You’ve seen the slide decks. Your competitors are “leveraging AI.” Your board is asking why you aren’t. Consultants are circling like sharks, whispering that if you don’t act now, you’ll be disrupted into oblivion.
So you spend. You license tools. You hire AI strategy leads. You run pilots that generate impressive demos and underwhelming P&L impact.
And here’s what nobody on Wall Street wants to say out loud: there is zero evidence that AI is lifting profit margins for any company outside the tech sector. Zero.
Not a little evidence. Not “early signs.” Not “the runway is long.” Nothing.
Apollo’s own data shows it plainly — profit margins across the S&P 493 (everyone except the Magnificent 7) are flat. After hundreds of billions in AI spending, after every CFO has signed off on “digital transformation initiatives,” after every earnings call mentions “AI-powered” something — the margins haven’t moved.
But here’s where it gets uncomfortable. The analysts tell you to be patient. “These things take time,” they say. “The internet didn’t transform productivity overnight either.”
That’s a comforting story. It’s also potentially catastrophic advice.
Because there’s a simpler explanation staring everyone in the face: AI outside the tech sector isn’t a profit engine. It’s a cover charge. You pay it to stay in the building. Nobody gets richer from it except the bouncer.
Think about advertising. Every consumer brand spends billions on ads. Has advertising ever made the aggregate consumer goods sector more profitable? No — because when everyone advertises, the spending cancels out. It becomes table stakes. You don’t advertise to win. You advertise to not lose.
AI is following the exact same pattern outside tech. Your competitor implements an AI chatbot, so you implement one. They add AI-powered logistics optimization, so you do too. They cut 10% of customer service headcount, so you cut 10%. Net result? Everyone’s costs look the same. Everyone’s margins look the same. The only difference is that Nvidia, Microsoft, Google, and Amazon are now collecting a tax on every single transaction.
This is what evolutionary biologists call the Red Queen’s Race — you run as fast as you can just to stay in the same place. It’s not a failure of adoption. It’s not a timing issue. It’s the structural reality of a technology that diffuses evenly across competitors.
When everyone has the same tool, nobody has an advantage. The tool’s creator has the only advantage that matters.
And that’s the part that should make every non-tech CEO’s stomach drop. The entire valuation of AI companies — the trillions in market cap — rests on the assumption that AI will eventually expand margins across the broader economy. That promise is the foundation. Remove it, and the math doesn’t work.
But what if the margins never expand? What if AI is just… infrastructure? Like electricity, or the internet itself — necessary, transformative in how business gets done, but not a source of differential profit for the companies using it?
The internet didn’t make most companies more profitable either. It made some companies possible (Amazon, Google) and made every other company more efficient in ways that got competed away. The gains flowed to the platform, not the participants.
Sound familiar?
Now, the counterargument: “But AI will create entirely new business models!” Maybe. But that’s a different claim than “AI will improve your margins.” New business models benefit startups and disruptors, not incumbents pouring capital into defensive adoption. If you’re a regional bank spending $50M on AI infrastructure, you’re not building a new business model. You’re paying not to get killed.
And the token costs? The argument that compute will get cheaper and AI will become practically free? In the real world, early loss-leader pricing is already giving way to revenue-generating pricing. The subsidies are ending. The tax is going up.
So where does this leave you?
If you’re an investor, the implication is brutal: the AI companies you’re holding at 50x revenue are priced for a margin expansion story that may never arrive. And the non-tech companies you expect to benefit from AI adoption are spending money that will never come back.
If you’re a business leader, the honest move is to stop pretending your AI investments will show ROI in the traditional sense. They won’t. They’re insurance. They’re the cost of remaining competitive in a world where everyone else is also buying the same insurance. Budget them accordingly — as overhead, not as growth bets.
And if you’re one of the tech giants selling the picks and shovels? Enjoy it. You’ve convinced an entire economy to pay you a recurring fee for the privilege of standing still.
The AI revolution is real. It’s just not coming for your margins. It’s coming for your budget.
FAQ
Q: But didn't the internet also take years to show productivity gains? Why is AI different?
A: The internet did take time — but it also created genuinely new business models (e-commerce, search, social) that didn't exist before. AI, as currently deployed by non-tech firms, isn't creating new models. It's optimizing existing ones in ways that get competed away. Same tool, same advantage, nobody wins.
Q: So should non-tech companies just not invest in AI?
A: No — that's the trap. You have to invest, because if you don't and your competitors do, you lose. The point isn't to avoid spending. It's to stop expecting that spending to generate returns. Budget it as a cost of staying in business, not as a growth initiative.
Q: Does this mean AI companies like Nvidia and Microsoft are overvalued?
A: It means their valuations assume AI will eventually expand margins across the broader S&P 493. If that margin expansion never materializes — because AI is infrastructure, not a profit driver — then the growth story supporting those valuations has a serious hole in it. The Red Queen's Race benefits the arms dealer, but only as long as the war keeps escalating.