The data says one thing, the pink slips say another — and both can be true at the same time.
The Summary
- Apollo's chief economist claims "zero evidence of AI-related job losses" based on ADP employment data, while at least a dozen major companies have cited AI in 2026 layoff announcements
- Tech CEOs rallying around this take are mostly building or selling AI infrastructure — not dealing with the downstream displacement
- The disconnect reveals a timing gap: companies are hiring AI implementers while simultaneously cutting the roles those implementations will replace
The Signal
Torsten Sløk's analysis isn't wrong. It's incomplete. The ADP National Employment Report shows private payrolls grew by 110,000 in April 2026. Companies are absolutely hiring AI implementation experts, data center engineers, and semiconductor specialists. Salaries for these roles are climbing. This is the buildout phase, and buildout phases create jobs.
But here's what aggregate employment data can't show: which specific jobs are disappearing while the overall number stays flat or grows. When a company cuts 500 customer service reps and hires 50 AI engineers, the net loss is 450 jobs. But those 50 new roles pay better, require different skills, and show up in the data as "AI-related hiring." Both things are happening. The economist sees hiring. The laid-off workers see pink slips citing "operational efficiency through AI implementation."
"The bottom line is that the AI spending boom is stoking both employment and inflation."
At least a dozen major companies have explicitly named AI as a factor in 2026 layoffs. Not as the only factor, but as a factor. That's new language. Five years ago, restructuring announcements blamed "market conditions" or "strategic realignment." Now they say the quiet part out loud. AI lets us do more with less. The jobs being created and the jobs being destroyed aren't the same jobs, and they're not employing the same people.
The enthusiasm from Box CEO Aaron Levie, Dell CEO Michael Dell, and White House AI Czar David Sacks makes sense when you map their incentives. Box sells enterprise software that needs AI integration. Dell sells the hardware running those data centers. Sacks is tasked with making America competitive in AI infrastructure. They're all optimistic about the buildout because they profit from the buildout. They're not managing call centers or accounting departments or legal doc review teams.
Key dynamics at play:
- Macro employment data lags micro displacement by 12-24 months
- High-skill AI jobs pay 3-5x what they replace, distorting wage data upward
- Companies hire for implementation first, automate second — we're still in phase one
The Implication
If you're trying to figure out where you fit in this, stop listening to economists with 30,000-foot views and start watching what companies actually do with their AI budgets after the implementation teams finish their work. The buildout is real. The displacement is also real. They're just happening in sequence, not simultaneously.
The people who will win in Web4 are those building agent infrastructure, owning the tokens that fund it, or developing skills that complement AI rather than compete with it. If your job is repeatable enough to describe in a prompt, start learning the thing that writes the prompts — or the thing that owns the system running them.