AI Trends

Microsoft offers voluntary buyouts to about 60,000 employees as tech giants trade manpower for computing power

HuxiuUpdated Nov 17 11:15

AI summary

Microsoft is offering a selective voluntary buyout plan to about 60,000 employees; those eligible account for more than a quarter of the total workforce, targeting employees at job grade 70 and below on teams in Canada, India, the UK, and the US. Microsoft's fiscal 2026 revenue was $331.8 billion and operating profit was $155.2 billion, up 18% and 21% year over year, but capital expenditure for fiscal 2026 is projected at $190 billion, mainly directed at AI chips and data centers. The article also cites Oracle's roughly 21,000 layoffs in the previous fiscal year, Amazon's roughly 16,000, and Meta's roughly 8,000, and discusses the expansion of computing power demand under the Jevons paradox and how individuals can build irreplaceability.

Why read this

AI score 57

The article groups Microsoft's buyout plan together with workforce reductions at Oracle, Amazon, and Meta, outlining an industry trend in which labor costs give way to AI computing power.

Topics

Tags

Body · AI translation

**Key Points**

**1. Microsoft's profits rise on both counts, yet it still cuts staff in disguise**

In fiscal 2026, revenue was $331.8 billion and operating profit was $155.2 billion, up 18% and 21% year over year; but spending on data centers and chips was $145 billion in the previous fiscal year and is projected at $190 billion in fiscal 2026, with headcount making way for AI compute first.

**2. Tech giants collectively trade people for AI compute**

Oracle cut about 21,000 jobs last fiscal year, a 13% reduction; Amazon cut about 16,000; Meta cut about 8,000. Under the Jevons paradox, falling AI costs actually push demand up, and once the compute race slows, a company may be out of the game.

**3. The buyout threshold shows a revaluation of organizational value**

The buyout is aimed at level 70 and below, signaling that the core layer of senior vice presidents and above will be retained; some employees in data centers, sales, marketing, cybersecurity, quantum computing and other areas are ineligible, and those close to customers and able to solve real problems are more likely to be kept.

**4. Individuals build irreplaceability through creativity, warmth and accountability**

Creative talent who can harness AI, people who can build trust and empathy, and people willing to take responsibility for complex problems and results are harder to replace and more likely to gain bargaining power in the market. 2026-10-28 17:39 Cutting 60,000 people sounds a warning bell Zhenghe Island Quick View

This article comes from the WeChat public account: Zhenghe Island, author: Guazao

Over the past two days, a personnel adjustment at Microsoft has caused quite a stir online.

Microsoft offered about 60,000 employees a "selective voluntary departure buyout" plan: if they leave voluntarily, they can get a sum of money.

The wording seems mild and restrained, but so-called voluntary departure is, in the end, a "variant" of downsizing.

Cutting 60,000 people, Microsoft's big game

This is actually not the first time Microsoft has done this; Microsoft has already carried out several rounds of personnel adjustments this year. This one is just the largest, with eligible employees exceeding a quarter of its total workforce.

There is another detail very much worth noting: Microsoft's fiscal 2026 (ended June 30) revenue was $331.8 billion, up 18% year over year; operating profit was $155.2 billion, up 21% year over year.

Why is Microsoft cutting people so frequently even though both revenue and profit are rising?

Microsoft spent $145 billion on data centers and chips in the previous fiscal year, and capital expenditure in fiscal 2026 is expected to be $190 billion, most of it going to AI chips and data centers. Microsoft will continue to spend heavily going forward, so every link has to be squeezed, and cutting labor costs is often the easiest to do.

In fact, Microsoft is not the only one doing this. Oracle cut about 21,000 jobs last fiscal year, a reduction of 13%. Oracle said in its annual financial regulatory filing that "the adoption and deployment of AI technologies across our businesses has already led, and may continue to lead, to reductions in headcount."

Amazon announced in January this year that it would cut about 16,000 corporate positions, and Meta launched layoffs of about 8,000 people in May this year, about 10% of its total workforce.

...

Giants dare not leave the table

So why are these giants trading people for compute?

Here, one economic term has to be mentioned: the Jevons paradox.

In 1865, economist William Stanley Jevons pointed out a paradox in The Coal Question: Watt's improvement of the steam engine raised coal efficiency, but Britain's total coal consumption rose rather than fell.

The earliest Newcomen steam engine had very low energy efficiency, with more than 99% of the energy wasted and an actual energy utilization rate of less than 1%. Because efficiency was so low, the machine could only pump water next to mines, because fuel was very easy to obtain and transport costs were almost zero; once moved away from the mining area, transport costs would eat up all the gains.

Watt later improved the energy efficiency of the steam engine. Logically, with higher coal utilization, coal consumption should have fallen. But after Watt's improvement, his partner Boulton brought it to market, and ultimately the steam engine moved out of the mining areas: urban textile mills could use it, and flour mills, ironworks and other places could use it too.

As a result, more industries used coal, more people used coal, and in the end more coal was burned.

At first, Britain's annual coal output was about 2.8 million tons. A few years after Watt improved the steam engine, annual coal output rose to about 10 million tons.

By the time Jevons summarized this phenomenon, Britain's annual coal output was already close to 95 million tons.

Behind this, the Jevons paradox illustrates a phenomenon: technological progress does not shrink demand; instead, it opens up suppressed demand space.

Today, almost no one considers the cost of calling large models; more people are calling large models, and demand not only does not shrink but grows wildly.

As the well-known corporate strategy expert Zeng Ming puts it: after infrastructure crosses a certain usability threshold, native applications begin to emerge densely, and once applications appear, they in turn drive the spread of infrastructure.

Just like electricity: people were willing to connect to electricity only after they saw electric lights; after connection, products such as electric irons, vacuum cleaners and washing machines appeared; and these products in turn promoted the spread of electricity, which then gave rise to more native products; finally, electricity became underlying infrastructure.

AI is the same: after calling costs cross the threshold, native applications such as agents and coding assistants emerge densely; the richer the applications, the higher the return for enterprises connecting to compute, more enterprises run their businesses on the platform, and market expansion in turn gives rise to more native applications, pushing model capabilities and compute technology to further maturity and ultimately penetrating every industry.

Now, the metric unit of Token has been proposed. Like a kilowatt-hour of electricity, a ton of water, or a GB of data traffic, it is fully measurable, proving that AI is becoming infrastructure.

So the situation now is this: the price per unit of Token is falling, but global compute demand, GPU rental prices and total AI spending are soaring.

Compute, model capabilities, the developer ecosystem, and the embedding of enterprise customers' workflows all have economies of scale and lock-in effects. Whoever builds out compute first, whoever pushes model capabilities to the frontier first, and whoever gets customers to run their businesses on its platform first will be able to capture the incremental growth that has been unlocked. Conversely, whoever is one step slower in expanding is not just earning a little less, but handing an already opened market to rivals.

The giants dare not lose; they are fighting a survival war over frontier position and compute scale. But for most companies, the way to participate in the fourth industrial revolution is not to build large models, but to borrow platform capabilities to improve their own room to survive.

Although it seems to have nothing to do with most enterprises, behind Microsoft's "disguised workforce reduction" there is also something worth deep reflection by enterprises and entrepreneurs, and it sounds a wake-up call for everyone.

This time, Microsoft's "voluntary-exit-style workforce reduction" states the conditions clearly: "eligible employees in teams in Canada, India, the UK and the US at job level 70 and below."

Setting the threshold at level 70 conveys at least one signal: Microsoft is in fact excluding the company's most core decision-making layer, senior vice presidents and above. This ensures that while the company "reshuffles its blood" and cuts costs, it can retain the people who set strategy and make key decisions.

In addition, there is another noteworthy point: "some employees working in data centers, sales, marketing, cybersecurity and quantum computing are not eligible."

Translated, that means: these departments are not being cut.

Putting these two points together makes it clear:

The core decision-making layer, as well as those frontline people closest to customers and those who can solve real problems, remain very valuable assets to the organization.

In other words, in the AI era, the logic of organizations has completely changed. In the past, judging a person's value meant looking at which level they were at, how many people they managed, how many resources they could coordinate, and how much weight they carried in the upward and downward transmission of information; keeping a person was calculated by how much work they did and whether they were worth the salary. Now it looks at whether, after AI has taken over information processing and process coordination, they still have abilities AI cannot replace, and whether this person can still make things AI cannot make.

Just like Anthropic, which does not look like a company in the traditional sense. The organization is extremely flat; it has no clear boundaries and no stable division of responsibilities. AI takes on work related to organizational context and processes, while people concentrate on the key judgments and accountability that AI cannot handle.

Now that Microsoft is "optimizing" its organization, how many positions at "experience level 70 and below" are there in your organization?

Of course, this is not telling you to reduce staff. Rather, it is asking operators and senior managers to think:

The core point is that business operators should, according to changes in business, markets and the environment, proactively adjust their organizations and sort out their talent structure.

Build irreplaceability

Adjustments need to be made at the organizational level, and for workplace professionals, improving their own irreplaceability is crucial.

In the past, no matter how excellent a person was, their personal time was limited and knowledge had barriers. Today, mobilizing intelligence can cross these barriers. At this point, a creative person who can harness AI can demonstrate the ability of "one general matching a thousand soldiers."

That is why we often see fragmented news. Some big companies, on the one hand, are carrying out large-scale layoffs, while on the other hand they are poaching people with high salaries, willing to pay annual salaries of millions or even tens of millions of dollars to get one talent.

In fact, this is not really contradictory; rather, companies are more willing to spend money on people who are worth it, on people with high "ROI."

2. Possess human warmth.

That is, the connections between people.

Take sales as an example. In the sales industry there is a classic saying: what customers appear to buy is the product, but what they actually buy is their trust in you.

Sales is about dealing with people and building connections with people. Only if you can convey warmth to others and solve their problems will they pay you, and this is especially true the higher the customer transaction value. This is also why "sales" was not eligible in this Microsoft layoff.

3. Take responsibility proactively.

In "AI Superpowers," Kai-Fu Lee drew an action map for individuals responding to the AI era, pointing in three directions: move toward "openness," to places where problems have not yet been defined and standards do not yet exist; move toward "complexity, risk and gatekeeping," putting yourself in the position where "if something goes wrong, you are responsible"; and move toward "warmth and trust," doing things that can build trust with people.

The author believes that all three directions have value, but the one about "complexity, risk and gatekeeping" appears extremely critical.

When AI does something wrong, it only apologizes but does not take responsibility; this is its most fundamental limitation. But responsibility must fall on a person or an organization. If you push yourself into the position where "if something goes wrong" you are responsible, then you are irreplaceable in a company.

The deeper meaning of this sentence is that you need to identify which places in the organization are important, then stand up and shoulder responsibility there. You need to make judgments about direction and be responsible for the results of things.

Here we might as well give an example. Suppose you are in charge of the company's official account and present the brand externally. Now AI can be used to write articles. If you can only write articles, then your irreplaceability is very low. But if you are the editor-in-chief, you need to control the tone and direction of the account and also be responsible for the final result, then relatively speaking your value is higher. So even if you are not the editor-in-chief, you should hold yourself to the mindset of an editor-in-chief.

When you have a certain degree of irreplaceability, you will have bargaining power in the market.

Conclusion

In his book "The Black Swan," Taleb tells a story about a turkey's happiness index:

A turkey was raised on a farm from a young age. Every day at noon, the owner would come on time to feed it. 10 days, 20 days, 100 days, never interrupted. The turkey therefore concluded that the owner was the person in the world who loved it most, and its happiness index rose straight up, and it believed such days would continue forever.

Until Thanksgiving Day, when the owner walked in holding a knife. The turkey became a dish on the plate. Although the story seems tragic, it is a portrayal of reality.

For enterprises, they should take advantage of this window to re-sort out value and find their core competitiveness. The same is true for individuals: they need to proactively evolve and enhance their own value. Only with high irreplaceability will they be relatively safe.

References:

[1]. "Intelligence," CITIC Press Group

Has published 248 articles on Huxiu. This content comes from the internet, original link; the views represent only the author and do not represent Huxiu's position. If there is any copyright issue, please contact hezuo@huxiu.com, and we will verify and handle it promptly. Everyone who is changing and wants to change the world is on the Huxiu APP[ ]

Source

Huxiu