Oatly: The $202m Cost of Building Capacity Too Early

In 2022 Oatly spent $206m on capital expenditure, most of it on new factories. By the end of 2023 it had cancelled two of them, in Peterborough in the UK and Dallas-Fort Worth in Texas, and recorded $172.6m of non-cash impairments and $29.0m of exit costs against them. Roughly $202m, written off on buildings that never made a litre of oat milk (Oatly 2023 Annual Report).

The easy read is that Oatly got it wrong. The useful read is that it got it wrong in a way every scaling brand is tempted to, and the way it recovered is the lesson.

Why the factories made sense at the time

In 2021 Oatly was the fastest-growing brand in its category and could not keep up. Its own filings say that "during 2021 and 2022, our failure to meet growing demand resulted in a failure to deliver products on a timely basis" (Oatly 20-F, FY2023). Orders were coming in that the company could not fill. Capacity, not demand, was the constraint on growth.

When that is your problem, building capacity is the obvious answer. Every unit you cannot make is a sale you do not get, and a factory of your own promises to close that gap while cutting the per-unit cost of production at the same time. So the company committed: $206m of capex in 2022, on top of heavy spending in 2021, for a network of large, owned plants on three continents.

The logic was sound. It just assumed the demand curve of 2021 would keep its shape.

What a factory actually commits you to

A factory is the most extreme version of a decision every product business makes. You spend cash now, on buildings, equipment and the people to run them, in exchange for lower unit costs later. The return depends entirely on volume. Run the plant near capacity and the economics are excellent. Run it half empty and you are paying the full fixed cost to produce half the output, and your unit cost goes up, not down.

That is what makes capacity different from most growth spending. Marketing can be turned off next month. A factory, a long lease, a production commitment or a container of stock cannot. The cash is committed before the sales arrive, and it stays committed whether they arrive or not.

By 2023 growth had slowed, the category had cooled, and Oatly had plants under construction that the volume no longer justified. It took the write-down and changed course.

The pivot: use what you have first

In 2023 Oatly moved to what it calls an asset-light model. Its filings put it plainly: "rather than continuing to build new manufacturing and production facilities, we will partner with co-manufacturing partners" (Oatly 20-F, FY2023). Existing plants and third-party manufacturers first. New sites only when the existing network is full.

The decision that best captures the new logic came in December 2024, when Oatly announced the closure of its Singapore facility. The reasoning in the announcement is the whole operational lesson in one paragraph: growth in Asia-Pacific would be served from the existing European plants, which would raise utilisation at those factories and lower future costs and capital needs (Oatly 6-K, 18 December 2024). The closure itself cost money, a $20m to $25m impairment and $25m to $30m of cash exit costs through 2027. The company judged that cheaper than carrying another under-used site.

Before adding capacity, understand what the capacity you already have can deliver. Oatly learned that the expensive way, then applied it.

The numbers

Sources: Oatly 2023 Annual Report; Oatly 20-F FY2023 and FY2024; Oatly 6-K, 18 December 2024; Oatly Q2 2026 results. 2024 capex shown as guided in the 2023 report.

Capex went from $206m to a guided $20m to $30m in four years. That is the shape of a company that stopped buying capacity ahead of demand and started filling the capacity it had.

Where it is now

Oatly's second quarter of 2026 is the clearest evidence the model is working. Revenue was $240.1m, up 15.2%. Gross margin reached 33.9%, up 143 basis points from 32.5% a year earlier, which the company attributes to supply chain efficiencies and a better channel, market and product mix. Adjusted EBITDA turned positive at $0.4m, against a $3.6m loss a year before. Full-year revenue guidance was raised (Oatly Q2 2026 results).

Capex in the quarter was $4.2m. Sold volume was 156.1 million litres against 153.6 million litres produced, which is a network running close to balance rather than one with empty plants waiting for growth.

Two honest cautions. The net loss for the quarter was still $31.3m, narrowed from $55.9m but a loss all the same, and the company has lost money every year since listing. And the gross margin gain was partly offset by higher energy and logistics costs from the Middle East conflict, which is a reminder that supply chain efficiency has to be re-won every quarter. Oatly has fixed its capacity problem. It has not yet fixed its profit problem.

What this means if you run a scaling brand

You will never build a factory in Peterborough. But you will face the same decision, scaled down, more than once a year, and it looks like this: a bigger warehouse because the current one is tight; a production commitment to a supplier because the MOQ brings the unit cost down; six months of inventory bought in one order because the price was right. Each one is a capacity commitment. Each one looks sensible inside a growth forecast. Each one commits cash before the sales arrive.

Model the slower-growth case before you sign. Oatly's plan worked if 2021 demand continued. It did not. Before any capacity commitment, run the numbers at half the growth rate you are planning for. How much volume do you need to cover the fixed cost? How much cash is tied up if the volume takes a year longer than you think? If the slow case still works, sign. If it only works in the fast case, you are not buying capacity, you are betting on a forecast.

Know your utilisation before you add a site. The Singapore decision rested on one number: how much headroom the European plants had. Most scaling brands cannot say what their current warehouse, their current supplier or their current 3PL could deliver if pushed, so they add a second one before the first is full. Our note on weeks of cover is the inventory version of this; the same question applies to every fixed asset you pay for.

Prefer commitments you can unwind. Co-manufacturers, 3PLs and shorter supplier commitments cost more per unit than owned capacity at full volume. They cost far less when volume disappoints, because you can step down instead of writing off. Oatly's asset-light model is a deliberate trade: a slightly higher unit cost in exchange for the ability to change course. At your scale, that trade is almost always worth making until the volume is proven, and we wrote about the own-versus-operate version of it in the PUMA and Maersk piece.

Write-downs are the price of certainty bought too early. $202m sounds like a disaster. It was the cost of exiting a commitment that no longer fitted, and taking it allowed the company to run a tighter network that now shows up in the margin. The founder's equivalent is the warehouse lease you break, the supplier commitment you negotiate down, or the inventory you clear at a loss. The mistake is not taking the hit. The mistake is carrying the commitment for another two years because the write-down feels like admitting it was wrong.

Capacity should follow proven demand, not forecast demand. That is the whole lesson, and Oatly's $206m to $25m capex line is what it looks like when a company finally applies it.

If you are weighing a bigger warehouse, a supplier commitment or a large inventory buy and want the slower-growth case modelled before you sign, see how we work.