Mark your calendars, literally: May 20. That’s the date Meta has reportedly picked for its next round of mass layoffs, according to a Reuters exclusive picked up by Investing.com and AOL. First-wave target: roughly 8,000 employees, or about 10% of the global workforce. And that’s just round one.
Welcome back, Year of Efficiency
Zuckerberg already did this bit in 2022–23. That round — branded the “Year of Efficiency” in a memo that aged about as well as a crypto whitepaper — cut roughly 21,000 people. What’s coming in May, if Reuters’ sources are right, will be the biggest single cut since.
Except this time Meta isn’t even pretending to be efficient. The company just finished a quarter where it spent hundreds of billions on AI infrastructure, committing to capex that would have made a 2022 CFO cry. The layoffs aren’t because Meta is broke. They’re because Zuckerberg wants the savings to pay for more H100s.
The sources told Reuters the plan “could shift depending on how artificial intelligence capabilities develop in the coming months” — which is corporate-speak for we’ll cut more if the models get better, and probably also if they don’t.
The 20% problem
Here is the part nobody is reading closely enough: the May 20 wave is reportedly only about half of Meta’s total planned cuts for 2026. Per BusinessToday and the Honolulu Star-Advertiser, the company is aiming for 20% or more of its global headcount over the course of the year. A second wave is already slotted for H2, date and magnitude TBD.
Meta’s global employee count sits somewhere around 76,000 today. Twenty percent is roughly 15,000 people — cumulatively a bigger bloodletting than the 2022–23 round. It just comes with a different brand voice this time: no punchy memo, no CEO podcast tour, just a leak to Reuters and a date on the calendar.
What ‘AI-driven’ actually means at Meta
The thing to notice is what’s not being said. Zuckerberg is not claiming AI has replaced these 8,000 jobs. He’s not even claiming it will. What he’s claiming is that AI is a spending priority, and to fund it, other things — including people — have to go.
This is a meaningful semantic shift from 2024’s “AI is our copilot” era. In 2026, AI is your replacement at the budget meeting: you don’t need to lose your job to a model; you need to lose your job because your salary is now earmarked for GPU leases. The McKinsey slide has been updated accordingly.
The “crucible” epidemic
Meta joins a very crowded room. Just this week, Snap CEO Evan Spiegel told his 1,000 laid-off employees the company was facing a “crucible moment” and that AI was generating 65% of Snap’s new code. Oracle, a week earlier, cut 30,000 jobs in a 6 a.m. group email and redirected the savings to an AI data center buildout. The “crucible moment” is, apparently, now quarterly.
Meta’s particular innovation is the advance scheduling. Most companies still prefer the surprise-attack model — Tuesday morning, cold coffee, badge stops working mid-keystroke. Zuckerberg’s team chose a scheduled-release approach: here is the date, here is the number, prepare your spreadsheets. It is the layoff version of a product launch.
What to watch
A few things worth tracking between now and May 20: whether Meta gives affected teams any division-level preview (tip: they won’t), whether the second-half wave gets moved forward or back based on Q2 earnings, and whether any VPs who championed AI-led “efficiency” are themselves in the 8,000.
Because the one thing that’s genuinely efficient about an AI-pivot layoff round is that it doesn’t require the executives who ordered it to feel a thing. That’s the point. That’s always been the point.
Fourteen months ago, Zuckerberg said AI agents would do the work of a mid-level engineer by the end of 2025. Eight thousand mid-level engineers are about to find out what he meant.