
On 29 October 2026, Microsoft began offering voluntary buyouts to roughly 60,000 employees across the United States, the United Kingdom and Canada — about a quarter of its global workforce and its largest restructuring in a decade. The move aligns with two years of public comments by CEO Satya Nadella that AI is changing how many people Microsoft needs and what kind of work they do. Liu Gong reads it less as cost cutting than as a structural shift of budget from payroll to compute, one whose
According to reporting by The Verge, Microsoft is offering voluntary buyouts to roughly 60,000 employees in the United States, the United Kingdom and Canada, with packages tied to tenure and job grade. The company disclosed the programme on 29 October 2026 but published neither an application deadline nor a target headcount.
The buyout follows a clear escalation. In May 2025 Microsoft cut about 6,000 roles, concentrated in product and management layers; in July 2025 it removed another 9,000 across sales, consulting and parts of engineering, together roughly 6% of the workforce at the time. The new programme reaches close to four times as many people as those two rounds combined.
Nadella has spent two years preparing the ground. He has said publicly that 20% to 30% of the code inside Microsoft is now AI-generated, and that the company intends to do more with less. Capital spending
Why now? Three preconditions converged in 2026.
First, the tooling matured. Copilot, GitHub Copilot Agent and internal code agents moved from demo to daily workflow, giving managers a quantitative read on per-role output for the first time instead of a gut feeling.
Second, the financials pushed back. AI infrastructure is heavy, depreciating over roughly three years, while payroll is a recurring expense. Once capex exceeds 30% of revenue, the finance side inevitably demands an offset at the other end, or server depreciation steadily erodes margins.
Third, the industry supplied cover. Since 2025 Salesforce has said it is slowing hiring because of AI, Amazon and Meta have framed
Liu Gong's take: the 60,000 buyout is not the end point. Within 12 months Microsoft's global headcount is likely to fall below 190,000, and the next round will be framed as role restructuring rather than headcount reduction.
On the business side, this is a swap of cost structures. Microsoft's quarterly capex passed $30 billion in fiscal 2026 while payroll runs at roughly $70–80 billion a year. Letting part of that 60,000 leave voluntarily saves cash that is small next to compute spending; the real point is converting a fixed cost into a variable one — servers can be idled, headcount cannot. Choosing buyouts over layoffs also smooths the cost curve on the income statement.
From a technical standpoint, substitution is happening in specific layers rather than across the board. Nadella's own figure is that 20% to 30% of code is AI-generated, a share that runs higher in testing, documentation and boilerplate and lower in architecture and cross-team coordination. The first layer compressed is the middle — coordination roles that neither decide nor execute, precisely the category Microsoft expanded fastest over the past decade.
For the industry, the symbolic weight of 60,000 exceeds its financial weight: about 26% of global headcount, nearly four times the combined 2025 cuts, and the first time a large software company has used voluntary buyouts to handle AI substitution. Liu Gong predicts that within six months at least one of Oracle, SAP or IBM will announce a comparable programme. Their role mix closely mirrors Microsoft's, while their AI infrastructure spending is far smaller, which makes the cost pressure sharper rather than lighter.
In the industrial inspection work Liu Gong knows best, the pattern is plainer: models absorb repetitive judgement while annotation, model maintenance and exception handling roles appear, and the ratio is rarely one to one — three to one or lower is typical. Software will not be the exception, only one or two quarters slower.
- The buyout covers about 60,000 people, or 26% of Microsoft's roughly 228,000 global employees; the May and July 2025 rounds together removed about 15,000, or 6.5%, making this programme nearly four times larger.
- Nadella said publicly in 2025 that 20% to 30% of code inside Microsoft is AI-generated; GitHub Copilot passed 20 million cumulative users over the same period on Microsoft's disclosed figures.
- Quarterly capex rose from about $19 billion in Q4 of fiscal 2024 to more than $30 billion in fiscal 2026, a gain of over 50% in two years, while total headcount has been flat or falling for several consecutive quarters.
- The programme spans the US, UK and Canada; Microsoft's annual
For the industry, in the short term (three months) large software firms in North America and Europe will treat voluntary buyouts as a standard tool, lifting demand for HR services and outplacement support. Over twelve months, hiring will tilt further toward AI infrastructure, data governance and security, mid-level coordination roles will keep shrinking, and graduate intake will be reallocated.
For ordinary users, Copilot-style products will not slow down because of the restructuring; they may speed up as resources concentrate. Over the medium term, enterprise software pricing will tie more closely to usage and documented automation savings, and buyers will demand that AI premiums be
Questions worth exploring:
- How voluntary buyouts versus forced layoffs differ on cost, legal risk and employer brand, and which path other tech firms copy
- What happens to software margins when payroll is swapped for compute depreciation, and whether reporting conventions need to change
- With the middle layer thinned, entry-level staff lose the promotion ladder — where will managers of AI-native organisations come from in five years
Further reading: 2026 Frontier LLM Comparison: A Practical Guide — a durable, evergreen reference. Bookmark it.
This item comes from The Verge AI (published 2026-10-29 16:00:00). Wenfeng Radar provides a Chinese/English summary and commentary on overseas AI developments; the original content is owned by its author.
Daily aggregation of overseas AI developments with in-depth analysis. Bookmark us to never miss an important signal; back to homepage.
