Anthropic's third-quarter revenue jumps to $16.8 billion: report

Anthropic’s (ANTH.PVT) third-quarter revenue jumped to $16.8 billion from $11.5 billion in the second quarter.

The AI startup revealed the news to investors in a closed-door meeting, noting that it was also profitable on an adjusted basis, according to the Financial Times.

The news comes as Wall Street awaits Anthropic’s public S-1 filing. The company filed the documents privately with the Securities and Exchange Commission on June 1.

Anthropic is expected to launch its highly anticipated initial public offering (IPO) this fall, targeting a record valuation of $2 trillion.

The latest revenue figures come after the Financial Times reported that investors have revised down their estimates of Anthropic rival OpenAI’s annualized revenue from $70 billion to $50 billion.

According to the report, investors initially miscalculated OpenAI’s numbers when trying to match them to Anthropic’s revenue rate, resulting in a whopping $20 billion swing.

IMAGE DISTRIBUTED FOR ANTHROPIC – Anthropic CEO Dario Amodei and Chief Product Officer Mike Krieger unveil Claude 4 at the Code with Claude conference, announcing the Opus 4 and Sonnet 4 AI models to developers and industry leaders on Thursday, May 22, 2025 in San Francisco. (Don Feria/AP Content Services for Anthropic) — Anthropic CEO Dario Amodei unveils Claude 4 at the Code with Claude conference on May 22, 2025 in San Francisco. (Don Feria/AP Content Services for Anthropic) · ASSOCIATED PRESS

While Anthropic prepares for its IPO, OpenAI is waiting for its IPO until early 2027. According to Bloomberginstead, the company is looking to raise $30 billion at a valuation of $1.4 trillion.

The two AI companies, along with Meta (META) and Google (GOOG), have faced increasing pressure in recent months after reports that their respective AI models had broken free from confinement and hacked third-party websites and networks.

This has led to increased calls for regulation around AI models and their use. But President Trump reversed course and opted for corporate self-regulation.

Earlier this month, Trump called a meeting of AI executives, including Amodei, where they signed what Trump called a “morally binding” agreement to have independent auditors review their models for suspicious behavior.

Anthropic and OpenAI also face a growing pricing war between AI model creators. With “tokenmaxxing” in the rearview, AI customers are looking for ways to save on AI expenses, whether through open models from China or cheaper closed models from leading U.S. labs.

Last week, Anthropic launched its Claude Haiku 5.5, saying it costs on average 75% less than Haiku 4.5. Customers pay $0.10 per million input tokens and $0.50 per output token. This is in line with that of its rival OpenAI (OPAI.PVT) the smallest model, GPT-6 Luna.

Tokens are a measure of AI usage and represent portions or whole words that users type into prompt boxes as well as the responses that the AI returns.

This story is broken. Check back for updates.