Inventory rose while sales fell at 66 US companies

Inventory is the line almost nobody reads. It is not in the earnings headline, it is not what analysts ask about on the call, and it sits on the balance sheet rather than the income statement. It is also the only line that records, in dollars, the difference between what a company decided to make and what its customers decided to buy.

We asked the SEC’s data a narrow question: which companies finished fiscal 2025 selling less than the year before and holding more stock than the year before?

Sixty-six of them, with revenue above $1bn.

Why the second condition matters

Screens for “inventory growing faster than revenue” are easy to run and nearly useless. We ran that one first. It returned 532 companies out of the 1,323 we could test — forty percent of the market.

That is not a signal. That is what growth looks like. A company selling more next year has to build more this year, and the build shows up before the sale does.

Requiring revenue to fall removes the explanation. If sales went backwards, the extra stock cannot be inventory for growth that arrived. Something else happened.

Comparable filers                     1,323
inventory outgrew revenue             532   ← 40%, mostly growth
revenue above $1bn                  256
excl. utilities, unlisted, misfiled 244
revenue actually fell            66

What the 66 look like together

Combined inventory rose $7.1bn. Combined revenue fell $50.2bn. The median company added 8 days of stock. The largest single increase was 93 days.

The list, by days of stock added

Days of inventory is inventory divided by the year’s cost of goods sold, times 365 — how many days of production cost is standing still. The number only means something against the same company’s prior year. A drugmaker and an appliance maker have cost structures that are not comparable, which is why the column below shows movement rather than level.

Twenty of 66, ordered by days added. Dashes mark companies that report inventory as a single line without a components breakdown.

The line that separates them

Two companies can both show inventory rising while sales fall and mean opposite things.

Alamo Group’s inventory is 92% finished goods. Farm machinery that was built to be sold, and was not sold. Whatever the reason, the pile is product.

Illumina’s inventory is 8% finished goods. The other 92% is raw material and work in process — reagents and components that have not become a sellable kit yet. A build of that shape is a production decision, not an unsold shelf.

Of the 66, 42 tag the split. Their median finished-goods share is 53%. Sixteen are above 60%, and eight are below 30%.

The 24 that do not break it out are not hiding anything — a single-line presentation is permitted, and retailers in particular have little to break out. But for those 24, this article cannot tell you which kind of pile it is, and neither can anyone else working from the structured data.

A tagged split is also only as good as the costing behind it. Days after this screen ran, Solésence told the SEC that its own allocation of labour and overhead across raw materials, work in process and finished goods had not complied with ASC 330, leaving inventories overstated — one of 17 companies that said this summer their published numbers should no longer be relied on.

Same industry, opposite direction

Ten semiconductor companies cleared the first screen. Sorting them by revenue growth shows how little the inventory line means on its own.

Every one of them built inventory faster than revenue. At the top of the table that is a company adding 12 days of stock while sales grow two-thirds — NVIDIA finished the year with $21.4bn of inventory against $62.5bn of cost of goods sold, which is the same build-ahead visible across its supply agreements. At the bottom it is a company adding 20 days while sales shrink by a ninth.

The inventory column is identical in direction and opposite in meaning. Reading it without the revenue column next to it produces the wrong answer nine times out of ten.

What this costs the companies

Inventory is cash a company has already spent and has not got back. That is the same mechanism behind the gap between reported profit and cash generated — earnings can rise while cash falls precisely because working capital absorbs it.

For the 66, the $7.1bn added is money sitting in warehouses instead of in the business. It is not a loss. It becomes one only if the goods are eventually sold below cost, and nothing in this screen says they will be.

What would make this wrong

• One year is not a trend. Every company here is compared against a single prior year. A business recovering from a stock-out in fiscal 2024 will appear to be piling up in fiscal 2025 when it is only returning to normal.
• Days of inventory is not comparable across industries. Alkermes shows 365 days largely because a drugmaker’s cost of goods sold is small relative to revenue. The level says little; the change says more.
• The finished-goods share is unavailable for 24 of the 66. Where it is missing we have said so rather than estimating it.
• Fiscal calendars cost us companies. We required the inventory date to equal the fiscal year end, which dropped 321 filers whose data the frames API files at a calendar quarter instead. Brown-Forman and General Mills are among them. That is a deliberate trade — including them would have meant dividing a November balance by a fiscal year of costs ending the following May.
• Two companies were dropped for disagreeing with their own filings. The frames figure differed from the annual report by more than 1%, so they were excluded under the check we apply to every screen. Neither gap was large; both were removed anyway.
• We have not read 66 annual reports. The finished-goods split comes from XBRL tagging, not from management’s explanation of it. Where a company has told its shareholders why inventory rose, that explanation is in the 10-K and is not reflected here.

Check it yourself

The two frames links below return the raw figures for every filer. The screen is a script in this site’s repository, screen-inventory-outpacing-revenue.mjs, and the finished-goods enrichment is a second one beside it. Both write their cache to disk so a re-run reproduces the same table.

Every company that survived was checked against its own annual report before appearing here. If a figure on this page does not match a filing, tell us and it will be corrected on the article and on the corrections log, with the date.