The data says companies are all in on AI. The data also says most of them are getting nothing back.

The Summary

The Signal

American companies are doubling down on AI spending. The Atlanta Fed survey shows investment has nearly doubled across US firms in recent months. But here's the uncomfortable truth: for most companies outside the Big Tech club, that spending is going into a black hole.

Apollo's Torsten Slok laid out the disparity in stark terms. AI is paying off for the hyperscalers who build the infrastructure. For the rest of the S&P 500, it's showing up as an expense line with no corresponding revenue bump or productivity gain. This isn't a temporary lag. It's a structural divide.

"AI's benefits are concentrated in Big Tech, highlighting a growing disparity in productivity and profitability across the broader market."

The perception problem is worse than the spending problem. Apollo's analysis found that companies see AI as a cost rather than a growth driver. When CFOs view something as cost, they eventually cut it. This could trigger a wave of reduced AI investments across mid-market firms, which would crater valuations for AI-focused companies that aren't named OpenAI or Anthropic.

Meanwhile, the job market implications are getting real. The Atlanta Fed data suggests rising AI investments will drive demand for tech services, but could also accelerate job reductions, especially for less-educated workers. This tracks with what we're seeing on the ground: enterprise software companies hiring prompt engineers while cutting customer success teams.

Key dynamics at play:

  • Big Tech captures AI productivity gains through infrastructure and models
  • Mid-market firms spend on AI tools without measurable return
  • Worker displacement concentrates among roles that don't require technical skills

Arm CEO Rene Haas called out the recent tech selloff as short-term thinking, arguing markets are missing AI's transformative potential. He's right that the long arc matters. But he's also selling chips into this buildout. The companies buying those chips and seeing no ROI have a different timeline.

The Implication

If you're building in this space, the split matters. Infrastructure plays and foundation model companies will keep attracting capital. Application layer companies serving mid-market customers need to show productivity gains fast, or that doubled spending becomes a rounding error in next year's budget cuts.

For workers, the message is clear: the skills gap isn't narrowing, it's accelerating. AI spending is concentrating in companies that already have technical talent. If you're not in that loop, you're not getting the productivity multiplier. You're getting the cost reduction.

Sources

Crypto Briefing