Google's $40B Anthropic Bet: The Most Expensive Insurance Policy in AI

When I saw the headline last Friday, I thought I’d misread a decimal point.

Google investing in Anthropic. Up to $40 billion.

Forty. Billion. Dollars.

$10 billion cash upfront, at a $380 billion valuation. If Anthropic hits performance milestones, Google adds another $30 billion. Plus 5 gigawatts of TPU compute, coming online from 2027.

5 gigawatts is roughly the output of five nuclear power plants.

Google Isn’t Alone

Pull up Anthropic’s funding history from the past six months and the picture gets surreal:

Amazon: $5B cash, up to $25B, plus 5GW of Trainium compute. Nvidia: up to $10B, 1GW GPU. Microsoft: up to $5B, with Anthropic committing $30B in Azure purchases. Google: $10B cash, up to $40B, 5GW TPU.

Combined compute commitments: over 11 gigawatts. About ten nuclear plants.

And here’s the twist: Microsoft is also OpenAI’s biggest backer. Yet it invested in Anthropic too.

This Isn’t Investment. It’s Insurance.

Traditional VC logic: I believe in you, I invest, you succeed, I profit.

Anthropic’s funding logic is completely different.

Amazon invests because Anthropic is one of AWS’s biggest AI customers. Google invests because Anthropic buys Google Cloud and TPU chips. Microsoft invests because Anthropic commits to $30B in Azure purchases. Nvidia invests because… well, Nvidia invests in everyone.

See the pattern? It’s a loop: I give you money, you buy my compute, my cloud business grows, I give you more money.

These giants aren’t investing in AI’s future. They’re buying insurance for their cloud and chip businesses. Anthropic’s ARR has crossed $30 billion - up from $9 billion at the end of 2025. More than tripled in a year.

At that growth rate, no cloud provider can afford not to invest. Because if you don’t, Anthropic takes its compute orders elsewhere.

The Two-Power Standoff

Analysts are saying the AI landscape has shifted from a “Big Three” to a two-power standoff: Team Anthropic vs. Team OpenAI.

The data backs this up. As of March 2026, OpenAI holds 35.2% market share, Anthropic 30.6%. The gap has narrowed from nearly 13 percentage points in early 2025 to just 4.5.

Claude Code deserves much of the credit for Anthropic’s surge. The AI coding tool has become near-standard among Silicon Valley engineers, driving explosive enterprise subscription growth.

But here’s the irony: Google is investing $40B in a company that’s eating into its own market. Google executives have reportedly expressed concern about Anthropic’s rapid rise in AI coding.

My Take

$40 billion for one company sounds insane. But reframe it as “the deposit cloud giants pay to lock in their biggest customer,” and it makes perfect sense.

The real story isn’t the dollar amount - it’s the industry logic it reveals. In the AI era, compute is oil, and companies like Anthropic are the biggest consumers. Whoever locks in their compute orders wins the second half of the cloud wars.

As for Anthropic itself - taking money and compute from four competing giants while maintaining independence and eyeing an IPO? One word: impressive.

But there’s a nagging question: when your shareholder list includes Amazon, Google, Microsoft, and Nvidia simultaneously, can you really stay independent?

That answer might have to wait until after the IPO.