Microsoft is "Laying Off" Again

Image Source | Internet (Please contact us for removal if infringing) Partially generated by AI

Microsoft is "laying off" again. But this time, the scale is too significant to casually brush off with the word 'again.'

On October 26, 2026, Microsoft offered a voluntary severance buyout to approximately 60,000 employees worldwide, covering the U.S., Canada, the U.K., and India—about 27% of its global workforce.

Nearly a quarter of Microsoft's global workforce found themselves faced with the question of whether to leave in a single email.

In the era of AI, the narrative of tech giant layoffs has consistently centered on AI costs.

However, the real focus this time isn't the figure of 60,000 itself, but rather its peculiar relationship with another number.

In the same year, Microsoft projected capital expenditures of $190 billion, with roughly two-thirds allocated to "short-term assets" like GPUs and CPUs. These assets begin depreciating immediately upon purchase, becoming worthless within three to five years.

On one side, 60,000 careers stand at a crossroads; on the other, $190 billion in computing power arms race. Both events unfold within the same company and year, yet they logically interlock.

This is a financial operation aimed at restructuring cost structures. What Microsoft is doing is far colder than mere "layoffs": It is transforming itself from a people-driven software company into an AI factory powered by computing throughput. In this factory's ledger, human resources are shifting from "assets" to "old capacity" in need of replacement.

Behind Microsoft's "Grace"

Let's clarify the nature of this scheme: It is not forced layoffs but voluntary severance buyouts.

Eligible employees must decide by early December whether to accept, with departures scheduled before year-end. Compensation terms resemble Microsoft's standard severance package: at least 60 days' basic salary, up to 39 weeks' basic salary based on seniority and tenure, continued stock vesting, medical coverage, and career transition support for a period.

This sounds far more "dignified" than receiving a cold meeting invitation on Monday morning. However, the word "voluntary" is precisely the key to understanding Microsoft's latest move.

This marks Microsoft's third round of staff adjustments in 2026. In April, voluntary retirement was offered to employees whose age plus tenure exceeded 70, with about 30% of eligible staff accepting. In July, 4,800 employees were laid off directly, with Xbox being the hardest hit. In October, the scope expanded to 60,000 globally.

Each round grew larger and more "voluntary." Yet, a Microsoft spokesperson stated bluntly: This scheme "does not replace layoffs; future adjustments will still be made based on business needs."

In plain terms: The door is open; you may leave with compensation, but no promises are made about whether others will be pushed out afterward.

More intriguing is who is excluded. Data centers and cybersecurity departments are not eligible for the buyout; neither are marketing and frontline sales teams; nor are the core engineering teams at the lowest levels.

Those covered include Windows, Copilot, Xbox, finance, human resources, and a vast middle ground of "vice president level and below."

Data centers cannot be "optimized" because they form the factory floor of the AI operation; core engineers cannot be "optimized" because they are the moat; frontline sales cannot be "optimized" because Azure orders still require human signatures.

Those invited to the table are precisely those who do not directly produce computing power, do not directly generate AI revenue, and were responsible for keeping the vast machinery "running smoothly" in the old software era.

Analysts at Wells Fargo calculated that if one-third of the 60,000 accept the buyout, it could boost operating profit margins by approximately 2 percentage points on an annualized basis.

Two percentage points are no small figure for a company of Microsoft's scale. And this saved money can almost seamlessly flow in another direction: depreciation and computing costs.

The "Depreciation Gravity" of a $190 Billion Investment

To understand why Microsoft must take such drastic action on the human resources front, one must first grasp the financial nature of AI infrastructure.

Traditional software companies spend money on people and code. Human costs are flexible—they can hire, freeze hiring, or "optimize." Code, once written, can be replicated infinitely at near-zero marginal cost.

But AI infrastructure is an entirely different beast. The $190 billion in capital expenditures buys data centers, GPU clusters, power, and cooling systems.

Once deployed, these assets follow a rigid depreciation schedule. GPUs do not depreciate less just because revenues fell short one quarter; data centers do not stop consuming power just because Copilot sales underperformed.

More critically, a significant portion of this $190 billion consists of "short-lived assets."

GPU and CPU technology iterates on an annual basis, rendering previous generations obsolete within a few years. This means Microsoft's profit statements are now burdened with an increasingly heavy layer of depreciation gravity.

Q3 2026 earnings showed Microsoft's gross margin dropping to 68%, the lowest since 2022, officially attributed to increased depreciation costs from AI infrastructure investments.

Free cash flow is also tightening: Goldman Sachs estimates Microsoft's free cash flow will plummet from approximately $67 billion in FY2026 to about $40.1 billion in FY2027—a nearly 40% decline.

Thus, every dollar saved on human costs is not "profit" in essence but a buffer against depreciation.

Microsoft is not replacing humans with AI and then distributing the savings as dividends. It is contracting on the human side to create breathing room for the expansion on the computing side.

The headcount structure of the old software era cannot support the asset structure of the new AI era.

The 60,000 buyouts essentially represent a cost structure swap: replacing predictable monthly human costs with rigidly scheduled computing depreciation costs.

What is being replaced are the carriers of the old model; what is being brought in is the throughput capacity of the AI factory.

A Buyout of an Entire Organizational Syntax

Focusing solely on the financial layer would overlook a more subtle dimension: What Microsoft is dismantling is not merely 60,000 positions but the organizational syntax that made those positions exist.

In May 2026, Business Insider revealed that Nadella had quietly abolished the Senior Leadership Team (SLT), a decades-old power center consisting of a dozen executives reporting directly to the CEO. It was replaced by a five-person core circle.

The old SLT was essentially a "fiefdom system": Windows, Office, Azure, Xbox, and Security each operated independently, with separate budgets, KPIs, and territories.

Copilot needed cross-departmental coordination to access Windows data; Azure required layered approvals to deeply integrate with Office. Even Nadella himself had complained internally about Microsoft's inability to seamlessly collaborate across its own products.

The consequence of this structure was that AI was treated as an "add-on" for each business line rather than the operating system itself.

Every fiefdom built its own AI, but no one took responsibility for Microsoft's overall AI experience. Abolishing the SLT meant dismantling the fiefdom system.

The five-person core circle no longer represents individual territories but the company as a whole; the 35-person engineering leadership group transcends old business lines with unified goals.

Xbox's actions are even more intuitive (intuitive): Management layers were compressed from 14 to no more than 5, ideally to 3.

What does 14 layers mean? Information passing from the frontline to decision-makers goes through 14 hands, each adding filters, delays, and "opinions." Reducing it to 5 layers means evaporating the roles of nine middle managers.

These individuals are precisely the core group covered by the buyout scheme: Level 70 and below, below vice president, sitting in the middle layers, not directly generating AI revenue, but occupying information and approval nodes in the old system—the "organizational lubricants."

Nadella publicly stated that Microsoft's massive size "has become a huge disadvantage in the AI era." The organizational efficiency of 228,000 employees cannot easily compete with Silicon Valley startups at the same pace.

Thus, what Microsoft is doing is not simply "reducing headcount"; it is transforming the organization from "managing people through layers" to "managing computing power and products through nodes."

Model research and product delivery have also been separated: Suleiman no longer oversees Copilot's daily development but focuses on superintelligence and proprietary models; Copilot products are handed over to Jacob Andreou, who reports directly to Nadella.

Models represent long-term exploration, while products require short-term iteration. The organizational cultures and evaluation logics needed for each are entirely different. Bundling them under one person weakened both in the past.

These actions and the 60,000 buyout scheme are different facets of the same initiative. The buyout clears people; organizational restructuring clears positions.

When 14 layers become 5, when the SLT becomes a five-person circle, when models and products split, those "voluntarily departing" are actually being ejected by a discarded organizational syntax.

The Entire Silicon Valley is Solving the Same Equation

Placing Microsoft's 60,000 layoffs within the 2026 tech industry landscape transforms it from an isolated "big tech story" into the loudest note in a collective movement.

The four hyperscalers—Alphabet, Meta, Amazon, and Microsoft—are projected to spend a combined $674 billion on AI infrastructure in 2026, more than double their 2024 levels.

Meanwhile, Meta announced layoffs of approximately 8,000 employees, 10% of its workforce; Amazon initiated its largest-ever round of layoffs in early 2026, putting 30,000 positions at risk; by 2026, the tech industry had already laid off over 90,000 workers.

Every company is solving the same equation: The denominator of AI capital expenditures is swelling, requiring cost compression on the molecular side to balance.

However, Microsoft's unique challenge lies in the still-open gap between AI investments and AI revenues.

Azure's revenue soared: It surpassed $100 billion for the first time in FY2026, with growth accelerating from 40% to 43%, and next quarter's guidance around 45%.

AI business annualized revenue exceeded $37 billion, up 123% year-over-year. From the computing supply side, Microsoft's AI narrative is solid—demand exceeds capacity, and Nadella once said, "If we had enough GPUs, growth would already exceed 40%."

But product penetration tells a different story. Copilot's paid seats surpassed 30 million in July 2026, a respectable growth rate, but still a single-digit percentage of Microsoft's 400 million Microsoft 365 commercial users.

The vast majority of commercial users have not paid extra for Copilot. Windows 11's Recall feature was forced offline for privacy overhauls after controversy. NPU chips in laptops lack valuable AI use cases. In developer surveys, GitHub Copilot trails behind Claude Code and Cursor in preference.

This is Microsoft's current dilemma: Computing-side revenues are accelerating, but the product-side "AI tax" has not yet reached ordinary users.

Azure's growth can absorb some depreciation but not all. With free cash flow declining, gross margins pressed to multi-year lows, and investors beginning to price stocks based on "capital expenditures/free cash flow" rather than "AI narratives," the human side must respond.

The 60,000 buyouts represent the time Microsoft bought itself before AI revenues fully cover computing costs.

Will AI Steal Jobs?

Returning to the question easily misled at the outset: Is Microsoft laying off employees because AI has taken their jobs?

Microsoft's official stance is nuanced. Chief Human Resources Officer Coleman framed the buyout scheme in an internal letter as an option to "help employees transition with company support."

During the July layoffs, Microsoft stated, "The laid-off positions were not replaced by AI, but AI is changing how work is done."

Together, these statements clarify: AI is not directly eliminating these roles but redefining which roles are worth keeping.

The middle managers covered by the buyout are not being "replaced" by a single large model. Rather, their existence relied on an organizational form—14 layers of approval, fiefdom-based budgets, promotions by jurisdiction—that no longer holds in the AI factory's cost structure.

When computing power becomes the core production resource, when model and product iteration cycles are measured in weeks, and when Nadella needs to review AI metrics weekly and meet with Azure infrastructure teams biweekly, the space for "information relay layers" and "approval lubrication layers" is structurally compressed.

AI didn't sit in their chairs, but AI has changed the way chairs are arranged.

For the industry, Microsoft's buyout plan serves as a mirror: During the window period of cost structure replacement, what is priced is not 'whether you can use AI,' but rather 'your position—whether it represents an asset or outdated capacity in the new cost structure.'

Data centers and cybersecurity are excluded, as are core engineers and frontline sales personnel, primarily because these roles are directly tied to computing power throughput and AI revenue.

Those invited to the table are the ones who kept the massive machinery running smoothly in the old structure but do not directly engage with computing power or revenue in the new structure.

This logic is not unique to Microsoft. As the AI capital expenditures of the four tech giants swell into the hundreds of billions of dollars, and as depreciation becomes an increasingly heavy line item on profit statements, the entire workforce structure will be forced to answer the same question: Are you producing for the computing power factory, or are you merely maintaining costs for the sake of the old company's dignity?

Microsoft has not provided an answer. It has simply placed 60,000 buyout offers on the table while continuing to sign $190 billion worth of computing power orders.