Mixed Signals on Anthropic
Public narrative around Anthropic is conflicted: a venture podcast and YouTube commentary claim dramatic revenue growth and much lower training costs compared with rivals, while a separate report says Anthropic’s tender offer fell short of an estimated $6 billion. Those opposing signals — sensational growth claims versus a lukewarm tender outcome — underline how noisy and speculative market coverage can be for high‑growth AI firms. (youtube.com) (benzinga.com)
Anthropic just gave the market two completely different stories in the same week: on April 6 it said its revenue run rate had passed $30 billion, and by April 8 reports said its employee share sale still came in below the roughly $6 billion some investors had expected. Those are not actually the same thing. Revenue run rate is a speedometer that annualizes current sales, while a tender offer is a private stock sale where employees decide whether to cash out shares. Anthropic’s own April 6 announcement gave the bullish case in hard numbers: run-rate revenue up from about $9 billion at the end of 2025 to more than $30 billion, and more than 1,000 business customers now spending over $1 million a year, up from more than 500 in February. That growth story did not come out of nowhere. Anthropic said on February 12 that it had raised $30 billion in Series G funding at a $380 billion post-money valuation, after saying in September 2025 that run-rate revenue had already climbed from about $1 billion at the start of 2025 to more than $5 billion by August 2025. Then the hotter claims started spreading through podcasts and YouTube clips. Lenny Rachitsky’s podcast page for an April 5 episode with Anthropic growth head Amol Avasare says Anthropic scaled from $1 billion to over $19 billion in annual recurring revenue in 14 months, and Bloomberg reported on April 6 that Anthropic had already moved on to a $30 billion run rate. A separate set of commentary pushed an even flashier comparison on costs. SaaStr, citing financial disclosures discussed after reports from The Wall Street Journal, said Anthropic’s training costs could peak around $30 billion in the same period that OpenAI projects roughly four times more, but Anthropic has not published those side-by-side numbers itself. Now the colder signal. Bloomberg and Benzinga both reported that Anthropic completed the tender offer below the hoped-for $6 billion because employees sold fewer shares than some buyers wanted to purchase. That shortfall does not automatically mean weak demand for Anthropic. Bloomberg said some investors could not get as many shares as they planned because employees were reluctant sellers, which is closer to a crowded auction with too little inventory than a failed product launch. It also does not automatically prove the opposite, because private-market numbers are slippery. A run-rate figure can surge in one quarter and cool in the next, and a tender offer can miss a target simply because workers think an initial public offering later in 2026 could fetch a higher price. So the cleanest read is narrower than the internet version. Anthropic has documented very fast business growth and very large fundraising, while many of the loudest claims about beating rivals on revenue and spending still depend on media reports, leaked financials, or commentary built around them. That is why the week felt so contradictory. One set of numbers came straight from Anthropic’s newsroom, another came from secondary-market mechanics, and the gap between those two is exactly where hype, leaks, and guesswork rush in for private artificial intelligence companies.
