On 28 September 2026 Novonesis announced the completion of its West Allis, Wisconsin expansion with two numbers and no volumes: fermentation capacity up “approximately 66%,” packaging capacity up “17%.” The release gives no litres, no tonnes, no starting point and no project cost. For a site the company simultaneously describes as “one of the world’s largest lactobacillus production facilities,” the public record now contains a superlative and two ratios to an undisclosed denominator.
That would be unremarkable — most ingredient companies do not publish plant volumes — except that the same project was described in some detail two years ago. In October 2024, BizTimes Milwaukee published a site visit with three Novonesis executives that carried dollar figures, a square-footage build sequence, a completion date and a capacity claim. The claim was that the expansion would double capacity and make West Allis the largest fermentation site of lactic bacteria in the world. It completed on schedule, and it completed at 66% and “one of.”
This is not a story about a company missing a target. It is a story about what a dairy manufacturer evaluating North American culture supply can and cannot compute from what has been published — and about where the only hard numbers for this project actually live, which is a local business weekly and a group interim report, not the completion release.
What the completion release discloses
The primary text, distributed through GlobeNewswire on Monday 28 September and reproduced in full by DairyBusiness three weeks later, is short. The quantitative content is one sentence:
The investment increases the company’s North American production capacity for dairy cultures, boosting fermentation capacity by approximately 66% and packaging capacity by 17%, while establishing West Allis as one of the world’s largest lactobacillus production facilities.
Note what that sentence leaves open. It is grammatically ambiguous as to whether the 66% measures the company’s North American dairy-culture capacity or the site’s fermentation capacity, and the trade press split on exactly that point. Food Ingredients First read it as the company level, reporting that Novonesis “has scaled its North American dairy culture fermentation capacity by 66%”. Construction Review Online read it as the site, writing that the expansion “increased the facility’s fermentation capacity by approximately 66%”. Both readings are defensible from the sentence. They are not the same claim, and they do not have the same base.
| Disclosed in the completion release | Figure | Base stated? |
|---|---|---|
| Fermentation capacity increase | Approximately 66% | No |
| Packaging capacity increase | 17% | No |
| Whether 66% is the site’s or North America’s | Not resolved by the text | Not applicable |
| Post-expansion fermentation volume | Not published | No |
| Pre-expansion fermentation volume | Not published | No |
| Project capital cost | Not published — Food Ingredients First: “Financial details of the investment have not been disclosed” | No |
| Site standing | “One of the world’s largest lactobacillus production facilities” | Company claim, no ranking basis |
| Site opened | 1929 | Yes |
| Site headcount | 360 | Yes |
| Ribbon-cutting | Friday 25 September 2026, ribbon cut by COO Anders Lund | Yes |
| Route to further capacity | “Leveraging existing infrastructure along with automation and robotics” | Yes, qualitatively |
| Sustainability features | Energy-efficient equipment, water reuse, waste heat recovery | Yes, qualitatively |
Practical consequence: a release that gives a percentage and a superlative but no volume cannot be used in a supply assessment. The question to put to the account team is not “how much did capacity grow” — that has been answered and is useless — but “what is the annual output of this site in tonnes of finished culture, how much of it is contractually committed, and what is the lead time on an incremental tonne.”
What the 2024 record said the project would do
BizTimes Milwaukee’s 6 October 2024 piece, reported from inside the plant with Theis Bacher (then vice president of food and beverage, North America), Keith Grover (head of the West Allis and Madison plants) and Julien Biolley (director of marketing and business development, North America), is the most detailed public account of the build. It sets out a three-phase sequence with costs attached:
| Phase | Year | Scope | Cost as reported |
|---|---|---|---|
| Phase one | Completed 2021 | 20,000 sq ft on the west of the campus; added fermentation capacity and warehouse space | $25 million |
| Phase two | Begun 2023 | Two-storey 81,000 sq ft addition east of the existing plant, including a 27,000 sq ft freezer at −67°F and space for future labs | $35 million |
| Phase three | Plans submitted May 2024 | Two-storey 50,000 sq ft building to house new fermentation equipment | Not separately stated |
| Five-year campus total, including phase three | As at October 2024 | All phases combined, not a single building | “$165 million will have been invested in the West Allis facility within the past five years” |
| Stated forward pipeline beyond that total | Unspecified timing | Site-wide, scope not broken out | Bacher: “There’s another roughly $100 million flowing to just this site alone” |
Bacher also told BizTimes: “We are investing in Milwaukee more than in any other of our global sites because of the people here.” On the capacity outcome, the article states:
The latest expansion to Novonesis’ fermentation facility will be finished by fall 2026. It will double the company’s production capacity and become the largest fermentation site of lactic bacteria in the world.
Two things about that sentence. First, it is the publication’s reporting from the visit, not a quotation marked as Novonesis’s own words — so it should be read as how the project was described to a reporter in 2024, not as a company commitment on a wire. Second, “double the company’s production capacity” is itself ambiguous in the same way the 2026 release is: the company’s capacity in what, of what, measured where. We cannot resolve it and do not.
The timing, at least, held. “Finished by fall 2026” and a ribbon-cutting on 25 September 2026 is a project delivered on the schedule it was given two years earlier — which, in a sector where announced capacity routinely arrives late or not at all, is the most impressive fact in this story and the one the release itself does not make.
Reading the gap between “double” and 66%
Three readings fit the evidence, and we cannot choose between them from public sources.
The denominators differ. If 2024’s “double” meant the site’s fermentation capacity and 2026’s 66% means the company’s North American dairy-culture capacity, the two numbers are not in conflict at all — they are ratios on different bases, and a site doubling could easily show up as a two-thirds lift at regional level. The release’s own wording points this way. This is the reading we consider most likely, and it is also the reading that makes the 2026 number the less flattering of the two to publish.
The scope was trimmed. A 50,000 sq ft fermentation building specified in May 2024 and commissioned in September 2026 passed through two years of cost estimation. Tanks are the line item most often reduced. Nothing in the record supports this reading; nothing rules it out either.
66% is simply the measured result. Pre-construction capacity claims are models. A doubling modelled on nameplate volume can land at two-thirds once real titres, changeover times, cleaning cycles and product mix are in the number. On this reading 66% is not a shortfall but a more honest figure than the one it replaced, and the same goes for “one of the world’s largest” against “the largest.”
Practical consequence: when a pre-build capacity claim and a completion claim disagree, do not assume failure. Ask which one is nameplate and which is demonstrated, and on what product mix. The gap between the two is usually the most useful number in the exchange, and it is almost never published.
Fermentation up 66%, packaging up 17%
The asymmetry inside the release is the part an operator should look at hardest, and none of the coverage we read addressed it.
Fermentation capacity rose by roughly four times as much as packaging capacity. If the two were balanced before the project — and we cannot confirm that they were — then after it, packaging covers about 70% of what fermentation can produce. The arithmetic is 1.17 divided by 1.66. This is our inference, not a disclosed figure, and it is conditional on a starting balance nobody has published. If packaging carried slack beforehand, the 17% may be exactly the top-up required and there is no constraint here at all.
But the structure of this particular plant makes the question non-trivial. Per the 2024 site visit, West Allis fermentation runs in six fermenters on cycles that “typically” take about eight hours, with some batches running several days. Output is concentrated, separated from the fermentation water and formed into frozen pellets using liquid nitrogen, then held at −67°F. The one element of phase two identified by size in that account was a 27,000 sq ft freezer at that temperature. In other words, the downstream of this site is a cryogenic pelletising and cold-storage train, not a drum-filling line — the kind of downstream where capacity is expensive, lumpy, and not obviously expandable by running the same equipment harder. That characterisation is ours, drawn from the 2024 description of the process; Novonesis has not described its downstream constraints.
The release’s own answer to “where does the next increment come from” is consistent with this. Further growth, it says, comes from “leveraging existing infrastructure along with automation and robotics.” That is a downstream and materials-handling answer. It is not a tank answer.
Practical consequence: ask which step is the binding constraint on an incremental order, and ask it about the pack format you actually buy. A site with headroom in fermentation and none in pelletising will quote long on a new SKU or a new pack size while insisting capacity is available — and both statements will be true.
The capital is visible at group level and nowhere else
Novonesis has not disclosed what West Allis cost. It has, however, disclosed a great deal about the spending programme the project sits inside, and the group numbers are specific.
From the H1 2026 interim report, published 19 August 2026 as Company Announcement no. 15:
| Metric | H1 2026 | H1 2025 |
|---|---|---|
| Net sales | EUR 2,235.5m | EUR 2,096.1m |
| Organic sales growth | 8% | 9% |
| Food & Beverages sales | EUR 748.0m | EUR 701.4m |
| Food & Beverages organic growth | 11% | 10% |
| Cash flow from net investments excl. acquisitions | EUR 215.7m | EUR 119.4m |
| Purchase of property, plant and equipment | EUR 177.6m | EUR 118.0m |
| Capex ratio (% of sales) | 9.7% | 6.6% |
| Capex ratio, Q2 | 11.0% | 8.1% |
| Property, plant and equipment, balance sheet | EUR 3,149.0m | EUR 2,834.1m |
The report attributes the step-up directly: “The higher investment level reflects planned capacity expansions to support accelerated growth.” For the full year it guides capex to “the 12-14% range relative to sales,” and says that range “relates to significant production-capacity expansion projects as also announced in relation to the 2030 GROW strategy launch in August 2025.”
So the capital intensity is real and rising — from 6.6% of sales a year ago to a guided 12-14% — and the demand side supports it. Food & Beverages grew 11% organically in H1, despite around four percentage points of drag from exiting certain countries, with the report citing “strong momentum in Dairy” driven by “increasing demand for high protein fresh dairy products” and “solid growth in Cheese.” That is precisely the demand West Allis serves.
What the group disclosure cannot tell you is how much of it landed in Wisconsin. And the comparison that makes the omission conspicuous is internal: earlier in the same month as the West Allis completion release, Novonesis put a number on a different plant. The Patalganga enzyme facility in Maharashtra was announced at EUR 600 million (about $699 million), to be fully operational in 2030. Same company, same month, same category of announcement — one carries a capital figure to nine figures of precision, the other carries two percentages.
The same interim report also approved an inaugural EUR 1 billion share buyback, to run from the second half of 2026 to the end of 2029, with net interest-bearing debt to EBITDA at 1.8x. The capex step-up and the buyback are being funded concurrently. We note this as a fact about capital allocation rather than a criticism; at 37.7% adjusted EBITDA margin and free cash flow before acquisitions at 14% of sales, both are evidently affordable.
Practical consequence: when a site-level cost is withheld, the group capex line and its guidance are the next-best instrument, and they are usually available for a listed supplier. A capex ratio nearly doubling year on year tells you a supplier is in a build cycle — which is useful for judging whether a quoted lead time is likely to improve or deteriorate.
Why a percentage with no base is a procurement problem
This publication has now documented the same pattern at several companies: a 4x capacity multiple with no starting volume, a headline fermenter volume that the company’s own website contradicted, and a tonnage claim that rested on a partner’s capacity rather than the supplier’s. Novonesis is a different case from all three — a listed incumbent with a 1929 plant, 360 people on site, more than 11,000 employees worldwide and nearly a century of dairy-culture production, not a scale-up hoping to be believed. The reputational stakes of publishing a volume are correspondingly lower, and the commercial stakes of publishing one are correspondingly higher: its competitors would find the number genuinely useful.
That is the honest reason for withholding it, and it is a good one. But the consequence falls on the buyer either way. A formulator deciding whether to single-source a culture from West Allis cannot calculate what share of the site’s output their volume represents, cannot tell whether a 66% increase has moved them from a priority customer to a rounding error or the reverse, and cannot compare this site against any other on a common basis — because the industry’s capacity figures are quoted on incompatible bases when they are quoted at all.
The counter-argument
Three points cut against the framing above.
Percentages are the right disclosure for a customer-facing release. This announcement is addressed to dairy manufacturers, not analysts. What a customer wants to know is whether supply has got easier, and “66% more fermentation” answers that directly without handing a competitor a production figure. Absolute volumes belong in a confidential capacity commitment, which is where a serious buyer would get them anyway.
The project delivered. Plans submitted May 2024, reported for completion by “fall 2026,” commissioned with a ribbon-cutting on 25 September 2026. Against a sector in which announced fermentation capacity frequently slips by years or is quietly abandoned — see Meati’s tanks sold for $75,000 and ADM’s conversion of underutilised capacity — on-time delivery of real steel is the headline, and a two-thirds lift on a site that has been fermenting commercially since 1929 is a substantial absolute addition even though we cannot size it.
“One of the world’s largest” may be the more defensible claim. There is no published register of lactic-acid-bacteria fermentation capacity by site against which “the largest” could be verified. A company that drops a superlative it cannot substantiate in favour of a hedge is behaving better, not worse, and should not be penalised for it.
What we could not establish
- The base for either percentage. No absolute fermentation or packaging volume for West Allis, before or after, appears in any source we read.
- Whether the 66% is the site’s or North America’s. The release sentence supports both readings, and two trade outlets took it differently. We have not resolved it.
- The project cost. Not disclosed. The $165 million five-year figure and the “roughly $100 million” forward pipeline are BizTimes’s 2024 reporting of the campus programme, not a cost for the phase that completed in September 2026, and the two should not be added or equated.
- Whether phase three as built matches the May 2024 filing. We did not obtain West Allis municipal permit records; the 50,000 sq ft figure is from the 2024 report of the submitted plans.
- Headcount comparability. The 2026 release states 360 people at the West Allis site. The 2024 article states 460 in metro Milwaukee — a figure covering multiple Wisconsin plants — and reported an estimate that it would grow 10% to 15% after the expansion. These two figures measure different populations. They are not evidence of a workforce reduction and should not be read as a comparison.
- The number of fermenters post-expansion. Six, per the 2024 site visit, pre-phase-three. No post-completion count has been published.
- Franklinton, North Carolina. Construction Review Online refers to Novonesis having “also expanded facilities in other markets, including Franklinton, North Carolina.” We did not verify that project and have not relied on it.
- The pre-project fermentation-to-packaging balance, on which our 70% coverage inference depends entirely.
What to watch
Four testable claims, so this piece can be revisited honestly.
- FY2026 capex lands in the 12-14% guided range. The annual report will show whether the build cycle ran to plan. H1 came in at 9.7%, implying a materially heavier second half.
- A volume figure eventually surfaces. Site-level capacity numbers at this company have historically reached the public through local business press and trade interviews rather than wires. That is where to look.
- Packaging is addressed next. If the 66%/17% split is a genuine constraint, the next West Allis announcement should concern automation, filling or cold storage rather than tanks. The release already signals “automation and robotics” as the route to further growth.
- Dairy momentum holds. Food & Beverages at 11% organic growth is what underwrites this capex. A deceleration in the Dairy sub-segment would make a 66% capacity addition look early rather than timely, and the quarterly reports will say which.