Apollo Chief Economist: AI Returns Show Zero Spillover Beyond Tech Sector
nashnova research
Apollo's chief economist Torsten Slok says AI keeps lifting revenue per employee at the Magnificent Seven, but the remaining 493 S&P 500 companies see zero return — meaning AI is still an inside game for Big Tech, not yet an economy-wide productivity driver.
Big Tech is cashing in — what about everyone else?
Revenue per employee at the Magnificent Seven keeps rising; margins are expanding in step.
The remaining 493 S&P 500 companies show flat revenue per head and flat margins.
This means → AI's economic payoff is concentrated in a handful of giants, with no spillover to the broader corporate sector.
Why are mainstream companies pulling back?
Slok notes that mainstream firms are cutting AI spend, treating token spending — the cost of calling AI models — as a margin-compressing cost centre, not a growth engine.
VC-backed startups keep increasing token spend, but mainstream companies "don't need the latest models and aren't willing to pay for them."
In plain terms = for most companies, AI is an expense with no visible return — money goes out, revenue stays flat.
What is AI fundamentally missing?
Slok warns that AI has a fundamental gap in understanding accountability, ethical trade-offs, and trust relationships.
These factors drive reputational risk — and ultimately determine revenue growth and margins.
This means → AI is good at processing information, but the most valuable capabilities in business — judgment, trust, accountability — remain beyond its reach.
Is the market already voting with its feet?
Meta reported earnings first last week, announcing a more aggressive AI bet and raising capex guidance — and the market punished it.
Its share price fell; credit-risk indicators hit record highs.
This reflects a shift in investor sentiment: the focus is no longer on how much is spent on AI, but when the returns will materialise.
Is the $3.5 trillion AI bet at risk?
Gartner estimates that if AI revenue cannot support the spending, over $3.5 trillion in AI infrastructure investment by 2030 could face impairment risk.
Slok concludes: "As long as agentic AI and token spending remain confined to a small part of the economy, AI's macroeconomic impact will be extremely limited."
In plain terms = a lot of money has been spent, but if only a few companies profit, the reckoning is coming — this earnings season is the first real test.
市场有风险，内容仅供研究参考，不构成投资建议。