Jack Dorsey cut 4,000 jobs at Block. Oracle's headcount fell by 21,000. Meta let go of 8,000 people. All three companies blamed the same thing: artificial intelligence. So why did a study of more than 21,000 U.S. firms find that the companies spending the most on AI are actually growing their payrolls?
The layoffs are real — and AI is the stated reason
In February 2026, Block co-founder Jack Dorsey announced he was cutting nearly half the company's workforce, more than 4,000 of its roughly 10,200 employees. He framed it as inevitable industry-wide change.
I believe the majority of companies will reach the same conclusion — Jack Dorsey, Block co-founder and CEO
Oracle's workforce dropped by about 21,000 people, roughly 13%, over fiscal 2026, even as the company spent $55.7 billion on AI data centers and disclosed in a regulatory filing that AI adoption had directly reduced headcount. Meta cut about 8,000 employees, around 10% of its workforce, while ramping up AI infrastructure spending.
| Company | Jobs Cut (2026) | AI Spending Signal |
|---|---|---|
| Block | 4,000+ (~40% of staff) | Cited "intelligence tools" as the driver |
| Oracle | ~21,000 (~13% of staff) | $55.7B capex on AI data centers, FY2026 |
| Meta | ~8,000 (~10% of staff) | Aggressive AI infrastructure and talent spend |
Models are quietly clearing a human bar
Part of the shift is capability. OpenAI's GDPval benchmark scores AI systems against real deliverables from 44 occupations across nine major industries, built with professionals averaging 14 years of experience. On the independently run GDPval-AA v2 version of that test, Claude Opus 5 posted the top score to date.
1,845 Claude Opus 5's GDPval-AA v2 score, vs. a 1,000 human baseline
GLM-5.3 and Grok 4.6 followed at 1,769 and 1,747, with GPT-5.6 Sol at 1,723 — every major frontier model now clears the human-professional baseline. A separate long-horizon test, Vending-Bench, asks models to run a simulated vending machine business for a year starting with $500. Claude Opus 5 finished with a reported balance over $11,000, and GPT-5.6 Sol topped $9,600, though comparable figures for some other current models were not independently confirmed at time of writing.
So why are heavy AI spenders hiring more?
A study from Ramp and workforce-data firm Revelio Labs, covering more than 21,000 U.S. companies, found that firms with the heaviest AI spending grew total headcount by roughly 10% and entry-level hiring by about 12% over the two years after adopting AI. Low-intensity adopters saw no significant change. The researchers were careful to note this shows correlation, not proof that AI caused the growth — heavy adopters tended to already be larger, faster-growing, and more technical companies.
The likely explanation, and the one raised across the layoff announcements themselves, is that AI is changing what companies can afford to attempt, not just what they can afford to keep doing. Cheaper execution can mean smaller teams doing the same work, or it can mean the same team taking on far more work than before. Which path a company takes may say more about its strategy than about AI itself.
Is your job built on execution AI can now match, or on judgment and ownership it still can't?
Source: layoffs, and hiring trends, 2026. Additional reporting and figures verified against CNBC, Fortune, Bloomberg, CNN, Yahoo Finance, Artificial Analysis (GDPval-AA v2 leaderboard), Andon Labs (Vending-Bench), and Ramp/Revelio Labs research.
