There is currently disagreement within OpenAI about when to IPO: Altman hopes for the fourth quarter, but the CFO believes they are not yet ready.

The internal tension between OpenAI's aggressive expansion strategy and financial prudence is coming to the surface. According to The Information, CEO Sam Altman privately expressed a desire to complete an IPO as early as the fourth quarter of this year, while CFO Sarah Friar had expressed the opposite view earlier this year, believing that the company is not yet ready to go public. Friar's concerns are focused on the company's financial reality: OpenAI has committed to spending over $60 billion on server rentals in the next five years and has warned investors that the cash burn will exceed previous forecasts by 2030, potentially surpassing $200 billion. Sources who have communicated with Friar revealed that she still has doubts about whether the company's slowing revenue growth can support such a scale of spending. Meanwhile, competitor Anthropic is eroding OpenAI's market share, further increasing external pressure. In response to reports of a rift in their relationship, Altman and Friar jointly issued a statement saying they are "completely aligned" on the importance of persistent access to computing power as a core strategy for OpenAI, and that both have been "directly involved in every major computing decision" over the past year. Divergence in IPO timeline Despite Friar's reservations, OpenAI has begun preparations for an IPO. The company has hired law firms Cooley and Wachtell Lipton Rosen & Katz, and has had informal discussions with the IPO underwriting teams of Goldman Sachs and Morgan Stanley. Altman's desire to go public also has a clear competitive motivation, as he has expressed a desire for OpenAI's IPO to be completed before Anthropic, which is currently discussing going public in the fourth quarter of this year. If the IPO goes through, it could be one of the largest in history. Friar's reluctance to go public quickly is not without traces. In an interview with The Wall Street Journal in November of last year, she openly stated that an IPO was "currently not in the plans," as the company is still focused on "aligning the company with its current scale." Changes in internal power structure At the organizational level, an unusual change has already occurred: since August of last year, Friar no longer reports directly to Altman, but instead reports to the head of the applied business, Fidji Simo. CFO reporting directly to a senior executive other than the CEO is a rare arrangement in large companies. According to multiple sources who have closely worked with both of them, Altman has excluded Friar from several key financial planning discussions. In one instance, Altman did not invite Friar to a recent discussion with a major investor about server expenses, despite her attendance at an earlier meeting on the same topic. One attendee described her absence as "noticeable and awkward." Another source stated that Friar was also not invited to a high-level executive meeting earlier this year on a significant financial decision. Financial risks hidden in high computing power commitments OpenAI's current scale of computing expenses is unprecedented. OpenAI has signed server rental contracts totaling approximately $665 billion, including agreements with companies such as Oracle (about $300 billion, five-year term starting in 2027), Microsoft ($250 billion until 2032), Amazon Web Services (about $138 billion, eight-year term), CoreWeave ($22 billion, five-year term), Cerebras ($10 billion), and others. These commitments are not ordinary cloud computing contracts. Friar explained that AI data center construction takes several years, so OpenAI must book capacity in advance. "I have to make decisions today to make sure we have enough computing power in 2028, 2029, and 2030. If I don't order today, the data centers won't come," she said. In one case, according to The Information, OpenAI and Oracle signed a risk-sharing agreement on overspending on data center construction costs, a rare clause in cloud computing customer contracts. Currently, OpenAI has shelved its previous plans to build its own data centers. Dual pressures of revenue growth and cash burn The competitive landscape is tightening. Anthropic has surpassed OpenAI in selling AI models to enterprises and app developers, while Google's Gemini continues to erode ChatGPT's dominance in the consumer chatbot market. OpenAI has raised its revenue expectations for the next five years by 27%, but has also privately disclosed to investors that cash burn until 2030 will exceed twice the forecast made last summer. Additionally, the company informed investors that last year's gross profit margin was lower than expected due to higher-than-expected user demand, leading the company to temporarily purchase computing power at higher prices. Anthropic's co-founder and CEO Dario Amodei bluntly discussed the dangers of ahead-of-schedule investments in data centers on the Dwarkesh Patel podcast, indirectly referring to OpenAI: "If I'm wrong by even just a year...you're bankrupt. My impression is that some companies haven't really done the math, they don't truly understand the risks they are taking." This resonates strongly with the concerns expressed internally by Friar. A source who has closely worked with both of them described the CFO's situation as, "She is facing a founder with ambitious goals who wants to go all-in on spending, which is a tough job." This article is selected from "Wall Street News," edited by GMTEight: Lee Ford.