The AI Industry’s Dirty Secret: They’re Counting the Same Dollar Three Times

You’ve probably noticed the headlines: AI is getting cheaper. API prices are plummeting. The future is finally affordable. But here’s the thing — it’s a lie. And the people selling you that lie are hoping you don’t look too closely at the math.

If you’re a developer building on OpenAI or Anthropic, you’ve felt the sting. The per-token cost might be lower, but somehow your monthly bills are higher. You’re not alone. You’re not bad at reading spreadsheets. The game is rigged.

The AI industry isn’t crashing because of hallucinations. It’s crashing because of accounting.

Let me show you what I mean. Last quarter, Google reported $53.4 billion in revenue from its deals with Anthropic. Impressive, right? But follow one Anthropic dollar through Google’s earnings release. It shows up in three different places: AI business revenue, chip business revenue, and AWS segment revenue. That’s not growth — that’s a shell game.

This isn’t a bug. It’s a feature of the current AI bubble. Cloud providers, AI startups, and chip makers are all counting the same transaction in their own revenue lines. The result? A market that looks like it’s exploding, when in reality, the underlying demand is being artificially inflated by financial engineering.

And the consequences are real. Every developer building critical infrastructure on subsidized API pricing is building on quicksand. When the subsidies stop — and they will — the unit economics will collapse. The companies that survive will be the ones that own the hardware and the cloud, not the ones selling the models.

I’ve watched this pattern before. In the dot-com era, companies counted the same ad dollar across multiple subsidiaries. In the crypto boom, exchanges counted the same trading volume across different platforms. Now, AI is doing the same thing — only with more zeros and a glossier narrative.

So what’s the real story? The AI bubble isn’t deflating because of technical limits. It’s not because of hallucinations. It’s because the entire industry is built on a pyramid of recycled revenue. When the music stops, the only survivors will be the monopolists: Google, Amazon, and Microsoft. Everyone else? They’re renting space on a sandcastle that’s about to be washed away.

The next time you see a headline about AI revenue surging, ask yourself: how many times is that dollar being counted? And then ask yourself: are you building your future on a lie?

FAQ

Q: If API prices are dropping, how can it be a bubble?

A: Because token consumption is exploding. The per-token cost is lower, but the total number of tokens used per task is much higher. Companies are subsidizing usage to create the illusion of adoption, while the real cost to the user doesn't actually decrease.

Q: What should I do as a developer or investor?

A: Don't build your entire stack on a single subsidized API. Hedge your bets with open-source models and local inference. For investors, look at unit economics, not top-line revenue. If the same dollar is counted three times, the growth is fake.

Q: Isn't this just the normal tech cycle of boom and bust?

A: No. This is different because the revenue is artificially inflated by triple-counting. In a normal cycle, the revenue is real but overvalued. Here, the revenue itself is an illusion. When the accounting stops, the entire house of cards collapses — not just valuations.

📎 Source: View Source