The numbers are staggering — and the timing is impossible to ignore. As Big Tech pours an unprecedented $725 billion into AI infrastructure in 2026, tens of thousands of the very workers who built these companies are being shown the door.
The $725 Billion Breakdown
In Q1 2026 earnings calls, the four tech giants set records that analysts had never seen before. Here's how the spending stacks up:
- Amazon: ~$200 billion capex — almost entirely data centres and custom AI chips
- Microsoft: $190 billion — including $25 billion attributed to soaring memory and GPU component costs
- Alphabet (Google): $190 billion — Google Cloud AI infrastructure and quantum compute
- Meta: $125–145 billion — GPU clusters, custom silicon, and AI research labs
Combined, that's up 77% from 2025's already-record $410 billion. Analysts at several investment banks have called it the largest single-year capex surge in corporate history.
92,000 Jobs — Gone
While the GPU orders pile up, so do the pink slips:
- Meta: 8,000 layoffs in May 2026 — Zuckerberg told employees directly: the company chose GPUs over headcount
- Amazon: ~30,000 role reductions across five months
- Microsoft: ~125,000 through "voluntary" departures — a polite euphemism for buyout pressure
The roles hit hardest: content moderation, software testing, customer support, and mid-level engineering. The functions, notably, that AI tools have made cheapest to automate first.
The Skills Mismatch Nobody Talks About
Here's the cruel paradox buried in the data: while 92,000 tech workers have been let go, there are an estimated 275,000 open AI-related roles that companies cannot fill. The skills gap isn't a future problem — it's already here.
The laid-off software tester with five years of experience at Amazon isn't automatically qualified to be an AI prompt engineer or an ML infrastructure specialist. Retraining programmes exist, but they're slow, expensive, and often insufficient.
Is This the AI Labour Crisis?
Economists are split. Optimists argue this is no different from past industrial revolutions — jobs are destroyed in one column and created in another. Pessimists point out the pace is different this time: AI can learn new functions in weeks, not decades.
What's clear is that Big Tech has made its choice. The $725 billion is committed. The GPUs are ordered. And the workers who don't upskill fast enough may find themselves on the wrong side of the largest capital reallocation in tech history.
The question isn't whether AI is replacing jobs. The question is whether displaced workers will have enough time — and support — to catch up.
