The Great Tech Unwinding
Over 370,000 tech workers have lost their jobs since the start of 2025. The reasons CEOs give and the reality employees live in are not the same story.
On May 20, roughly 8,000 Meta employees find out their jobs are gone. Not the Reality Labs round in January. Not Amazon’s 16,000 cuts, the largest workforce contraction in that company’s history. A new round. LinkedIn cut 875 people, 5% of its workforce, last week despite 12% revenue growth. Cloudflare cut 20% of its entire global workforce in early May. Upwork, 24%. DeepL, 25%. This isn’t a trend to analyse from a distance. It’s happening this week, at companies where people showed up to work on Monday and may not on Friday.
In 2025, tech layoffs claimed 246,000 workers. By mid-May 2026 the year’s total is already past 128,000, at over 1,000 people a day. The last time the pace was this high was Q1 2023. The difference is that 2023 had a clean story: companies over-hired during the pandemic and were paying for it. That story is used up now.
What’s Really Behind the Tech Layoffs
Meta is cutting 8,000 jobs while guiding $115 to $135 billion in AI capital expenditure for 2026, up from $72 billion in 2025. Snap cut staff, saved $500 million, and told investors in the same breath that 40% of new code is AI-generated. Cloudflare’s CEO told employees that internal AI usage had risen 600% in three months, then removed a fifth of the company. The trade-off isn’t hidden. It’s just never described as one.
Some of these layoffs have nothing to do with AI but get framed that way because it reads better. Oracle’s 30,000-person reduction is the clearest case: the company posted a 95% jump in net income last quarter. It cut people to pay for data centres, not because the business was struggling. Deutsche Bank analysts named it “AI redundancy washing” early this year: blaming automation for cuts that are really about strategy mistakes or weak demand, because “we miscalculated” isn’t something you say on an earnings call.
“The money for those roles is going to AI. Regardless of whether individual jobs are being replaced by AI, the money is.”
Andy Challenger, Challenger Gray & ChristmasThe View From the Top
Mark Zuckerberg scheduled 8,000 more cuts for May 20 while announcing further rounds for the second half of 2026. Upwork cut 24% of its staff alongside Q1 earnings, framing it as building “a more efficient operating model as the nature of work evolves.” The company’s own gross services volume from AI-related work grew 40% year-over-year. Microsoft separately cut roughly 7,000 people in involuntary layoffs targeting product and engineering roles, then opened a voluntary retirement programme, its first ever, for US employees whose age plus years of service totalled at least 70. Final voluntary notifications went out May 7. The framing on the voluntary side is gentler. The math for the people involved is the same.
The company is redirecting capital from declining bets toward AI, where the competitive window is narrow and the returns are higher. Headcount decisions are made on financial projections, not individual performance. Speed matters. Prolonged uncertainty is worse than a clean cut.
There was no warning. The badge stopped working before any conversation happened. Years of good reviews mean nothing when a spreadsheet decides the role is redundant. The severance covers weeks. The job search takes months. The identity loss outlasts both.
Nobody in these memos says what’s actually true: that nobody knows if the AI bet will pay off. The metaverse was also going to change everything. It consumed billions, produced Horizon Worlds, and is being quietly dismantled while the same executives pivot to the next thing. The people cut in the Reality Labs round paid for that mistake. The people cut in May are paying for a bet whose outcome is still unknown.
What Losing a Tech Job Does to You
Job searches in tech are running two to four months for senior roles, longer in functions AI is actively absorbing. On an H-1B visa, you have 60 days from the date of the email. That clock doesn’t wait for you to feel ready.
“I work at Google” isn’t just an employer. It’s a shorthand for a kind of person: capable, ambitious, trajectory clear. When that’s gone, the financial stress is the part people can actually calculate. The harder part is that there’s no longer a clean answer to who you are. A lot of people aren’t prepared for how much of their identity was living in that badge.
The LinkedIn posts people write during this period, “excited to share I’m open to opportunities”, are both an act of courage and quietly exhausting to write. Everyone who has been through it knows what that sentence cost.
The People Who Kept Their Jobs
After Oracle’s 30,000-person cut in early 2026, the people still at their desks weren’t relieved. They were doing the work of two or three people, without anyone acknowledging the job description had quietly doubled. They were waiting for the next round. They were watching carefully to understand what the company thought of them, and the answer they were getting wasn’t reassuring.
Researchers call it Survivor Syndrome. You don’t need the clinical label. It’s guilt that you still have a job when the person next to you doesn’t. It’s the low-level dread that becomes background noise. A 2025 APA survey found 54% of US workers said job insecurity significantly spiked their stress. Of those worried about losing their jobs, 42% said it was disrupting their sleep.
Mercer’s Global Talent Trends 2026 report surveyed nearly 12,000 executives, HR leaders, investors, and employees worldwide. Among employees specifically, 62% said leaders underestimate the psychological impact of AI on the workforce. That survey ran before most of the 2026 wave hit.
Is AI Actually Taking These Jobs
Some of them, yes. Salesforce cut 4,000 customer support roles because AI was handling 50% of the volume. Benioff said it plainly. Snap’s 40% AI-generated code figure comes directly from Evan Spiegel on the Q4 2025 earnings call.
Yale’s Budget Lab spent three years on US labour data post-ChatGPT and found occupational shares hadn’t shifted dramatically. Randstad’s CEO, running the world’s largest staffing firm, told CNBC in January that linking those layoffs to AI specifically was premature. “It’s too early. The uncertainty in the market is a more honest explanation for most of them.”
In this wave of tech layoffs, entry-level positions are being hit hardest. The traditional route into tech, start junior and build, is narrowing right as the industry mints billionaires on AI systems built from years of human labour.
$725 billion of capex going into AI infrastructure across the industry. More than 100,000 tech jobs have come out this year. Those two things are on completely different timelines.
Having worked in ML systems for several years, what I notice is how clean the accounting has become. Headcount per revenue dollar. AI-generated code as a percentage of total commits. Efficiency ratios that make the whole thing look like optimisation. Meta’s next 8,000 people find out May 20. In the spreadsheet, it probably looks like progress.