Google's $40B Anthropic Bet and the Strange Economics of AI's Circular Money

Bloomberg dropped a one-line scoop on April 24 that rattled the industry: Google is in talks to invest up to $40 billion more into Anthropic. On the surface, it’s just another headline-grabbing AI deal. Look closer, and it’s a perfect snapshot of the weirdest dynamic in tech right now — money isn’t really flowing anywhere. It’s spinning in tight circles between a handful of Big Tech firms and AI labs, and the loops keep getting bigger.

$40 Billion Is an Unreal Number, Even by 2026 Standards

Per Bloomberg Technology’s reporting, Google’s proposed top-up would reach $40 billion. The clip pulled over 13,000 views in a day, which tells you the number startled even a market that’s gotten numb to ten-figure rounds.

Google is already one of Anthropic’s largest backers. It has poured several billion dollars in since 2023, layered on top of a deep cloud partnership. Adding $40 billion on top of that would make this the largest single-company investment into one AI startup in history — by a wide margin.

For context: Microsoft’s cumulative investment into OpenAI sits around $13 billion. Google is reportedly considering roughly three times that — in a single round.

Anthropic Just Passed OpenAI on Revenue (Allegedly)

To make sense of the price tag, you need to know where Anthropic actually is right now. Recent industry analysis claims Anthropic has crossed $30 billion in annualized revenue, edging past OpenAI on that metric.

A year ago, Anthropic was still framed as “the OpenAI alternative.” Then Claude quietly became the default model in coding tools and enterprise stacks. Cursor, Windsurf, GitHub-adjacent IDEs, and a flood of B2B platforms run on Claude. On enterprise revenue, Anthropic appears to have leapfrogged the competition.

That’s why Google’s $40 billion isn’t a hedge anymore. Anthropic has stopped being a defensive bet and become a revenue-generating asset. Every additional Claude workload running on Google Cloud is a workload that doesn’t run on AWS or Azure.

Now Follow the Money

Here’s where it gets interesting. If Google sends $40 billion to Anthropic, a huge chunk of that cash flows right back to Google as cloud spend. Anthropic doesn’t own data centers. It rents them. Some of the money goes to Nvidia for GPUs. Some goes to AWS, since Anthropic also runs on Amazon’s cloud (and Amazon is itself a major Anthropic investor).

The industry has a name for this now: circular deals. The pattern looks like this — Big Tech invests in an AI lab, the lab spends that money on Big Tech’s cloud, Big Tech books the cloud revenue, and the inflated valuation justifies the next round. Recent analyses tracking roughly $242 billion in interlocking AI deals have argued this loop is starting to distort what “competition” even means in the cloud market.

On the Books: Revenue. In Reality: ?

The accounting question is the uncomfortable one. If a meaningful slice of Google Cloud’s revenue is just money paid by a company Google funded, is that revenue capturing real market demand — or itself?

Stack the deals on top of each other and the picture gets stranger. Oracle’s reported $300B cloud contract with OpenAI. Microsoft’s OpenAI equity stake. Nvidia investing in CoreWeave while CoreWeave buys Nvidia GPUs. Same pool of capital, sloshing back and forth between the same five logos.

Real demand exists, of course. Enterprises swiping cards for Claude and GPT API access is genuine revenue. But the share that’s organic versus the share that’s intra-ecosystem recycling is getting harder to separate — and nobody outside these companies has the books to tell.

Bubble or New Industrial Structure?

Investors are split, and the split is philosophical.

The bull case: capex on GPUs and data centers is real, durable infrastructure. The dotcom era left behind the fiber that powered the next two decades of internet economics. Today’s GPU buildout will do the same for AI.

The bear case: when revenue depends on a closed loop between five companies, the loop is also the failure mode. If OpenAI stumbles, Microsoft’s cloud growth stumbles. If Anthropic stumbles, Google Cloud and AWS take simultaneous hits. The concentration that makes the numbers go up also makes them brittle.

The Takeaway

$40 billion is a shocking number. The more shocking thing is that it’s barely shocking anymore. Two or three years ago, a $100M seed was a story. Now nine- and ten-figure rounds drop quarterly and barely move the conversation.

So which is it — capital expenditure laying the rails for the next two decades of computing, or accounting theater dressed up as a market? The honest answer might be that both are happening at once, in the same checks, between the same companies. That’s what makes this moment so hard to read.