No Meat Factory USA Inc. told Washington State it would close its Stanwood plant on 16 September 2026, affecting approximately 123 employees. A week later it filed again. The amended notice, received by the state on 28 July, said operations had closed on 27 July — the day before the notice arrived — and put the number affected at approximately 113.

The statutory notice period is 60 days. What the workers received was a notice dated after the plant had already stopped.

That is the whole story of the Stanwood closure in two filings, and it is more instructive than the closure itself. The figure that travelled — 123 workers, 16 September — comes from the superseded notice. It is still the version in general circulation more than five weeks later.

What the two notices say

Both The Daily Herald, which reviewed the filings in the state’s WARN database, and TheStreet, which reviewed the notices separately, describe the same sequence. Their accounts agree on every point where they overlap.

First notice Amended notice
Received by Washington State Week of 20 July 2026 28 July 2026
Stated closure date 16 September 2026 27 July 2026
Employees affected ~123 ~113
Largest single classification 61 production workers 55 production workers
Stated reason Not given as an exception “unforeseen business circumstances that led to immediate financial distress”

The two closure dates are 51 days apart. The amended notice was received one day after the closure it announced.

Other details are consistent across both filings. The site is 10120 269th Place NW, Stanwood. Night sanitation technicians were the second-largest classification in the first notice, at 14. The affected classifications include plant management, maintenance, warehouse, safety, human resources and supply chain — and the chief executive’s own position. The employees are not represented by a union. Both notices say workers will receive information on benefits, final pay, continuation of health coverage, unemployment insurance and workforce transition resources. Neither says whether production moves elsewhere.

No Meat Factory did not respond to The Daily Herald’s request for comment on 23 July.

Practical consequence: if you are tracking closures in this sector from press coverage, the first-filed number is the one that propagates and the amended number is the one that is true. Check whether a WARN entry has been superseded before using its headcount in any tally.

What the 60-day rule actually requires

The relevant regulation is 20 CFR 639.9. It is worth reading precisely, because the intuitive reading — that a closure without 60 days’ notice is automatically a violation — is not what it says.

The rule sets out three conditions under which the notification period may be reduced: the “faltering company” exception, “unforeseeable business circumstances”, and natural disaster. On the mechanics of late notice it is explicit:

If one of the exceptions is applicable, the employer must give as much notice as is practicable to the union, non-represented employees, the State dislocated worker unit, and the unit of local government and this may, in some circumstances, be notice after the fact.

So notice after the fact is contemplated by the regulation itself. The amended filing is not, on its face, irregular. The rule also requires that the employer, at the time notice is actually given, provide a brief statement of the reason for reducing the notice period — and the amended notice does exactly that, in language that tracks the regulation’s own vocabulary.

Two features of the rule are worth holding onto. First, the employer bears the burden of proof that the conditions for an exception have been met. Second, the unforeseeable-business-circumstances test is specific about what qualifies:

An important indicator of a business circumstance that is not reasonably foreseeable is that the circumstance is caused by some sudden, dramatic, and unexpected action or condition outside the employer’s control.

The regulation offers examples: a principal client’s sudden termination of a major contract, a strike at a major supplier, an unanticipated and dramatic major economic downturn, a government-ordered closing. It then adds a qualification that cuts the other way — the employer “is not required, however, to accurately predict general economic conditions that also may affect demand for its products or services.”

That qualification matters here, and it is why this article does not assert that any rule was broken. A multi-year contraction in plant-based meat demand is a general economic condition, and the regulation expressly declines to require employers to forecast those accurately. A sudden financing failure, by contrast, could sit squarely inside the exception. The notice says “immediate financial distress” without saying what caused it, and no filing in the public record resolves which it was.

Practical consequence: for a co-manufacturer, the “principal client’s sudden and unexpected termination of a major contract” example is the live one. A contract manufacturer’s revenue is concentrated in a handful of brand-owner accounts, and losing one can be genuinely sudden in a way that a demand curve is not.

Four disclosures, one company

The clearer way to read Stanwood is as a sequence of the company’s own stated numbers, held against each other.

Date Source What was stated
12 Jan 2023 TechCrunch $42m Series B; $60m raised to date; new 200,000 sq ft US plant; online “toward the end of 2023”; 40 employees then
15 Mar 2023 WA Dept of Commerce Investment of “up to $20 million”; plant to “eventually employ up to 150 workers”; $200,000 state grant
~20 Jul 2026 WARN notice 1 Closure 16 September 2026; ~123 employees
28 Jul 2026 WARN notice 2 Operations closed 27 July 2026; ~113 employees

The plant was announced as a 200,000 sq ft facility that would eventually employ up to 150. The highest headcount that ever appears in a public document is the 123 in the closing notice. On the state’s own framing, the site reached roughly four-fifths of the employment it was announced for, then closed a little over two years after production began.

The capital context sharpens this. The Series B was $42m, announced in January 2023 with Tengelmann Growth Partners leading and existing investor Emil Capital Partners participating; TechCrunch reported total capital raised at $60m at that point. The Stanwood plant was the stated use of proceeds — the round was explicitly to “build a bigger manufacturing facility in the U.S.” The asset that money went into ran for about two years.

Washington State contributed a $200,000 economic development grant, made to Economic Alliance Snohomish County rather than to the company, to support site engineering, development costs and initial upgrades. It was not structured as a per-job subsidy and should not be read as one. Economic Alliance’s chief executive Garry Clark framed the project at the time as “the return of over 100 food processing jobs to our community.”

Why the co-manufacturing layer breaks first

A contract manufacturer does not sell to consumers. It sells capacity to brand owners, and its volumes are a derivative of theirs. That makes the demand data the relevant explanation, and one figure in it does more work than the headline decline.

The Good Food Institute’s 2025 US retail analysis, built on data commissioned from SPINS, puts plant-based meat and seafood at $1.0 billion, with dollar sales down 10% and unit sales down 11% year on year. Over the longer window the compounding is heavier: between 2023 and 2025, dollar sales fell 17% and unit sales fell 21%, to 173 million units.

The more diagnostic number is household count. GFI reports that 11% of US households purchased in the category in 2025, down from a high of 20% in 2021, while repeat purchase rates stayed relatively flat. Its own reading is that “sales declines were driven by households leaving the category” rather than by loyal buyers buying less. The average buying household purchased about once a month.

That distinction is the one that decides a co-manufacturer’s fate. A category losing frequency among a stable buyer base loses volume gradually and predictably, and a manufacturer can size down against it. A category losing roughly half its buyer base over four years while the remainder buys at an unchanged rate loses volume in steps, as individual brand-owner accounts stop ordering entirely. Steps are what strand a 200,000 sq ft asset — and they are also what make a co-manufacturer’s revenue loss look sudden from the inside even when the category trend was visible for years.

The share figures show how little room there was to absorb this. Plant-based meat and seafood accounted for 1.4% of total retail packaged meat dollar sales in 2025, or approximately 0.7% of the total meat category once random-weight meat is included — down 0.24 and 0.12 percentage points respectively against 2024.

A note on the dataset

GFI’s figures rest on SPINS data using “plant-based positioned and/or labeled vegan attribution”, with plant-based private-label products additionally included; GFI states this may not align with standard SPINS categories. For the share-of-total-meat calculation, SPINS does not report non-UPC animal-based meat counter sales, so GFI develops its own dollar and unit assumptions for that segment and adds them to the reported UPC sales. The 0.7% figure therefore contains a modelled denominator and should be treated as an estimate rather than a measurement. The 11% household penetration figure does not depend on that adjustment.

Practical consequence: when underwriting co-manufacturing capacity in a declining category, household penetration is the leading indicator and dollar sales are the lagging one. Penetration fell from 20% to 11% while repeat rates held — that pattern was legible before the volumes were.

The asset

The Stanwood site was not built for this. It is a 200,000 sq ft plant with 55,000 sq ft of freezer capacity, pasteurising and co-extrusion capability, and lines for sausage and deli alternatives — a specification the Commerce announcement described as covering veggie burgers and nuggets, a veggie-based ground beef substitute, seafood alternatives and extruded products including sausages and wieners.

It has now been idled twice in under a decade. Twin City Foods, a frozen vegetable processor, closed the site in 2017 before No Meat Factory reopened it. Green Queen reports the company acquired it for $19.4m; on 200,000 sq ft that is roughly $97 per square foot before the renovation spend.

With Stanwood shut, the company’s British Columbia site becomes its only facility. Whether that plant absorbs the US volume is not addressed in either notice.

This is the third distinct pattern of stranded alternative-protein capacity we have tracked. It differs from the others in a way that matters: Believer Meats’ plant went to auction and drew no qualifying bids, and the cultivated meat supply layer failed upstream of any commercial product. Stanwood is a conventional food plant with conventional buyers — freezer space, extrusion and pasteurising have uses far outside plant-based meat. Its redeployment odds are structurally better than a purpose-built bioreactor hall’s, which is precisely why the price it eventually fetches will be a cleaner read on the sector than any of the distressed fermentation assets have been.

The counter-argument

The strongest case against reading much into the two notices is that this is ordinary corporate housekeeping. Companies amend WARN filings routinely; the regulation anticipates late and even after-the-fact notice; the reason given tracks the statutory language; and the count moved by ten, which is within the range of ordinary attrition between a notice and a closure — some workers leave once a closure is announced, which mechanically reduces the number affected.

On that reading the only genuine problem is a reporting one: outlets published the first notice and did not revisit it. The Daily Herald did update, appending a line that its article “has been updated to reflect new information filed in the amended notice.” Others did not.

That case is reasonable and this piece does not dismiss it. The reason the sequence is still worth recording is that the two filings are about a week apart and the closure dates are 51 days apart. A company that on roughly 20 July believed it could operate until 16 September, and on 27 July had stopped, either learned something in that week or had already known it. The regulation’s burden of proof sits with the employer on exactly that question, and nothing in the public record answers it.

What we could not establish

  • The individual WARN documents. Washington’s WARN database is the primary repository, but the specific notices are not retrievable as documents from its public landing page. Every figure attributed to the filings here comes from two outlets that each state they reviewed them, and which agree with one another.
  • The exact filing date of the first notice. The Daily Herald reported it on 23 July and says the amendment came “a week later, on July 28”; Green Queen described the first notice as filed “last week” on 27 July. We have written “week of 20 July” rather than assert a date, and “about a week apart” rather than compute an interval from an inferred one.
  • Which notice the 55-worker figure belongs to. TheStreet gives 55 production workers as the largest affected classification in an article that states the ~113 total, but does not explicitly say the figure is drawn from the amended notice rather than the first. We have placed it against the amended notice on that basis; the 61 figure is explicitly tied to the first notice by Green Queen, which used the 123 total.
  • What the “immediate financial distress” was. The notice states the conclusion, not the cause. Whether it was an account loss, a financing failure or something else is not in the public record, and we have not assumed.
  • Whether the exception was properly invoked. That is a legal determination requiring facts we do not have, and we make no assertion either way.
  • The British Columbia footprint. TechCrunch described a 30,000 sq ft BC facility in January 2023; Commerce referenced a Surrey, BC plant opened in 2022 plus a Surrey innovation centre; Green Queen describes a 40,000 sq ft co-manufacturing plant in Coldstream, BC in July 2026. These may be different sites or a restated figure. We could not reconcile them.
  • The Stanwood purchase price. The $19.4m figure appears in Green Queen’s account only. Neither Commerce nor TechCrunch states a purchase price, and we could not corroborate it.
  • Actual capital deployed. Commerce announced investment of “up to $20 million”; The Daily Herald describes $20 million as invested. “Up to” and “invested” are not the same claim and we have not treated them as one.
  • Twin City Foods’ closure year. Commerce states 2017; Green Queen states 2018. We have used 2017, from the primary source.
  • Peak employment. No document states the highest headcount the site reached. The 123 in the first notice is the highest figure on record, not a confirmed peak.

What to watch

Four testable claims:

  1. Whether the Stanwood site sells, and at what price per square foot. Against a reported $19.4m acquisition — roughly $97/sq ft — the resale price is the cleanest available read on whether food-grade extrusion and freezer capacity retains value when its original category contracts. This asset should clear more readily than the fermentation and cultivated assets we have tracked, because its capability is generic.
  2. Whether any WARN-related claim is filed. The employer bears the burden of proof on the exception. A claim would test it; the absence of one over the coming months is itself informative, particularly given the workforce was non-union.
  3. Whether the British Columbia plant absorbs US volume. Neither notice addresses it. If the company’s US brand-owner accounts continue to be served from Coldstream, this was a footprint consolidation. If they are not, it was an exit.
  4. Whether GFI’s next release shows household penetration stabilising near 11%. Penetration falling from 20% in 2021 while repeat rates held is the single number that best explains why co-manufacturing capacity stranded ahead of the headline sales decline. If it flattens, the remaining capacity is roughly correctly sized. If it keeps falling, more of it is not.

We have argued before that the counts of alternative-protein consolidation diverge depending on what each tracker includes. Stanwood is a small illustration of the same problem one level down: a single closure produced two official employee counts and two official dates within about a week, and the superseded pair is the one that entered general circulation.