Here's a pattern Wall Street doesn't want to talk about: Amazon, Microsoft, and Google are systematically acquiring ownership of every major AI company while getting their money back immediately.

It's not venture capital. It's not strategic investment. It's something more elegant and more ruthless.

Here's the playbook:

Step 1: AI startup needs $1 billion to train models. They can't afford the infrastructure.

Step 2: Amazon "invests" $1 billion in exchange for equity.

Step 3: Startup immediately spends that $1 billion on AWS cloud services.

Step 4: Amazon gets its cash back. Keeps the equity forever. Cost basis: zero.

Rinse and repeat across the entire AI industry.

Let me show you the actual numbers from Anthropic:

Amazon, Google, and Microsoft collectively own 34-45% of Anthropic, currently valued at $380 billion. They invested roughly $25 billion total. But Anthropic has committed to spend $75-105 billion on their cloud infrastructure over the next few years.

Do the math: They're getting $75-105 billion back in infrastructure fees while keeping equity stakes worth $129-171 billion. Total value extracted from a single company: $204-276 billion.

The same structure exists with OpenAI (Microsoft owns ~27%), and it's spreading to every AI startup that wants to operate at scale.

Now here's where people get it wrong:

Some will say "this is just like Cisco in the 1990s doing vendor financing." Wrong.

Cisco gave loans. When dot-com companies went bust, Cisco held worthless IOUs. The stock dropped 86%.

The hyperscalers are buying equity, not extending credit. If the AI company fails, they already got their money back through infrastructure fees. If it succeeds, they own a massive stake. Heads they win, tails they break even.

Cisco financed telecom infrastructure builders (WorldCom, Global Crossing), not the platform winners. Cisco got zero equity in Google, Amazon, Facebook, or Netflix, the companies that actually won the internet.

Amazon, Microsoft, and Google own pieces of Anthropic, OpenAI, and every other AI company that could potentially disrupt them. The potential disruptors are owned by the incumbents.

Here's the other difference: Starting a web company in the 1990s cost $50,000. Google started in a garage. Facebook in a dorm room.

Building frontier AI today costs $100 million to $1 billion just to train a model, then $5-15 billion per year to run it at scale. You cannot build in a garage. There is no cheap alternative.

Only three companies can provide that infrastructure, and they're extracting equity in exchange for access.

What about regulation?

Each company takes minority stakes to avoid majority-control scrutiny. They can all claim they're "competing" with each other. To stop this, regulators would need to prove coordinated behavior across all three companies simultaneously. That's unprecedented.

By the time antitrust authorities figure this out in 5-10 years, the equity stakes will be locked in and the value already extracted.

So what do you do with this information?

If you own Amazon, Microsoft, or Google, you're not just betting on cloud growth. You own a portfolio of AI equity stakes acquired at near-zero cost. That's a hell of a hidden asset.

If you own pure-play AI stocks, understand who really controls your company's destiny. Check the cap table. See who owns 15%, 20%, 30%. Those aren't investors, they're landlords.

This isn't the dot-com bubble. This is Standard Oil's playbook perfected: control the infrastructure, extract equity from everyone who needs it, own the future at zero net cost.

Except better than Standard Oil, because three companies operating an oligopoly is harder to regulate than one monopoly.