Anthropic revenue claim
A market commentator reported that Anthropic’s disclosed annualized revenue run rate topped $30 billion — a near tripling from a cited $9 billion figure in 2025 — a claim investors used to reprice AI bets. (Those revenue and growth claims were discussed on market podcasts and summarized in the media briefing). (youtube.com 1) (youtube.com 2)
Anthropic said on April 6 that its revenue run rate had passed $30 billion, up from about $9 billion at the end of 2025, and that single number was big enough to jolt the whole artificial intelligence trade. (anthropic.com) A revenue run rate is not booked yearly sales. It is a snapshot that takes recent revenue and stretches it over 12 months, like taking one busy restaurant weekend and asking what the year would look like if every weekend stayed that full. (sherwood.news) Anthropic did not drop the figure in a fundraising leak or a podcast aside. The company put it in its own announcement about a new compute deal with Google and Broadcom, which made the number look less like rumor and more like operating data tied to future capacity needs. (anthropic.com) The same post said more than 1,000 business customers are now spending over $1 million a year on an annualized basis, up from more than 500 customers less than two months earlier in Anthropic’s February funding announcement. (anthropic.com 1) (anthropic.com 2) That customer mix helps explain why investors reacted so hard. Consumer chatbots can be flashy, but enterprise contracts in software engineering, finance, and health care usually renew in bigger chunks and can justify much larger spending per account. (anthropic.com) (economictimes.indiatimes.com) Anthropic’s own milestones show how steep the climb has been. The company said it was at roughly $1 billion in run-rate revenue at the start of 2025, above $5 billion by August 2025, about $9 billion by the end of 2025, and above $30 billion by April 2026. (anthropic.com 1) (anthropic.com 2) That growth is expensive, because every extra customer query needs chips, power, and data center space. Anthropic said its expanded agreement gives it multiple gigawatts of next-generation Tensor Processing Unit capacity from Google and Broadcom starting in 2027 and running through 2031. (anthropic.com) Multiple gigawatts is utility-scale electricity, not normal software-company infrastructure. Anthropic said the first phase alone is expected to use more than 3.5 gigawatts of power, which shows why investors now talk about artificial intelligence leaders partly as energy and hardware stories. (anthropic.com) The valuation backdrop made the revenue claim even louder. Anthropic announced a $30 billion Series G funding round on February 12 at a $380 billion post-money valuation, so a jump from $9 billion to more than $30 billion in run rate within weeks gave investors a new way to argue that private-market pricing was not pure hype. (anthropic.com) Public-market investors read this through suppliers as much as through Anthropic itself. If one model company is already signing for years of custom chips and power, then companies tied to cloud capacity, networking, and semiconductor design suddenly look less like speculative picks and more like toll collectors on a crowded highway. (anthropic.com) (sherwood.news) The catch is that a run rate can rise faster than durable profits, because it says nothing by itself about margins, retention, or how much of the spending is being burned back into compute. But Anthropic’s April 6 disclosure still changed the conversation from “can these companies monetize?” to “how many companies can physically supply the machines they need?” (sherwood.news) (anthropic.com)