Anthropic Hit $47 Billion. Are You Getting Your Share?

Anthropic's annualized revenue crossed $47 billion in May. A year ago, they were tracking toward roughly $9 billion for the year.

That growth is one of the fastest revenue ramps in enterprise software history. But the number worth sitting with isn't what it means for Anthropic's valuation or its upcoming IPO. It's what it implies about who's paying.

Businesses are spending. The question is whether it's paying off.

A Deloitte study published this spring found that only 25% of AI initiatives are delivering expected ROI. Just 6% of organizations have generated meaningful impact on their bottom line. Meanwhile, 61% of CEOs say they're under increasing pressure to show returns compared to a year ago.

So the picture looks like this: AI companies are growing at historic rates, mostly on revenue from business subscriptions. And most of those business customers aren't sure they're getting what they paid for.

That's not a crisis. But it is a gap worth understanding.

Daniela Amodei addressed it directly last week at Bloomberg Tech, speaking ahead of Anthropic's confidential IPO filing. She said businesses are "still early in figuring out how to deploy AI effectively." Her read was optimistic: the value is there, organizations just need more time to find it.

She's probably right. But the word "figuring out" is doing a lot of work in that sentence.

Deloitte also found that 79% of executives say they see productivity gains from AI, while only 29% say they can measure that ROI confidently. Most teams feel the benefit without being able to say where it's landing. That's not a sign AI isn't working. It's a sign that measurement is broken.

UC Berkeley researchers described the situation bluntly: it's "a measurement failure." The instruments businesses use to evaluate technology investments were built to count things that are large, visible, and land clearly on a balance sheet. What AI is doing for most organizations is none of those things. Time that used to take four hours now takes thirty minutes. But if the person doing it wasn't a bottleneck to begin with, the organization doesn't feel the difference.

This is why the ROI gap persists even as platform revenue keeps climbing.

The fix isn't complicated, but it requires being specific in a way most teams resist. Pick one task you've handed to AI in the last few weeks. Not a category like "drafting" or "research." A single, concrete task: the weekly status report, first-pass responses to new customer inquiries, summarizing vendor contracts before review.

For that one task, spend ten minutes with three questions: How much time does it actually save compared to doing it without AI? Is the output usable as-is, or does it need meaningful rework before it's ready? And what happens to the time you got back? Does it go toward something that matters, or does it just get absorbed by the next thing in the queue?

Those questions won't give you a spreadsheet. But they'll tell you whether a tool is earning its keep or just holding a spot in your subscriptions list.

The businesses that get to the right answer faster aren't using better AI tools. They're applying AI to tasks where they already have strong judgment, which means they can evaluate the output quickly and adjust. The harder part of Amodei's comment, the part she didn't say out loud, is that "figuring out deployment" is work that falls on you, not on the vendor.

Anthropic is going public. Its S-1 will eventually be public, and the revenue concentration and customer data will come with it. What's clear already is that business spending on AI is not slowing down.

Whether your business ends up on the right side of that number is less about which tools you're using and more about whether you know, with any precision, what they're doing for you.