29 restructurings in 2026. Five said what they would save.

In one week of May, two companies told the SEC they were restructuring. Starbucks put the cost at about $400 million. Autoliv, four days earlier, put its own at $142 million.

Autoliv’s is the one that takes more cash out of the business.

That is not a trick of accounting. It is written into both filings, and it is the reason the headline number on a restructuring charge tells you less than it appears to.

What the form asks for, and what it doesn’t

When a US-listed company commits to closing a plant or cutting a workforce, it files Item 2.05 of Form 8-K. The item is unusually specific about what must be said: the date the board or management committed, an estimate of the total charge, an estimate by major cost type, and an estimate of the portion that will result in future cash expenditures.

Read across enough of them and the omission becomes as informative as the disclosure. The item asks what the plan will cost. It does not ask what the plan will return.

We screened the 600 largest US filers by revenue and read the structured item field on every 8-K each of them filed between 1 January and 26 August 2026. Twenty-nine filings carried Item 2.05, from 25 companies — Goodyear filed twice, Lamb Weston twice, Estée Lauder three times.

Total charge stated                    29 of 29
Dollar figure for the cash portion      9 of 29
Headcount or percentage of workforce   13 of 29
Expected annual saving                  5 of 29

Every one of them will tell you the bill. Five of twenty-nine tell you what it buys.

The same charge, two different things

The split between cash and non-cash is where the meaning sits. A charge made of severance is money leaving the building. A charge made of asset impairment and accelerated depreciation is a write-down of something the company already paid for years ago — painful in the income statement, invisible in the cash flow statement.

Which brings the opening back. Starbucks’ charge is 2.8 times the size of Autoliv’s, and about 70% of it is non-cash — impairment of long-lived assets, the filing says, “including right-of-use lease assets.” That leaves roughly $120 million of cash. Autoliv’s charge is 91% cash: $129 million of severance and retention for approximately 2,200 employees in Türkiye, against $13 million of write-offs.

The smaller number is the bigger cash cost. A screener sorting on restructuring charges would rank these the other way round.

The five that said what it was for

Only five filings put a number on the benefit. Because they also state the cost, they are the only five where a payback period can be worked out at all.

Kyndryl            $200m  ÷  $450m/yr from FY2028   =  0.4 years
Boston Scientific  $750m  ÷  $500m/yr gross         =  1.5 years
Goodyear (Jul)     $550m  ÷  $270m/yr from 2028     =  2.0 years
Goodyear (Mar)     $105m  ÷   $50m/yr after 2028    =  2.1 years
Autoliv            $142m  ÷   $40m/yr from 2027     =  3.6 years

Kyndryl is the outlier and the filing is blunt about it: approximately $200 million of charges, “primarily consisting of future cash expenditures for severance,” against “annualized run-rate operating expense savings of approximately $400 to $500 million” in fiscal 2028. Spend once, save twice as much every year.

Goodyear is the useful case because it filed twice with different shapes. The March plan in EMEA was pure cash — $100–110 million out, $100–110 million of cash outflows, no asset write-down — for a net reduction of 400 positions after creating 200 new ones. The July plan closing Fayetteville, North Carolina was the opposite: $535–565 million of which only $190–210 million is cash, the rest accelerated depreciation on a plant that will stop running, plus $40–50 million of pension special termination benefits. Both land at roughly a two-year payback, from completely different ledgers.

The other 24 filings give you the numerator and no denominator.

What the item lets you not say

Three of the 25 companies state a total and nothing else about its composition — Nike’s approximately $300 million, Synopsys’ raised estimate of $425–500 million, and Estée Lauder’s.

Estée Lauder is the instructive one. Its programme was committed to on 1 February 2024, and the 2026 filings still carry the original language: the company “was unable to make a determination of the estimated amount or range of amounts to be incurred by major cost type and future cash expenditures.” That sentence is not evasion — it is the escape hatch Item 2.05 explicitly provides, with a duty to amend once the numbers exist. Estée Lauder filed three times in 2026 under this programme, in April, June and July.

Under Armour shows what the fuller version looks like. Its plan is $305 million, split up to $139 million cash and up to $166 million non-cash — and, unusually, it says what has already landed: $261 million recognised at 31 March 2026, of which $109 million cash and $152 million non-cash. Almost every other filing here is an estimate of the future. This one is a progress report.

The headcount numbers, where they exist

Thirteen of the 29 quantify the people. The range is wide enough that the percentages matter more than the totals.

Block expects to reduce its workforce by more than 40%. Intuit’s plan cuts approximately 17% and considers site closures. Coinbase is cutting approximately 700 people, or 14%. Viatris up to approximately 10%. Autodesk approximately 7%, or 1,000 employees. Kenvue approximately 3.5% net; Workday approximately 2%; Insperity approximately 4% of non-sales positions.

In absolute terms Dow is the largest that gives a number — approximately 4,500 roles — followed by Autoliv’s 2,200 in Türkiye and Goodyear’s 1,750 in Fayetteville. FedEx’s French subsidiary may reduce up to 500 positions and change the working locations or schedules of up to 800 more, a distinction no other filing in the group draws.

What this is not

• Not every restructuring in 2026. The population is the 600 largest US filers by revenue. Under Armour’s plan sits outside it and is included above precisely to show the edge of the screen. A mid-cap that closed a plant this year will not appear.
• These are estimates, not outcomes. Item 2.05 is filed at commitment, before the money moves. Synopsys’ filing in this screen is an amendment raising an earlier estimate; Corteva’s raises its own by $100–115 million; Colgate’s raises a programme from $200–300 million to $350–550 million. Estimates move, and the amendments are where you see it.
• No aggregate is computed here. Some of these figures are single actions and some are cumulative programme totals spanning years. Adding them would produce a number that means nothing, so this article does not produce one.
• Payback is not return. The five ratios above divide a stated cost by a stated saving. They ignore the timing of both, tax, and — in Boston Scientific’s case explicitly — the portion of savings the company intends to reinvest rather than keep.
• Silence is not concealment. A company that does not forecast savings may simply not have a reliable estimate. The point is that the filing requirement does not force the question, so the answer arrives unevenly.

Check it yourself

Every figure here is in Item 2.05 of the filings linked in the sources, and most of these documents are two pages. The Starbucks non-cash figure, the Autoliv split, the Goodyear ranges and the Kyndryl savings are all quoted above in the filings’ own words.

The population came from reading the structured item field on each company’s EDGAR submissions file rather than searching filing text — the same method we used to trace six revolving credit facilities in August and the 364-day renewal calendar. When a single company writes assets down on this scale it deserves its own reading, as with Air Products’ $6.6 billion of hydrogen project exits.

No share prices appear in this article. Every calculation above is written out in this article’s front matter and re-checked when the site builds.

If a figure here does not match a filing, tell us and it will be corrected on the article and on the corrections log, with the date.