which makes the trade-off explicit — every point of cost uncertainty costs five basis points of additional safe harbour spend.

The break-even table worth building

Run overrun percentage down the rows and planned spend across the columns, with pass/fail in the cells:

Max_Overrun_Survivable = (Planned_Spend / (Base_Cost × 5%)) − 1
                       = (13,500,000 / 12,500,000) − 1        = 8.0%

One number, and it is the one to report: this position survives an 8% cost overrun and no more. That is far more useful to a board than "we spent 5.4%," because it states the risk in the units the risk actually arrives in.

ℹ️ Note: Model the physical work position in parallel as a binary flag with an evidence checklist rather than as a number. The two routes fail in completely different ways — one to arithmetic, one to judgement — and a single combined "BOC: yes" cell hides which one is carrying the position.

To run the full version — safe harbour cushion solved against your own cost distribution, both pathways tracked separately with their evidence, and the placed-in-service clock started from the established date — prompt Dezzmond with your development budget and spend schedule.

What Do Lenders and Tax Equity Actually Check?

The question is never whether a date was established. It is whether the position survives scrutiny by someone hostile.

• Which route is the position built on? And is the other one documented as a fallback.
• If it is the 5% safe harbour, what cost overrun does it survive? Stated as a percentage, not as a dollar spend.
• Was the spend paid or incurred on the correct basis? Including whether title and delivery requirements were met for equipment.
• If it is physical work, what exactly was done and when? Contemporaneous records, not a later reconstruction.
• Was any of the claimed work on the excluded list? Permitting, engineering and site clearing are the recurring offenders.
• What is the position if the vacatur is reversed on appeal, retroactively? This is now a standard diligence question and it needs an answer.
• Who bears the risk if the position fails? Sponsor indemnity, purchase price adjustment, or a reduced credit assumption in the model.

Frequently Asked Questions

What are the two ways to establish beginning of construction?

Demonstrating physical work of a significant nature, or paying or incurring at least 5% of the total cost of the facility before the deadline. They are alternatives — satisfying either establishes the date.

Is the 5% safe harbour currently available for large solar and wind?

As of the June 2026 decision, yes. The US District Court for the District of Columbia vacated Notice 2025-42 on 6 June 2026 with universal effect, restoring the safe harbour. The government is expected to seek a stay pending appeal, and a reversal could apply retroactively.

Does spending money that goes into the project's tax basis count as physical work?

No. Preliminary activities are excluded even where their cost is properly included in the depreciable basis — planning, designing, financing, permitting, environmental and engineering studies, site clearing, test drilling unrelated to foundations, and removal of old equipment.

Can work performed off site establish beginning of construction?

Yes. Manufacture of components, mounting equipment, racks and rails, inverters and step-up transformers can constitute physical work of a significant nature, typically where the equipment is custom-designed for the facility rather than held as general inventory.

Why do developers spend more than 5% for the safe harbour?

Because the test measures against total facility cost, which is not final when the spending occurs. A cost overrun raises the denominator and can retroactively defeat a position that was comfortable against the budget.

Closing: A Date, Two Tests, and an Unresolved Appeal

Beginning of construction is the most consequential date in a US renewable project's tax life, and for the moment it rests on a legal position that has changed twice in a year. The physical work test has been constant throughout. The 5% safe harbour was removed for most of the market in August 2025 and restored by universal vacatur in June 2026, with an appeal outstanding and retroactive effect on the table.

The disciplined response is not to pick the winner. It is to document both pathways, size the safe harbour against a cost outcome rather than a cost estimate, and state the position in terms of what it survives rather than what it satisfies.

The next post takes the clock this date starts: the continuity requirement, the four-year safe harbour, and the excusable disruptions that stop it running.

Sources: McGuireWoods — Federal Court Vacates IRS Notice 2025-42 · Orrick — Court Throws Out Notice 2025-42 · McGuireWoods — Notice 2025-42 Limits 5% Safe Harbor to Low-Output Solar Facilities · Troutman Pepper Locke — IRS Beginning of Construction Guidance for Solar