Anthropic has told investors it expects to post the first operating profit in its history this quarter — a projected $559 million on $10.9 billion in revenue — a milestone that, if it holds, would arrive roughly two years ahead of the company's own earlier guidance and reframe the central question hanging over the generative-AI boom: whether anyone building frontier models can actually make money.
According to financial projections reviewed by The Wall Street Journal, the company expects $10.9 billion in revenue for the June quarter, up 130% from $4.8 billion in Q1. The figures were shared with backers during a funding round that closed at a $965 billion post-money valuation, and they land days after Anthropic confidentially filed a draft S-1 with the SEC on June 1, positioning it to beat rival OpenAI to the public markets.
The growth curve is genuinely steep. Anthropic's annualized revenue run rate reached roughly $47 billion in May, up from about $9 billion at the start of the year — a fivefold jump in five months. The company's prior guidance, given to investors last summer, had not projected a full-year profit until 2028 at the earliest.
What's driving the numbers
The primary engine, by Anthropic's own account, is enterprise demand for its coding tools. Claude Code and the broader enterprise API business have pulled in corporate customers across industries, and the company has leaned into agentic workflows — software that can execute multi-step tasks rather than merely answer questions. CEO Dario Amodei captured the mood at a developer conference this month, saying revenue growth had become "too hard to handle."
Just as important to the profit line is unit economics. In the first quarter, Anthropic spent 71 cents on compute for every dollar of revenue; that ratio is projected to fall to 56 cents this quarter. The company runs primarily on Google and Amazon silicon rather than Nvidia, has made more conservative data-center commitments than OpenAI, and carries far fewer free consumer users — all of which trim the subsidy burden that weighs on competitors.
The bull case versus the bear case
For optimists, this is the moment the AI-lab business model stops being a faith-based exercise. A frontier developer is showing operating leverage at scale, with costs falling as a share of revenue precisely when revenue is exploding. If that trend continues, the trillion-dollar valuations attached to AI start to look like multiples on a real business rather than a story.
The skeptics are not buying it — at least not as stated. The $559 million figure includes model-training costs but excludes stock-based compensation, and because Anthropic is still private, it is not bound by public-company accounting standards. The numbers are unaudited internal projections prepared in a fundraising context.
The sharpest critique comes from technology writer Ed Zitron, who in an essay titled "Anthropic's 'Profitability' Swindle" argued the profit is "a result of accountancy rather than any improvements to its business model." Zitron points to compute-supply disclosures indicating Anthropic is paying around $1.25 billion a month for infrastructure — roughly $15 billion a year — but at a discounted ramp-up rate during the early months of a multi-year contract. Those discounted months, he notes, are the very quarter Anthropic is using to claim its first operating profit. Anthropic itself cautioned investors it may not sustain profitability across the full year, citing planned increases in computing-infrastructure spend.
That caveat is the crux. An operating profit that exists only while compute is temporarily cheap, and only after equity-funded chip commitments and stock comp are set aside, is a different animal from durable profitability. Both readings can be true at once: revenue growth is real and the cost trajectory is improving, and the headline profit number is flattered by timing and accounting choices that an audited filing would force into the open.
What to watch
The resolution is coming, and it has a name: the S-1. Anthropic's confidential filing will eventually become public ahead of a targeted October listing, and an SEC-reviewed prospectus answers the questions a fundraising deck does not. It will show audited financials, the full compute-cost schedule, the treatment of stock-based compensation, and whether the 56-cent compute ratio survives contact with disclosure rules. It will also reveal whether Q3 and Q4 — when the discounted compute period is set to lapse — stay in the black.
Until then, the $559 million is best understood as a projection with an asterisk: a credible signal of momentum, dressed in numbers that have not yet been tested by anyone obligated to check them.
"Anthropic's 'profitability' is a result of accountancy rather than any improvements to its business model."- Ed Zitron, Technology writer, Where's Your Ed At