Anthropic’s Q2 Revenue Reported Above $11.5B — Profitable, and Ahead on the IPO Track – WOWTALE

Anthropic‘s second-quarter revenue exceeded $11.5 billion, Bloomberg reported on August 15. The figure did not come from the company; Bloomberg obtained internal documents, and the numbers are preliminary and could still change.

Buried in the same documents is the more consequential line: adjusted operating income turned positive. It is the first signal of a frontier AI lab posting an operating profit on a quarterly basis.

14x in a year

Revenue in the same quarter a year earlier was $787 million — a more than fourteenfold increase. The prior quarter, Q1 2026, came in at $4.73 billion, meaning the company more than doubled in three months. First-half revenue totals roughly $16.2 billion.

For scale: Anthropic’s full-year 2025 revenue was about $10 billion. In six months this year it earned 1.6 times what it made in all of last year.

“$11.5 billion” and “$40 billion” are not the same measure

OpenAI also produced a number this week. Its annualized run-rate revenue topped $40 billion, Bloomberg reported on August 13, roughly double where it stood at the end of 2025.

Setting those two figures side by side and declaring Anthropic ahead requires care, because they measure different things.

Anthropic’s $11.5 billion is quarterly revenue actually earned over three months. OpenAI’s $40 billion is annualized run-rate revenue — a recent revenue pace extrapolated across twelve months. It is not realized revenue; it carries an implicit “if this pace holds for a year.”

Multiply Anthropic’s quarter by four and you get roughly $46 billion, above OpenAI’s $40 billion. But at that moment Anthropic’s figure becomes the same extrapolation. Annualizing a company doubling every quarter understates it; annualizing one whose growth is flattening overstates it.

The only clean comparison is realized revenue. For full-year 2025, OpenAI booked $13 billion against Anthropic’s roughly $10 billion — OpenAI ahead. Then Anthropic alone booked $16.2 billion in the first half of 2026. OpenAI has not disclosed its first-half figure; only Q1 revenue of $5.7 billion is known, so a like-for-like comparison of the same period isn’t yet possible.

The gap is wider in profit than in revenue

Profitability is where the two diverge most. OpenAI burned $3.7 billion in cash during the first quarter against $5.7 billion in revenue — more than half of what it earned. Anthropic, over the same stretch, pushed operating income into the black, on an adjusted basis.

That qualifier matters. The documents do not specify what was adjusted, and the company has said nothing officially. Treatment of long-term compute commitments and stock-based compensation could change the picture considerably. Anthropic struck a five-gigawatt compute agreement with Amazon in April tied to a stated $100 billion, ten-year investment plan. Whether profitability survives fixed costs of that magnitude is the next test.

S-1s filed a week apart; two months later, diverging

Both companies started the listing process at nearly the same moment. Anthropic filed a confidential draft S-1 with the SEC on June 1; OpenAI followed on June 8. One week apart.

Two months on, the gap has opened. Anthropic CFO Krishna Rao is holding early investor meetings — the customary step before formal pricing — without discussing valuation. Some investors are working toward an October listing at a $2 trillion valuation, but that figure is their own arithmetic, not anything the company has put forward. Factoring in SEC review and bookrunner selection, an October listing implies a roadshow in September.

OpenAI is in a different position. It has not set a listing date and is reportedly weighing a wait until 2027. The deeper obstacle is structural: founded as a nonprofit and later capped-profit, it has yet to complete conversion into a standard Delaware for-profit corporation, a prerequisite for going public. Its March financing implied an $852 billion valuation — a private-market price, not an offering price.

What the reporting establishes isn’t a ranking. Two companies entered the same process in the same week, and the one that built a working profit model first is the one ready to face the market first. How public investors will price an AI lab remains unknown. Which lab goes first is becoming clear.