On October 26, Microsoft opened a voluntary exit offer to roughly 60,000 employees, about 27% of its global workforce. Human resources expects fewer than 5% to accept it, which works out to something near 3,000 people.
The offer carries a name inside the company, VEO, and a clear author: Satya Nadella, who has spent two years telling investors that the company’s cost base has to follow its workload. The option is being presented as choice. The arithmetic behind it is a filter.
The comparison that explains it is six months old. In April, Microsoft offered a retirement buyout to about 8,750 senior US employees, and more than 30% took it. That package carried extended medical coverage and cost the company a $900 million charge. The October offer carries the standard severance formula and nothing above it.
That gap is the whole story. Employees facing a voluntary exit run the same calculation any seller runs: what do I get now against what I give up. In April the answer included health insurance at a moment when the individual market is expensive and uncertain. In October the answer is a cheque that a standard layoff would have produced anyway.
The medical coverage detail is not incidental. For a US employee in their late fifties, bridging the years to Medicare is worth more than several months of salary, which is why the April number and the October number are not comparable offers at all.
A rational workforce responded rationally. Where the premium was real, three in ten left. Where it was absent, most people stayed, because staying costs nothing and the job market for a mid-career engineer is not obviously better than the one they already have.
That result is easy to read as a failed programme. It is better read as a designed one. A company that wanted 60,000 departures would have paid for them, and Microsoft knows exactly what a premium costs because it wrote the April cheque. What it appears to want is a self-selection mechanism that removes the people already planning to leave, at the price of an ordinary redundancy.
The second half of the design is on the hiring side. Nadella has said pay for systems engineers, infrastructure specialists and AI training researchers has risen noticeably, and that the forward deployed engineer, the role that sits inside a customer’s operations to make a model work, is the fastest growing job in the company.
Read those two statements together and the strategy stops being mysterious. Compensation is being reallocated rather than cut. Money freed in roles the workload no longer needs goes to the roles where the market is bidding up salaries, and the voluntary exit is the cheapest way to create that room without announcing a second round of layoffs inside a year.
The industry comparison supports the reading. Microsoft’s largest peers have spent three years trimming in waves, and each wave has produced the same sequence: announcement, severance, a quarter of goodwill in the press, then hiring in different functions eighteen months later. A voluntary offer at standard terms produces most of the headcount effect with less of the reputational cost.
Recurring buyouts have also become a fixture of large technology employers since 2023, and they serve a second purpose beyond cost. They reset the age and seniority mix of a division without a formal reduction in force, which is useful when the company expects to be hiring in the same unit a year later.
It does produce less certainty. A company cannot book a restructuring with the same confidence when it does not know how many people will walk, and analysts who model headcount off the announcement will be working with a range rather than a number. That uncertainty is part of the mechanism, not a flaw in its execution.
The cost side deserves attention too. If 3,000 people leave at standard severance, the charge is small against a company of Microsoft’s size, and the savings are smaller still. The real expense in this manoeuvre is wage inflation in the roles that matter, which shows up in operating margin over four to six quarters rather than in a single restructuring line.
There is a harder question about who stays. Voluntary programmes select on willingness to leave, not on performance, so the people most likely to accept are those with outside options, which is often the strongest engineers. Microsoft’s bet is that raises and the AI roles will hold the people it wants before the window closes.
Investors will get the first real read on that bet in the January earnings call, when headcount and operating expenses land in the same filing. A small decline in headcount with a rising cost per employee is the outcome management is describing. A large decline would mean the filter caught more than intended.
For the 57,000 people who stay, the message is plainer than any memo. The company is not shrinking. It is moving, and the offer document is a map of where it is going, priced at the value of a decision the company would prefer most employees decline.


