Anthropic Revenue Surges to $65 Billion, Implements Invisible AI Text Watermarks
FlatRigor 5410 sources
The Story
Anthropic’s annualized revenue run rate surpassed $65 billion at the end of July, marking a sevenfold increase from a year ago. This figure is up from $47 billion in May and $9 billion at the end of last year. Investors anticipate Anthropic will conclude 2026 with revenues between $100 billion and $120 billion. The company also reported a preliminary revenue figure of $11.5 billion for the second quarter. Anthropic has confidentially filed IPO paperwork and is expected to go public ahead of rival OpenAI, possibly this fall, seeking a public valuation of $2 trillion or more. The company was last valued at $965 billion in late May. In a separate development, Anthropic is implementing invisible watermarks within Claude-generated text, with models launched on or after August 2 featuring an imperceptible watermark. The AI lab plans to extend support to older models and release a detection tool, stating the mark travels with copied-and-pasted text. Anthropic asserts its watermarking cannot be traced to specific organizations or individual users and is being added to meet commitments under the European Union's AI Act. This move has divided the tech community, with supporters arguing readers deserve to know when AI is used, while some users express concern about potential issues with clients, bosses, schools, or contracts. Freelance developer Vladislav Rajtmajer worried about a Claude marker in code sent to clients, though Anthropic claims a negligible effect on code. Public policy worker Arturo Villarroya feared professional and academic repercussions. Anthropic previously declined to sign an open letter supporting permissive policies for open-weight AI models, citing public safety concerns, a stance that contrasts with many industry peers. The company also faced temporary access disablement for two advanced models in June due to an export control directive citing national security authorities, and was blacklisted by the Pentagon earlier this year.
The Spread
Flat
What they agree on
- Anthropic's annualized revenue run rate reached $65 billion at the end of July.
- Anthropic is implementing invisible watermarks in text generated by its Claude AI models.
- The watermarking is intended to comply with the European Union's AI Act.
- Anthropic has filed confidential IPO paperwork and is expected to go public soon.
- The company's co-founders have pledged to donate a portion of their wealth to charity.
Where they split
- Article 1 (New York Post) focuses on the philanthropic potential of Anthropic and OpenAI IPOs, detailing specific donation figures and comparisons to historical philanthropy, which is not the primary focus of other articles.
- Article 7 (The American Conservative) provides a critical perspective on Anthropic's refusal to sign an open letter supporting open-weight AI models and its relationship with the AI industry, a framing not present in other reports.
- Article 5 (CNBC) includes details about Anthropic's temporary disablement of advanced models due to a government export control directive and its blacklisting by the Pentagon, which are not mentioned in all other articles.
- Article 2 (Business Insider) details the arguments for and against AI watermarks, including specific concerns from users and quotes from industry professionals, offering a deeper dive into the debate than other sources.
Source Perspectives
Left1 source