The Continuity Requirement: Four Calendar Years, and Why One Day Can Cost You One
Establishing beginning of construction does not qualify a project. It starts a clock, and the project has to keep running against that clock until it is placed in service. Fail that second test and the first one evaporates — the date was established and then lost, which is worse than never having claimed it, because the financing was priced on the assumption it held.
The safe harbour that protects you is measured in calendar years, not elapsed time. A project that begins construction on 31 December 2025 has until 31 December 2029. One that begins a day later, on 1 January 2026, has until 31 December 2030. One day of difference, one full year of deadline.
This is the second post in the beginning-of-construction series. You will get what the continuity requirement actually demands, the two routes to satisfying it, exactly how the four-year deadline is computed and why the calendar-year mechanic matters more than it should, the excusable disruptions that stop the clock, and how disaggregation rescues a portfolio where only some phases finish in time.
ℹ️ Note: This describes how these rules work in practice. It is not tax or legal advice, and the governing guidance is currently unsettled following the June 2026 vacatur — confirm the position before relying on it.
What Does the Continuity Requirement Demand?
That you keep going. Having established a beginning-of-construction date, a taxpayer must maintain progress toward completion, and the question of whether progress was maintained is judged on facts and circumstances unless a safe harbour applies.
The requirement exists to stop the date being gamed. Without it, a developer could perform one qualifying act, stop for a decade, and claim a credit vintage that bore no relationship to when the project was actually built. Continuity is what ties the established date to a real construction programme.
Two routes satisfy it:
| Continuity safe harbour | Facts and circumstances | |
|---|---|---|
| The test | Placed in service within four calendar years | Did construction actually continue? |
| Evidence needed | A placed-in-service date | A full construction record |
| Certainty | Binary and provable | Judgemental, contestable |
| When it matters | Most projects, most of the time | Long-build projects and portfolios that slip |
Almost every project aims for the first. The second is what you fall back on, and it is considerably harder to close a tax equity investment against.
How Is the Four-Year Deadline Computed?
By calendar year, not by elapsed months. The formulation in the guidance is precise: a taxpayer satisfies the continuity requirement if the facility is "placed in service by the end of a calendar year that is no more than four calendar years after the calendar year during which construction began."
The worked example carried in the technical review by Current Federal Tax Developments makes the mechanic concrete: construction beginning on 20 August 2025 requires the facility to be placed in service by 31 December 2029.
Work through what that means and the asymmetry appears immediately.
- Begin construction on 5 January 2025 → deadline 31 December 2029 → 1,821 days
- Begin construction on 20 August 2025 → deadline 31 December 2029 → 1,594 days
- Begin construction on 31 December 2025 → deadline 31 December 2029 → 1,461 days
- Begin construction on 1 January 2026 → deadline 31 December 2030 → 1,825 days
The last two lines are the ones to sit with. Two BOC dates one day apart produce deadlines a full year apart. A project that establishes its date on 31 December gets four years; one that establishes it the following morning gets five.
That is not a loophole, it is the arithmetic of counting whole calendar years, and it cuts both ways. Late in a calendar year, establishing beginning of construction sooner can be actively worse for the continuity clock even though it is better for credit vintage — and those two considerations are frequently traded against each other without anyone naming the trade.
ℹ️ Note: Where a credit-vintage deadline and a continuity deadline both bear on the same BOC date, model them together. Pulling the date into December to catch a statutory cutoff buys eligibility and spends eleven months of build time, and the second cost rarely appears in the memo recommending the first.
What Is the Difference Between Continuous Efforts and Continuous Construction?
One asks whether you kept working. The other asks whether you kept building. The gap between them is wide enough to determine outcomes.
Continuous efforts is the broader standard. As K&L Gates describes, it could be satisfied through activities such as "entering into binding equipment contracts and obtaining permits" — commercial and administrative progress, not necessarily physical progress. A project waiting eighteen months for an interconnection upgrade while ordering equipment and closing permits is making continuous efforts.
Continuous program of construction is narrower. The requirement is to maintain "a continuous program of construction for the applicable wind or solar facility" by "continuing physical work of a significant nature." Ordering equipment does not satisfy it. Work has to be happening.
The distinction matters because of which projects it catches. A solar project with a two-year build and no interruptions satisfies either test without thinking about it. A project held up by a transmission upgrade, a permitting appeal or a supply chain failure may have excellent continuous efforts and no continuous construction at all, because there is nothing to build until the constraint clears.
K&L Gates notes that continuous efforts "had been extended in 2021 to projects relying on the Physical Work Test," which is the point at which the two standards effectively converged as available options.
What Did Notice 2025-42 Change, and What Did the Court Restore?
The notice narrowed the test. The vacatur removed the notice.
Notice 2025-42 preserved the four-year continuity safe harbour and the continuous program of construction test, but removed continuous efforts. The effect, per K&L Gates, was to replace the broader standard with "a stricter 'continuous construction' standard requiring 'continuing physical work of a significant nature'."
For a project outside the four-year safe harbour that is a material tightening. Under the old framework, a developer stalled on interconnection could point to binding contracts and permitting progress. Under the notice, it needed physical work — on a site where physical work was precisely what could not proceed.
Then, as covered in the previous post, the US District Court for the District of Columbia vacated the notice in full on 6 June 2026, nationwide, and remanded it to Treasury and the IRS. The court found the agency had "consistently applied the dual test framework" for over a decade before departing from it without adequate explanation.
A vacatur removes the instrument. What survives is the framework that preceded it — which means the narrowing of the continuity test falls away with the elimination of the 5% safe harbour, and both of the older standards stand again.
Two cautions carry over unchanged from the previous post. An appeal is expected, and a reversal could operate retroactively. So a project relying on continuous efforts today should be able to articulate its position under continuous construction as well, for the same reason it should document both BOC pathways: the law that will eventually judge the position is not yet fixed.
What Counts as an Excusable Disruption?
A defined but non-exclusive list of interruptions that do not break continuity. Current Federal Tax Developments sets out the items:
- Severe weather or natural disasters
- Permit or licence delays from government entities
- Government requests regarding safety or security
- Interconnection-related delays
- Custom component manufacturing delays
- Labour stoppages
- Specialised equipment unavailability
- Presence of endangered species
- Financing delays
- Supply shortages
Three features of that list are worth drawing out.
It is non-exclusive. The enumerated items are examples, not boundaries. A disruption of a similar character can qualify even if it is not named — which is more helpful than a closed list but harder to rely on in advance.
Interconnection is explicitly in it. Delays for new transmission lines and grid upgrades are named, which matters enormously given the queue dynamics covered later in this series. The single most common reason a US project stalls is expressly excusable.
Financing delays are in it too. That is less obvious than it looks, because financing delay is partly within the developer's control. Its inclusion reflects the reality that capital markets close to whole sectors at times, for reasons no individual sponsor caused.
What the list does not do is stop the four-year clock. An excusable disruption preserves continuity under the facts-and-circumstances test; it does not extend the safe harbour deadline. A project relying on excusable disruptions has already left the safe harbour and is arguing the harder case.
Is the Four-Year Safe Harbour Even Available to You?
Only if construction began early enough. The OBBBA introduced a second, harder deadline that overrides the continuity clock entirely for later starters — and for those projects the four calendar years discussed above simply do not apply.
The structure has two branches, and which one a project falls into is decided by a single date.
| Construction begins before 5 July 2026 | Construction begins on or after 5 July 2026 | |
|---|---|---|
| Placed-in-service requirement | Continuity safe harbour: four calendar years | Before 2028 |
| Effective window | Up to ~5 years | ~18 months or less |
| What governs | Continuity requirement | Statutory termination date |
Grant Thornton states the rule plainly: wind and solar facilities "must be placed in service before 2028 unless construction begins before July 5, 2026." Projects that start earlier are not subject to the accelerated timeline; projects that start later are, regardless of what the continuity safe harbour would otherwise have permitted.
Sit with the size of that cliff. A project establishing beginning of construction on 3 July 2026 falls into the first column and has until the end of 2030 — four calendar years after 2026. A project establishing it three days later falls into the second and must be generating before 2028. The same project, the same programme, a difference of three days, and roughly three years of permitted build time.
This is the single most consequential date in the current US renewables timetable, and it reframes everything in this post. For anything beginning construction after 4 July 2026, the continuity requirement is close to academic — the statutory deadline binds long before four calendar years elapse, and the practical question becomes whether the project can physically be built and energised inside eighteen months, including interconnection.
Three consequences for how a development programme should be run.
The BOC deadline is not a tax formality, it is a schedule constraint. Missing it does not reduce the credit; it compresses the entire build programme by roughly three years. That should appear in the construction schedule discussion, not only in the tax memo.
The safe harbour equipment order becomes a scheduling instrument. Establishing beginning of construction before 4 July 2026 — by physical work or, subject to the appeal discussed above, by the 5% safe harbour — is what buys the four-year window. The cost of that order is small relative to the value of the window it purchases.
Portfolio sequencing should be tested against the cliff, not the average. A five-phase project where three phases establish BOC in June 2026 and two in August 2026 is two different tax projects with two different deadlines, and the second pair may not be buildable at all.
ℹ️ Note: Model the beginning-of-construction date as a hard branch rather than a continuous variable. Almost every other date in a development schedule can slip by weeks with proportionate consequences. This one has a step function in it, and a model that interpolates across 4 July 2026 will report an answer that cannot occur.
How Does Disaggregation Rescue a Portfolio?
By letting the phases that finished on time keep the safe harbour while the stragglers argue continuity separately.
As covered in the previous post, multiple facilities operated as a single project may be aggregated for the purpose of establishing beginning of construction. The corollary matters here: multiple energy properties treated as a single project may be disaggregated and treated as separate properties for continuity safe harbour purposes.
The mechanic is that disaggregated properties placed in service before the continuity safe harbour deadline are eligible for the safe harbour, while the remaining properties may satisfy the continuity requirement under a facts-and-circumstances determination.
That is a genuinely useful piece of flexibility, and it changes how a portfolio should be modelled. Without disaggregation a five-phase project is a single pass/fail against one date — one late phase contaminates all five. With it, each phase is tested on its own placed-in-service date, and the exposure is limited to the phases that actually slipped.
The planning consequence is to sequence deliberately. If four of five phases can be placed in service inside the window and the fifth cannot, the fifth should be the one whose continuity story is strongest on the facts — the phase held up by an excusable interconnection delay rather than the one held up by a financing decision.
How Do You Model the Continuity Clock in Excel?
Three things: the deadline from the BOC date, the slack against the build programme, and the portfolio split at the deadline.
The inputs
Assumptions, labelled as such:
Beginning of construction 2025-08-20
Planned placed-in-service 2029-09-30
Phases in the project 5
Phase capacity (each) 50 MW
Phase 5 planned PIS 2030-04-15
The deadline and the slack
BOC_Year = YEAR(2025-08-20) = 2025
Deadline = DATE(BOC_Year + 4, 12, 31) = 2029-12-31
Slack = Deadline − Planned_PIS
= 2029-12-31 − 2029-09-30 = 92 days
Elapsed_Window = Deadline − BOC = 1,594 days
Ninety-two days of slack on a four-and-a-third-year programme is under six percent. State it that way rather than in days, because a board reads "6% schedule contingency against a credit-critical date" differently from "we have three months."
The calendar-year sensitivity
If BOC had been 2025-12-31 → Deadline 2029-12-31 → window 1,461 days
If BOC had been 2026-01-01 → Deadline 2030-12-31 → window 1,825 days
Difference from one day = 364 days
This is the cell that should exist in every development model and almost never does. Where a BOC date is being chosen rather than discovered — and with safe harbour equipment orders it frequently is — the calendar year it falls in is worth up to a year of build programme.
The portfolio test
Phases 1-4 PIS 2029-09-30 → before deadline → safe harbour → qualified
Phase 5 PIS 2030-04-15 → after deadline → facts and circumstances
Capacity_Safe = 4 × 50 = 200 MW
Capacity_At_Risk = 1 × 50 = 50 MW
Share_At_Risk = 50 / 250 = 20%
Without disaggregation the answer to this table is 250 MW at risk, not 50. The single line Disaggregate = TRUE is worth 200 MW of credit certainty, which makes it the most valuable assumption in the sheet and one that should be flagged rather than buried.
The sensitivity worth running
Run planned PIS date down the rows against BOC date across the columns, with days of slack in the cells and conditional formatting on negative values. The discontinuity at each 31 December is visible immediately, and it is the single most useful picture in a development schedule discussion.
ℹ️ Note: Model the excusable disruption argument as a separate flag with a short evidence note, not as an extension of the deadline. It preserves continuity under facts and circumstances; it does not move the safe harbour date, and conflating the two produces a model that reports comfort it does not have.
To run the full version — deadlines derived from each phase's own BOC date, disaggregation applied at the phase level, and slack tested against a probabilistic construction schedule rather than a single date — prompt Dezzmond with your development programme.
What Do Lenders and Tax Equity Actually Check?
- What is the continuity safe harbour deadline, computed from the BOC calendar year? Not four years from the BOC date.
- How much slack sits against the planned placed-in-service date? As a percentage of the window, not in days.
- Is the project relying on the safe harbour or on facts and circumstances? The second requires a full construction record and a much longer diligence conversation.
- If facts and circumstances, is the standard continuous efforts or continuous construction? The answer currently depends on an appeal.
- Are any excusable disruptions being relied on, and are they evidenced contemporaneously? Interconnection and permitting delays are named in the list; financing delays are too.
- Is the project disaggregated for continuity purposes, and is the analysis documented? It converts a portfolio-wide failure into a phase-level one.
- What happens to the model if the deadline is missed on one phase? Credit haircut, indemnity, or purchase price adjustment.
Frequently Asked Questions
How is the four-year continuity deadline calculated?
By calendar year. A facility satisfies the safe harbour if it is placed in service by the end of a calendar year no more than four calendar years after the year construction began — so construction beginning 20 August 2025 requires placed-in-service by 31 December 2029.
Does an excusable disruption extend the four-year deadline?
No. Excusable disruptions preserve continuity under the facts-and-circumstances test. They do not move the safe harbour date, so a project relying on them has already left the safe harbour.
What is the difference between continuous efforts and continuous construction?
Continuous efforts can be satisfied by activities such as entering binding equipment contracts and obtaining permits. Continuous construction requires continuing physical work of a significant nature. The second is materially harder for a project stalled on interconnection.
Which continuity test applies now?
Notice 2025-42 removed the continuous efforts test; the DC District Court vacated that notice in full on 6 June 2026, which removes the narrowing. An appeal is expected and could apply retroactively, so positions should be documented under both standards.
Can part of a project qualify if the rest is late?
Yes, through disaggregation. Properties treated as a single project may be disaggregated for continuity purposes, so those placed in service before the deadline keep the safe harbour while the remainder argue continuity on the facts.
Closing: A Clock That Counts in Years, Not Days
The continuity requirement is the half of beginning of construction that gets least attention and causes most of the damage, because it fails slowly. A BOC position is established on a single day and can be checked. Continuity is a condition maintained across four years, tested at the end, against a standard that changed twice during the period.
Three things are worth taking from it. The deadline is computed in calendar years, so where the date can be chosen the calendar year it falls in is worth up to eleven months of build programme. Excusable disruptions preserve the argument but never move the date. And disaggregation converts a single portfolio-wide test into a series of phase-level ones, which is usually the difference between a partial problem and a total one.
The next post takes the evidentiary side of both: what lenders and tax equity actually ask for when they diligence a beginning-of-construction position, and the documentation sponsors wish they had kept.
Sources: Current Federal Tax Developments — Technical Review of Beginning of Construction Requirements in Notice 2025-42 · K&L Gates — What Renewable Developers Need to Know on Beginning of Construction Rules · Troutman Pepper Locke — District Court Vacates IRS Notice 2025-42 · Grant Thornton — Energy Incentives Under OBBBA