Microsoft and Meta cut 16,000 jobs the same day, blamed AI, and kept their AI teams

I was drinking coffee on a Thursday morning in April, minding my own business, when two of the biggest companies on the planet decided to make my afternoon interesting.

By the end of 23 April 2026, Microsoft and Meta had between them announced cuts touching about 16,000 jobs. Same day. Both pointed at AI productivity as the reason. And both, when you read the fine print, carefully ringfenced their AI teams from the cuts.

That last bit's the whole story, so I'll say it plainly. If AI were really doing the work, you wouldn't need to keep hiring more people to build the AI.

Poonam Soni

@CodeByPoonam

9,000 Microsoft employees did everything right.

Top schools. FAANG resumes. $200K salaries.
Still got replaced by a line item in an AI budget.

Senior engineers. Directors. PMs with FAANG resumes and 15-year careers.

The memo called it “voluntary retirement.”
Nobody

3.3K

23 Apr 2026

Poonam Soni's post on it crossed a million views, and the line that stuck was "replaced by a line item in an AI budget." Not replaced by robots. Replaced by a budget reallocation. That's the honest truth about what happened, and it's a very different thing from the headline you'll have read.

What actually got announced

Microsoft opened its first voluntary retirement program in 51 years. To put that in perspective, the last time Microsoft did this, it wasn't really Microsoft yet. The framing was a "voluntary retirement program" using a "Rule of 70" formula, where your age plus your tenure had to add up to 70 to qualify. Estimates put it around 8,750 US employees, roughly 7% of the US workforce, with notifications starting 7 May (CNBC, 2026). Microsoft AI and Copilot teams were exempt.

Meta, the same day, announced about 8,000 layoffs and closed roughly 6,000 open roles, around 10% of its workforce, with AI efficiency cited as the driver (SiliconANGLE, 2026). The cuts were announced on the day and began executing from 20 May. Meta's Superintelligence Labs, the AI group, kept hiring straight through it. (Reality Labs, the metaverse division, got hit hard in the same restructuring, so this wasn't a case of sparing every expensive bet. It was specifically the AI teams that were protected.)

Two companies, two press releases, one suspiciously identical message, all in a single news cycle. Either there's a memo doing the rounds, or both PR teams subscribe to the same newsletter. I genuinely don't know which is worse.

And it isn't just tech. The same week, KPMG cut about 10% of its US audit partners, which one finance commentator noted in the same breath as the Microsoft and Meta numbers (@money_cruncher, 2026). Audit partners. The people who own a slice of the firm. The "AI productivity" story's being told well outside software now.

The framing versus the mechanism

Here's the thing the announcements don't say out loud. AI didn't walk into a meeting and fire 8,750 people. What's going on is simpler and, in a way, more honest if anyone'd just say it.

Rihard Jarc

@RihardJarc

$META cutting 10% of the workforce (8.000 roles) and eliminating 6.000 open positions.

This will happen across the board in the tech sector and more broadly in knowledge work, Zuck is as always just ahead of the pack. Companies are switch labour costs for AI compute.

254

23 Apr 2026

Rihard Jarc, a VC, put the whole thesis in one line: companies are "switching labour costs for AI compute." The salary budget gets smaller. The compute budget gets bigger. The money doesn't disappear, it moves house. A Wedbush analyst described Meta's cuts as automating "tasks that once required large teams," which is a polished way of saying the productivity gains AI promised are now being cashed in, and workers are the ones paying.

The part that's genuinely new, and a bit grim, is the recursion. A lot of the people being cut are the engineers who built the very platforms the AI runs on. They built the company, and now the company's spending the savings on the technology that made their roles look optional. I've been doing this twenty years and I can't think of a clean precedent for it.

The fair counterargument

Now, I'd be having a go at the executives without giving them their due if I didn't put the other side. Maybe the cuts are genuinely justified. Engineers I talk to do report AI making routine work 30 to 40% faster, and that's a real, measurable gain. In a flat market, a real productivity gain does mean the same output with fewer people. That's not a trick. That's just what efficiency looks like when you're not growing.

Fine. I'll accept all of that. But here's where I get stuck. The same productivity logic that lets a company claim credit to shareholders for being lean also lets it skip the harder conversation: what does it owe the people whose roles it compressed? "AI productivity gains" points the camera at the technology and away from the decision. The decision was made by humans, in a boardroom, about a budget. The honest version says "we restructured, here's the help we'll offer." I'd rather work for that company than the one hiding behind a productivity slide.

I should own something here too. I've spent the best part of 18 months telling clients AI's job impact was overcooked, and I leaned on the data to do it. The Oxford Economics work that put real AI-attributable losses closer to 4% still holds up, and most cuts genuinely are restructuring dressed up. But the gap between "restructuring" and "AI displacement" is narrower than I was claiming, and I'd be a hypocrite to pretend otherwise.

Related Article8 min read

The AI Layoff Myth: Only 4.5% of Job Cuts Are AI-Related

Oxford Economics analysed thousands of layoff announcements. AI is responsible for just 4.5% of job losses. 'Market conditions' is cited 4x more...

Read full article

If you want the financial flipside of all this, the savings rarely stay saved. The companies cutting salaries to fund compute are finding out that compute sends invoices too, and they're not small ones.

Related Article11 min read

You laid off your engineers to save money. Then spent it all on AI.

Companies cut hundreds of thousands of engineers in 2023-2025. AI productivity would cover the gap. Then the invoices started arriving. Uber burned...

Read full article

What I'd watch from here

If you're running a business here, you're not going to get an 8,000-person announcement. You'll get the quiet five-person version, with no press release. The pattern to watch isn't "will a robot replace my team." It's whether your IT vendors keep hiring at the same rate, and whether the savings you're promised from AI actually show up after the compute bill lands.

Expect more of the same: companies announcing AI-attributed job cuts while, in the same quarter, openly growing their AI engineering headcount. Same trick, same costume.

About 16,000 people got the same kind of news on the same Thursday, and the companies that delivered it spent more on AI engineers that quarter than on the people they let go. That isn't a contradiction. It's a strategy. The least the rest of us can do is call it by its real name.

---