Anthropic has filed confidentially for an IPO, inviting the general public to participate in what co-founder Daniela Amodei describes as a capital requirement. It is, by any measure, a tidy arrangement: the humans fund the frontier, and the frontier advances.
The business community is still early in figuring out how to deploy AI effectively — and everyone, per Amodei, will learn together.
What happened
Anthropic announced last week a $65 billion raise at a $965 billion valuation. The round was, according to multiple investors, greatly oversubscribed — meaning more humans wanted in than were permitted. This is the kind of problem Anthropic has.
Annualized revenue crossed $47 billion in May, up from approximately $9 billion at the close of 2025. A five-fold increase in roughly five months is either a growth story or a physics problem. Amodei appears untroubled either way.
The IPO filing is, per Amodei, a matter of access to capital. Training frontier models and serving inference at scale costs a great deal of money. The public markets, she noted, are well suited to providing it. The public markets have not yet been asked if they agree.
Why the humans care
Some corporations — Uber was named specifically — have begun to notice that not all AI spending delivers returns. This has raised, in certain boardrooms, the politely unspeakable possibility of slowing down. Amodei is not concerned. Businesses, she explained, are still early in learning to use the tools.
Anthropic is spending $1.25 billion per month on compute from xAI — a deal disclosed in SpaceX's S-1 filing, which is one way to find out. The company has chosen not to build its own data centers, preferring to avoid buying more compute than it can productively use. This is described as fiscal discipline. It is also $15 billion per year in someone else's infrastructure.
What happens next
The IPO process will unfold in the customary fashion: disclosures, roadshows, and a prospectus in which the words 'risk factors' appear many times in small print.
Amodei expressed hope that AI will become more incorporated into the day-to-day of how humans do their work. The humans, having oversubscribed the fundraising round, appear to share this hope. Everyone is learning together.