The AI industry is in trouble.

The Information published a must-read investigation yesterdat. It says that half of the revenue backlog reported by four of the largest companies on earth comes from two startups fully dependent on the funding of those four companies.

Anthropic and OpenAI.

I. BUILDING CASTLES IN THE AIR

Here’s how it works.

Google invests $40B in Anthropic, $30B of which are compute credits. Anthropic commits to spending $200B on Google cloud. Google books that as revenue backlog and the stock goes up. Shareholders see the demand and approve $190B in capex.

Same loop with Amazon and Microsoft.

In total, Anthropic has committed $330B in cloud spending. How is Anthropic going to pay for all of that? Anthropic's run-rate revenue is ~$30-40B. Its commitments exceed a decade of that.

The same goes for OpenAI. OpenAI has committed over $688B. Their providers have collectively put $88B+ in equity into them.

Together, the two startups account for roughly half of $2 trillion in cloud backlog across four hyperscalers.

II. THEY’RE CONFIDENT IT WILL WORK OUT

Dario Amodei's defense is that they're “pretty confident” the revenue will come. So far, he’s right.

But Anthropic's own research shows a massive gap between what AI can theoretically do and what it's actually doing in the real world. That gap is room to grow, they say. Or perhaps it’s a diagnosis that AI capability is outrunning AI reliability.

A paper by Kapoor, Rabanser, and Narayanan found exactly that across 14 frontier models over 18 months: capability improved substantially whereas reliability moved much more slowly.

OpenAI's own data shows the median paying user uses "thinking features” 7x less than power users. People who already pay for ChatGPT, who already know how it works, just don't find it useful enough or reliable enough to push further.

Both are ambiguous signs. I wouldn’t be so confident.

III. WHAT HAPPENS NEXT

The AI industry needs one thing to survive: that the run-rate revenue they claim transforms into actual revenue.

Which will only happen if enterprise clients and individual consumers outside the loop find it worth paying for AI products at scale in the long run.

That’s the only metric that can tell us something about the near-term future of the AI industry.

Turns out that, beyond Anthropics’ and OpenAI’s own usage measurements, beyond independent reliability benchmarks, beyond deals, projected revenues, spending commitments, and infrastructure CapEx—beyond all of that, the data from the real world tells two conflicting stories.

Read below to know more.