Since earlier coverage of Anthropic's AI self-coding disclosure, the company has now made a bigger move: filing confidentially for an IPO, according to TechCrunch reporting from June 4, 2026.

The Numbers

Annualized revenue crossed $47 billion in May 2026, per Anthropic's own figures. That's up from roughly $9 billion at the end of 2025. Five-fold growth in roughly five months is the kind of trajectory that attracts investor attention.

The company's $65 billion fundraise at a $965 billion valuation — announced last week — was reportedly oversubscribed, according to multiple investors who spoke to TechCrunch. Private demand is strong. Now Anthropic wants a piece of public capital too.

Why Go Public Now

Daniela Amodei, speaking Thursday at the Bloomberg Tech conference, was blunt about the motive: money.

"It's a really big upfront cost to train the models and to serve inference on them," she said. "My guess is that over time, the sort of core set of companies that are working to advance the frontier are just going to need access to capital, and I think the public market is very well suited to that."

This isn't a vanity listing — it's a capital-raising machine being pointed at the public markets because the compute bills are enormous.

The $1.25 Billion-Per-Month Compute Cost

Anthropic is spending $1.25 billion per month on compute — through its cloud partnerships with Google and Amazon Web Services, according to TechCrunch.

That's $15 billion a year in compute costs alone, from a company generating $47 billion in annualized revenue. This is not a startup burning cash recklessly — but it's also not a profit story yet. The margin math here deserves serious scrutiny, and neither Anthropic nor most press coverage of this IPO filing has rushed to provide it.

Amodei defended the decision NOT to build its own data centers — unlike OpenAI or xAI — saying Anthropic doesn't want to "overextend" by buying more compute than they can productively use. Renting compute at $1.25 billion a month carries its own risks.

The ROI Question

Uber has already gone on record saying not all of its AI spending has proven productive. That's a major enterprise client — exactly the category of customer Anthropic needs to sustain.

Amodei argued companies are still early in figuring out deployment. "As the business community gets more familiar with the tools, we're all going to learn together," she said.

But "we'll figure it out eventually" is a thin foundation for a near-trillion-dollar valuation. If enterprise customers start pulling back AI budgets — even slightly — that $47 billion annualized revenue number stops looking inevitable and starts looking fragile.

What the IPO Means

A confidential filing means Anthropic isn't required to go public with its full financials yet. That's a deliberate choice. It lets the company gauge market appetite without showing the world its actual cost structure, margins, and cash burn.

When the S-1 details become public, the real story emerges. Revenue growth is real. Whether this company is actually building toward profitability — or whether it's the most expensive science experiment in American corporate history — will be written in those numbers.

The Larger Picture

Most reporting on this treats the revenue trajectory as validation of the business model. Revenue growth at a company spending $15 billion a year on compute, while simultaneously calling for AI safety pauses, while also claiming its AI writes 80% of its own code, while filing for an IPO at a $965 billion valuation — that's a complicated set of claims to square.

Coverage has split along predictable lines: optimistic on the left, largely absent on the right. Neither framing serves readers well.

What's at Stake

If you hold index funds, Anthropic going public eventually lands in your portfolio. At a near-trillion-dollar valuation, the sustainability of enterprise AI spending matters.

The AI revenue numbers are real. Whether those numbers can sustain the current valuation is the question Daniela Amodei is betting billions that nobody scrutinizes too closely until it's too late.