Google Just Wrote Anthropic a $40 Billion Check

If you've been anywhere near tech news this past week, you've probably seen the number. Forty billion dollars. That's what Google is committing to put into Anthropic, the company behind Claude. ¹

It's the kind of headline that makes your eyes glaze over, because the dollar amounts flying around AI have stopped feeling like real money. So instead of getting lost in the figures, let's talk about what's actually happening, why it matters, and what it might mean for the AI tools you're starting to rely on at work.

What actually got announced

Here's the deal in plain English. Google is putting $10 billion into Anthropic right now, in cash, at a $350 billion valuation. ² Another $30 billion can follow if Anthropic hits certain performance milestones. Most of the broader package isn't really a check, though. It's compute capacity. Google Cloud is reserving 5 gigawatts of computing power on its TPU chips for Anthropic over the next five years. ³

To put 5 gigawatts in human terms, that's roughly the peak summer power draw of the entire San Francisco metro area. ¹ One company, training one family of AI models, eating that much electricity.

That last part is the real story here. The dollar figure got the headlines, but the compute commitment is what AI insiders are paying attention to.

Why compute is the new oil

Modern AI runs on three things: data, smart algorithms, and absurd amounts of computing power. Of those three, compute is now the choke point. There's only so much chip manufacturing capacity in the world, and only so much grid power available to run those chips. Every frontier AI company is in a race to lock down both.

Anthropic is in an unusual spot here. Most AI companies train their models on Nvidia GPUs (the famous chips that have made Nvidia one of the most valuable companies in the world). Anthropic is the only major AI lab that runs its training primarily on Google's TPUs instead. ² A TPU is Google's homegrown answer to Nvidia, and Google has been quietly chasing Nvidia in the AI chip business for years.

That's why this deal matters more than just the dollar figure. Google isn't simply an investor anymore. Google is Anthropic's primary infrastructure supplier, and now its biggest financial backer too.

For Anthropic, the deal solves an existential question: where do I get enough chips and electricity to train the next version of Claude? For Google, it locks in a flagship customer for its chip business and a strong AI hedge against OpenAI.

Anthropic isn't a small startup anymore

Tucked into the announcements was a number that probably matters more than $40 billion. Anthropic's annualized revenue rate passed $30 billion this month, up from about $9 billion at the end of 2025. ¹ That's roughly tripling in four months.

The growth isn't coming from chatbots. A big chunk of it is coming from Claude Code, the tool developers use to write software with Claude, plus the API that powers Claude inside apps like Cursor, Notion, and a long list of enterprise products. Anthropic has quietly become the AI company that real businesses pay real money for.

If you work in IT, your company probably has at least one team paying for Claude in some form. That's where the revenue is coming from.

The skeptical view

Not everyone is impressed. A growing chorus of analysts is calling these deals circular financing. ⁴ The argument goes like this. Google invests $40 billion in Anthropic. Anthropic uses the money to buy compute capacity from Google. The cash never really leaves Google's books, it just gets recorded as an investment instead of a customer purchase. Critics say it inflates both companies' apparent growth without much new value being created.

Defenders point out that this is just vendor financing dressed up for the AI era. ⁵ Banks did the same thing for railroads in the 1800s. Telcos did it during the broadband buildout of the late 1990s. The investment turns into customer purchases, the customer purchases turn into revenue, and as long as the underlying business is growing, the loop keeps spinning.

The honest answer is probably somewhere in the middle. AI's growth is real, since Anthropic wouldn't be tripling revenue if customers weren't paying. But the financial structure of the industry is unusual, and it's worth knowing that the headline investment numbers don't mean what they would in a normal sector.

What this means for working professionals

If you're an IT pro, business analyst, or someone who's recently started leaning on Claude or another AI tool to get work done, here's the practical takeaway.

First, Anthropic isn't going anywhere. The worry that your favorite AI tool might shut down in a year because it ran out of money is now off the table for the major labs. Google, Amazon, and others have collectively pledged enough capital to keep Anthropic well funded for years.

Second, expect Claude to keep getting better, faster. Five gigawatts of dedicated compute means Anthropic can train bigger models without rationing chips between customers and research. The next few quarters should bring some interesting capability jumps.

Third, the AI industry is consolidating. We're moving from "every Silicon Valley startup is making a foundation model" to "there are about four companies that matter, and they're all backed by the cloud giants." That's not necessarily bad news, but it's worth being aware of when you're picking tools or planning your skills development. The tools you learn today are very likely to still be around in five years, which wasn't always a safe bet in this industry.

The bigger picture

A few years ago, AI felt like a science experiment with uncertain commercial prospects. This week's news, sitting alongside similar mega deals from Amazon, Microsoft, and Nvidia, makes it clear that AI is now infrastructure. Like electricity or fiber optic cables, it's something that big stable companies are paying real money to build out, on the assumption that we'll all be using a lot more of it.

For the rest of us, that's a quiet bit of good news. The tools you're getting comfortable with aren't fads. They're going to be here, and they're going to keep improving. The right move isn't to panic about whether AI is going to replace you. The right move is to keep learning how to use it well, because the companies pouring billions into this aren't slowing down anytime soon.