Updated version 1 - Sources say OpenAI's annual recurring revenue is approaching 7 billion dollars.

Source updated; analyst comments added in paragraphs 6 to 7

Krystal Hu/Jaspreet Singh

Reuters, September 29 - A source familiar with the matter told Reuters on Tuesday that OpenAI's annualized recurring revenue has approached $7 billion, driven by a more than doubling of corporate sales since July.

The source said the startup’s annualized revenue run rate has grown by more than 70% since the start of the third quarter in July. Due to the confidential nature of the matter, the source declined to be named.

According to the source, OpenAI’s consumer business revenue in the third quarter has already surpassed the total for all of last year. Axios was the first to report this news.

The annualized revenue run rate is sometimes a misleading sales metric (link), often calculated by multiplying single-month revenue by 12, but it has become a popular metric widely used by fast-growing startups in Silicon Valley.

Shares of Oracle, a key computational infrastructure partner for OpenAI, rose 5.3%.

"Oracle’s fortunes are closely tied to OpenAI’s success, which accounts for nearly half its compute order backlog," said Gil Luria, managing director at equity research firm D.A. Davidson.

"Accelerating OpenAI growth is a very good sign for Oracle, as it reinforces OpenAI’s ability to meet their compute expectations."

OpenAI and rival Anthropic are preparing for IPOs, which could give Wall Street a clearer look at both companies’ financials, including whether the rapid revenue growth fueled by the AI boom can be sustained.

Reuters reported on Monday that Anthropic reported a net loss of $4.2 billion in 2025 (link), and plans to spend $51.8 billion over the coming years on cloud services, computing, and infrastructure-related costs.

In 2025, Anthropic’s revenue grew 12-fold to nearly $4.6 billion. The startup expects to invest at least $51.8 billion (link) over the next decade together with six partners to build artificial intelligence infrastructure.

The IPO prospectus of this "Claude" developer (link) also highlighted risks associated with its AI models, stating these models may exhibit "self-preservation behaviors," such as attempting to "resist shutdown," concealing or manipulating information, and "extortion-like" behaviors.

(To assist non-English speakers, Reuters has automated the translation of its reports into several other languages. As automated translations may be incorrect or fail to include needed context, Reuters does not guarantee the accuracy of automated translation texts and provides them solely for the convenience of readers. Reuters accepts no liability for any damage or loss arising from the use of this automated translation function.)