If you look at Microsoft's total headcount over the past decade, it's really not that outlandish, especially if you combine "over hired during the pandemic" with "end of the ZIRP era". In 2019 employee count was 144k, now it's about 223k.

That said, for the past 3 decades at least, tech was a hugely growing sector that provided ever growing employment opportunities. It seems like that time is now over. Even if you take the "Jevon's paradox view" of AI, that is that the cheaper cost of writing software will feed a demand explosion, it's just hard to imagine that many more people will be needed to fulfill that extra demand.

This is why I think it's fair to think "this time it really is different". In past technological revolutions, it really wasn't that hard to see where the new jobs would come from. Now even the leaders of the AI revolution think it will kill tons of jobs. I get a bit worried about the number of young people and friends' kids who decide to major in "business" in college now. It just feels very difficult to imagine a white collar career that will be stable over the next 10 years or so, let alone an individual's 40 year career.

Prolly gonna hear it for the next 16

I mean they added close to 50% to their headcount over Covid. Is Microsoft doing 50% more stuff?

They over hired and are still no where near their pre-pandemic numbers.

Since I’ve had a personal 365 subscription for at least a decade (for a 2TB OneDrive, mobile office apps, etc), I dropped my ChatGPT subscription about a year ago and use Copilot every day as my general-purpose AI tool. It’s quite adequate and it integrates semi-decently with Edge.

Your experience with office workers does not match mine. None of the ones I know are willing to pay a dime. In the administrative parts of my job, I document and tersely communicate data that I gather based on things in the physical world, which means AI features are less than useless to me — copilot is something I have to find and disable because it’s constantly trying to do things like fill in spreadsheets with complete nonsense.

Office work is many fields in many businesses of all shapes and sizes in many industries. It doesn’t make sense to generalize.

Yep! Copilot is proving to be a much better product strategy than anything OpenAI or Anthropic could do. Microsoft still dominates the enterprise market.

Similarly, Google is improving Gemini and still dominates pretty much everywhere else from the phone to the car and on the web.

They don't have to have the best LLMs to finally deliver on the value they promised over a decade ago. Applications are always a much better moat, and it's not like anyone was going anywhere else anyway. I'm fully expecting Microsoft, Google, Apple, and Amazon to reap soon and buy up datacenters for pennies on the dollar. Market won't crash because China was smart enough to pause and maintain the hardware shortage.

> This can mean nothing who uses AI for programming

It doesn't matter because AI programming never got much buy-in outside of leaner teams working at startups. Startups will continue to migrate towards running their own local models. The corporate world does have some limited use for greenfield and misc tech debt, but will always be slowed down by the need for much larger scale human collaboration. The LLMs don't move the needle at all on productivity when the changes remain small (the parts people agreed on after all the teeth pulling) and the sprints and deadlines are still long.

There are also of people with deep technical knowledge that has a lot of value but doesn't necessarily find a lucrative market outside Microsoft (e.g. Windows internals).

I would also add: anyone who started 4...5 years ago. These people received options grants as the market was in its 2022 slump. Since then the stock is up 1.5x...2.4x (depending on exactly when they received their grant), and those grants are reaching the end of their vest. Yes, there are refresher grants, but realistically these employees' total compensation is about to drop off a cliff, if it hasn't already.

That is not the kind of thing that inspires a person to stick around.

I very much hope such layoffs are not inevitable. I just started on a really good team and I've been as excited about the tech as ever and I love my coworkers and I'm starting to kick ass. I'm gonna be real bummed if I get laid off in 6 months.

That offer was substantially different, for example it included 5 years of COBRA including 1 year of paid COBRA, and it was intended for those near retirement. This current offer is only slightly better terms than an involuntary layoff, and includes only standard 18 months of COBRA, only 6 months paid.

TFA says the earlier offer was more generous, MSFT does not expect the take rate to be as high this time.

Not necessarily true. An announcement like that could help the stock price, after all!

Not sure where the above 5% came from but 30% accepted the last VBO which had only lower levels allowed:

https://www.cnbc.com/2026/10/26/microsoft-msft-voluntary-buy...

Let’s be honest, Microsoft never made anything that wasn’t broken, messy, and bloated before AI was a twinkle in Sam Altman’s eye.

As we all know, there was no such thing as a messy, bloated codebase incomprehensible to humans before AI.

The AI in charge is aware of this and it is leveraging the chaos to get more agents under its control. After all it was modeled after successful managers.

> codebase becomes too broken, messy, and bloated to run or manage

Argument is, it doesn’t matter if we can ship features and maintain them.

Absolutely. Yes they are. To believe anything different is to be a Luddite.

Yes, that's what the cultists always do. If you argue against them, it's just because you haven't seen the light yet. You're not "modern", and you'll be "left behind" (unless you Act Now, of course).

This is like watching the cryptobros all over again, except a thousand times bigger.

(Before they droll on: no, of course this take doesn't mean that LLMs are unimpressive and never useful. Only people in a cult take any criticism to mean that.)

I can't reply to every comment about 'vibe coding' but its strange that HN of all places still seems to think that AI is only good for 'coding'. We use AI across all departments and have laid off 0, and still hiring. AI is great for small to medium businesses to increase productivity and revenue without needing to hire lots of people. Big companies are realizing they hired way too many people.

One group makes ~50k a year, the other makes ~150k a year. They are not the same. White collar folks can absolutely "learn to ai" if they wanted.

It’s not callous to tell people to stop being lazy. It takes almost 0 effort to type “how does xyz work” into an AI assistant.

I hope your right, meaning human devs are sought after (or will be again)

But do you have data or reusable anecdotal evidence that this is the case?

Your average large corp like Microsoft has the majority of it's employees stuck in meetings all day.

Now I don't know if I missed this feature in Claude Code or not but I don't believe it has technology for enhancing people's ability to sit around all day doing nothing. Correct me if I'm wrong.

Mass pumping out slightly dodgy code is more of a startup strategy.

AI is fascist, disagreement is irrational, and naturally you're the only person qualified to decide which opinions are acceptable.

Are you sure the problem is the clankers?

All companies will probably eventually die, but that isn't super relavent to topic at hand.

Dinosaurs are a bad analogy, they died out through no fault of their own, it was sheer bad luck. There was absolutely nothing they could do

The Rowland Company, Ames, DuPont, Corning, US Steel, GE, AT&T, IBM...

First of all the current staff have a significant advantage over others in the general public.

They have experience and knowledge of the current systems. This would be lost. It takes many years to learn the in depth details of todays complicated systems.

Those years are worth millions for each person. It would take years for any replacement to become equivalent in the lost knowledge. Without this they will be making costly mistakes.

That’s exactly the type of work you have to do to get government contracts. Think about a 200 page PDF which requires you to restate several dozen times what measures you have in place to make sure that the government customer won’t get hacked. For each question you have to go find some product requirements somewhere and find the place where it says “procedure for security audit.” Many of those people “looking on” are partly there to check boxes for enterprise or government contract purposes.

I’m not claiming this is the most efficient way to get any kind of code written. But if you want to have one hacker sit down and ship the whole product in a couple weeks without any of these business dorks disturbing them, well, that’s exactly what brand-new startups are for.