AI-linked layoffs are no longer theoretical. Based on current 2026 US trends, and using a deliberately conservative model of roughly half recent monthly acceleration, around 300,000 US workers could be displaced in calendar 2026.
That represents approximately $70 billion per annum in fully burdened employment cost removed — not just wages, but salary, benefits, payroll burden, software, office space, HR, management, and infrastructure.
Where the cuts are likely to land – Where the AI $buildout budget comes from
| Segment | Estimated share |
|---|---|
| Tech / developers / QA / product | 20% |
| Corporate / HR / finance / marketing / admin | 25% |
| Support / back office / clerical | 35% |
| Operations / logistics / field admin | 15% |
| Management layers | 5% |
Why some segments matter more
A $55K support or clerical employee may actually cost closer to $115K/year once software, office, supervision, HR, and compliance are included.
A $160K tech employee in California or another major hub may represent $300K+/year in real organizational cost before stock compensation.
What AI is really removing
AI is increasingly targeting full cost per seat, not just wages:
- software licences
- office costs
- HR
- management layers
- recruiting
- training
- compliance
Conservative 2026 projection
Using only about half of recent monthly growth acceleration:
~300,000 displaced US workers
~$70 billion annual burdened cost removed
Three-year view
One year matters. Three years shows structural scale:
~900,000 worker-years displaced
~$210 billion removed over three years
Stock compensation caveat
This model excludes major stock-based compensation, largely concentrated in technology. Including that could push the annual figure closer to $75–80 billion.
Bottom line
AI is beginning to reprice which jobs are economically viable, with the greatest near-term pressure on support, clerical, corporate, and selected technology roles.