There is a particular number that landed on June 1 that the press releases have not yet caught up to. TechTimes’ Monday morning print put cumulative US tech layoffs for 2026 at 148,092 across 354 events, an average of 981 jobs a day — 46% above 2025’s 674-a-day baseline. If the curve holds, the sector ends the year north of 370,000. The headline is the cumulative number. The real story is the cohort it disproportionately ate.
The 2026 AI Index, released in April by the Stanford Institute for Human-Centered AI, found that employment for software developers aged 22 to 25 has fallen nearly 20% since 2024, even as headcount among their older colleagues kept growing. The decline, the report’s authors note, is concentrated in roles involving boilerplate coding, scripted testing, and routine bug fixes — the exact tasks that GPT-4-class autocomplete and Copilot-class agents now handle at near-junior-level reliability.
Both numbers are real. Both are from the same month. They describe the same arrow.
What 148,092 actually means
The TrueUp tracker prints a slightly different cut — 148,173 displaced across 355 events by midnight on May 31, or 975/day — but the two numbers triangulate. The pace is what matters. 2025 averaged 674 tech cuts a day. The 2026 line is running 46% above that, every day, for five months. There has been no Q2 deceleration; the May print was the worst single month since October 2023.
The cumulative composition reads like a who’s-who of the AI capex round. Oracle’s 30,000 in January was the single largest event and explicitly tied to data-center reorientation. Microsoft cut 6,000 in May on the first day of FY26, mostly software engineers in Washington state, with the AI and Copilot teams explicitly exempt from the program. Meta took out 8,000 on May 20 — 10% of staff, internal memo framing the move as offsetting AI investment. Intuit announced 3,000 the same day, 17% of global headcount, with locations in Reno and Woodland Hills shutting down. Wix took 1,000, 20%, on May 28. Cisco cut 4,000 on May 13 — the same day it reported a record $15.8B quarter.
Challenger Gray’s cause attribution puts AI at 49,135 of the YTD total — roughly 25–26% of cuts in March and April, the leading single cause for two consecutive months. “Market and economic conditions” still claim a larger cumulative share at 53,000+. But the AI line is the only one accelerating. Through the back half of 2025 it polled in single digits. By spring 2026 it polled at a quarter of every cut.
What Stanford’s chart actually says
The Index’s economy chapter is more careful than the layoff tracker headlines. Its claim is narrow and specific: between 2024 and the most recent quarterly cut, payroll employment among 22-to-25-year-old software developers fell roughly 20%. Employment among developers aged 26 and up grew over the same window. The collapse is not a software industry collapse. It is a junior-tier collapse inside a still-growing field.
The mechanism the authors point to is unromantic. The tasks that defined the junior-engineer role in 2022 — write the boilerplate, run the scripted tests, take the first pass at the routine bug, escalate when stuck — are precisely the tasks that the current generation of code-completion and code-agent tools now ships at near-junior reliability. The senior who used to delegate that work to a junior no longer needs to. The junior who needed those rotations to compound into a mid-level engineer no longer gets them.
This is the loop the Index does not have words for but the cuts do. The companies cutting headcount are simultaneously cutting the rungs of the ladder that produced the seniors they want to keep. The cohort that would have been senior in 2030 is the cohort whose entry-level seat just got eaten. The 35-year-old engineer at Microsoft, Meta, or Wix is, structurally, safe. The 23-year-old new hire in those same companies is the one the slide deck is referring to.
Stanford’s employer-survey component adds the forward-looking piece. One in three employers surveyed expects further workforce reductions in the next 12 months. The reductions they describe are not framed as “AI replaced the worker.” They are framed as “we don’t need to hire the worker we would have hired.” That is a different kind of disappearance — quieter, slower, and invisible to the layoff tracker.
The framing arbitrage Cisco proved is now permanent
The reason every June 2026 press release sounds the same is that Cisco’s 13% pop on May 13 — record revenue and a 4,000-person cut on the same day, with “AI realignment” attached — proved that the market pays a premium for the AI vocabulary. After that print, no CFO with a layoff to announce had a reason not to attach the AI label. The alternative now costs basis points.
What MIT’s Paul Osterman called “AI washing” in his May 31 Fortune interview — and what LostJobs covered yesterday — is the cover-story half of the same arrow. The junior-tier collapse Stanford documented is the production half. Both are running at once. One is what companies say. The other is what hiring does.
The Index makes the second one harder to argue with than the first. Press releases can be reframed. The 20% line cannot.
The cohort that won’t get its rotations back
The thing the layoff tracker misses is the hire that doesn’t happen. In a normal year, US tech adds roughly 100,000 entry-level developer positions. Stanford’s number implies that the 2026 hiring class is on track for somewhere closer to 80,000 — the layoffs are visible, the unfilled junior reqs are not. Over a five-year window, the missing 20,000-per-year cohort compounds. By 2030, the mid-level layer of every major tech employer is structurally thinner than the org chart anticipates, and the senior bench gets longer to refill.
This is the structural argument the AI-jobs apocalypse forecast missed and the AI-washing critique cannot fully absorb. Even if AI displaces zero current workers — and it has, demonstrably, displaced more than zero — the not-hiring tail rewires the labour market on a longer timeline than any quarterly press release. The 22-year-old who can’t get the first job in 2026 doesn’t become a 27-year-old senior engineer in 2031.
The Index puts a number on it. The June 1 tracker puts a date on it. The two together are the closest thing this cycle has to a junior-tier autopsy in real time.
What changes by Q3
Three calendars converge on Q3. Salesforce reports on June 4; its “AI-rebalancing” line is now in its fourth consecutive quarter, and analysts are demanding specific FTE numbers rather than the vocabulary. Atlassian’s late-July guide will either restate the 10% AI-era target or walk it back. Block’s H1 print, expected August, has to demonstrate that Dorsey’s 40% cut actually produced the margin he promised.
The second calendar is Stanford’s. The HAI team has confirmed that the September 2026 AI Index update will include the first quarterly refresh of the 22-25 employment series. If the cohort line keeps falling, the 20% number becomes 25%. If it stabilises, the most likely explanation is that hiring froze entirely. Neither is a recovery.
The third calendar is the regulatory one. California’s WARN-AI executive order from May 21 requires the first AI-tagged displacement filings in early September. The companies that voluntarily AI-labelled their spring cuts will have a clean filing path. Companies that quietly called the same cut “rightsizing” will not. The asymmetry shapes which vocabulary survives into Q4 and which one gets retired with the 10-K.
What June 1 made clear is that the two arrows that have been pointing at each other for eighteen months are now the same arrow. The press releases describe AI rebalancing. The Stanford economy chapter describes a junior-tier collapse. The Challenger attribution puts AI at the top of the cause column for two months running. The TrueUp tracker pegs the pace at 46% above last year and accelerating. None of these is the headline. The cohort it adds up to is.
The 23-year-old reading these numbers in June 2026 is the data point. The 33-year-old still writing code at Microsoft is, for now, not. The gap between those two sentences is what the rest of this decade will be about.