The $852 billion valuation of OpenAI is being questioned by its own investors.

      Some investors claim that OpenAI has changed its product roadmap twice in six months and risks losing direction ahead of a potential IPO anticipated as soon as Q4 2026. OpenAI's new Chief Revenue Officer has accused Anthropic of exaggerating its $30 billion run rate by $8 billion due to its gross accounting practices concerning cloud partner revenue. Both companies assert that they adhere to standard accounting methods.

      OpenAI’s valuation of $852 billion is under scrutiny from some of its investors as the company shifts its focus toward the enterprise sector, according to a report from the Financial Times on Tuesday.

      The concerns revolve around a phase of noticeable strategic instability: OpenAI has altered its product roadmap twice within six months—first due to competition from Google and then from Anthropic—and has recently abandoned several projects, including the Sora video generation launch and an 'adult' chatbot.

      Some investors expressed to the Financial Times that these rapid shifts might expose the company to Anthropic and a resurgent Google, even while it readies for a possible IPO by the end of 2026.

      The criticism is sharp. An early supporter of OpenAI remarked to the Financial Times: "You have ChatGPT, a billion-user business growing at 50-100% per year; why are you discussing enterprise and code? It’s a company lacking focus.”

      Jai Das, president of Sapphire Ventures, who isn’t an investor in either OpenAI or Anthropic, further characterized OpenAI as “the Netscape of AI,” comparing it to the once-dominant browser company that was ultimately surpassed by Microsoft and acquired by AOL.

      One investor who has supported both firms mentioned that to justify OpenAI’s latest funding round, they would need to project an IPO valuation of $1.2 trillion or more.

      OpenAI's leadership strongly defended the company. Chief Financial Officer Sarah Friar highlighted the $122 billion fundraising completed last month, deemed the largest private round in Silicon Valley history, supported by major investors like SoftBank, Amazon, Nvidia, Andreessen Horowitz, Sequoia Capital, and Thrive Capital, among over 25 others, as proof of investor confidence.

      “The notion that investors are not backing our strategy contradicts the facts,” Friar stated. “Our raise, the largest on record, was oversubscribed, completed in record time, and supported by a wide range of global investors.”

      Additionally, Friar shared with CNBC that enterprise now constitutes 40% of OpenAI’s overall revenue and is expected to equal its consumer business by the end of 2026.

      OpenAI also aims for 30 gigawatts of computing capacity by 2030 and informed investors last week that it has already secured 8 gigawatts, a capacity it claims Anthropic will not reach until late 2027.

      Central to the competitive concerns is Anthropic’s revenue growth. The maker of Claude saw its annualized run rate surge from about $9 billion at the end of 2025 to $30 billion by the end of March 2026, primarily driven by strong demand for its coding tools.

      As of February, OpenAI reported reaching $25 billion in annualized revenue. This apparent disparity prompted a notably forceful response from OpenAI’s new Chief Revenue Officer, Denise Dresser, who joined the company in December 2025 from her previous position as CEO of Slack.

      In an internal memo sent to staff on Sunday, Dresser charged that Anthropic had inflated its run rate by roughly $8 billion. This accusation hinges on a recognized accounting difference: Anthropic records the full value of revenue generated via its cloud distribution partners, Amazon Web Services and Google Cloud, on a gross basis, while OpenAI reports its share of revenue from Microsoft on a net basis, subtracting the partner's share prior to recognition.

      Both methods are valid under US GAAP. If Dresser's assessment holds true, it would suggest that Anthropic's actual run rate is closer to $22 billion rather than $30 billion.

      Anthropic contested this characterization. A source close to the company told the Financial Times that Anthropic “recognizes gross revenue on sales through partners because it is the principal in the transaction and its cloud partners are the distribution channel,” a common rationale for gross recognition under accounting standards.

      Dresser's memo acknowledged that Anthropic’s “coding focus provided them with an early advantage” in the enterprise market, but suggested that a narrow, developer-focused approach could become a drawback as AI extends beyond engineering teams.

      “You don’t want to be a single-product entity in a platform war,” the memo indicated. It also detailed OpenAI’s Q2 priorities: securing the enterprise model layer with a new model codenamed ‘Spud’, establishing its Frontier agent platform, expanding through a recently announced partnership with Amazon, and creating a deployment engine known as DeployCo.