Key Takeaways
A disagreement between OpenAI’s top executives over the timing of its initial public offering threatens to push the much-anticipated listing to 2027, as the company grapples with spending commitments that exceed $1 trillion.
Chief Financial Officer Sarah Friar has privately suggested delaying the IPO until 2027, according to reports in the Wall Street Journal, cautioning that the company is not yet prepared for the financial rigor required of public companies. This timeline is at odds with CEO Sam Altman, who has been pushing for a listing as early as the fourth quarter of 2026.
The internal friction highlights a fundamental imbalance in OpenAI’s finances. The company, which generates roughly $2 billion in monthly revenue, has locked in an estimated $1.15 trillion in non-negotiable, long-term spending commitments for data center infrastructure, according to an analysis from PitchBook. These deals include partners like Microsoft, Oracle, and Nvidia.
The stakes of the IPO’s timing are critical for the entire AI sector. Banks have reportedly advised OpenAI and its chief rival, Anthropic, that the first frontier AI model company to list on the public markets will get to define the industry’s valuation framework. A delay could hand that advantage to a competitor with a more stable financial structure.
The scale of OpenAI’s financial obligations creates a dangerous cash flow scenario. According to PitchBook, the company’s agreements include a deal with Oracle that requires $60 billion in annual spending starting in 2027, a figure that could surpass OpenAI’s entire net revenue for that year. These costs are fixed even if revenue growth slows, a risk that materialized earlier this year when the company reportedly missed targets in enterprise segments where Anthropic has gained ground.
This has put Friar, who joined from Nextdoor and previously oversaw Square’s IPO, in the position of reining in spending. The public emergence of her strategic friction with Altman signals deeper issues around capital allocation, especially as the company continues to sign new deals, including a reported $100 billion commitment from Amazon Web Services.
The pressure is magnified by a widening gap with competitors that operate on a leaner model. Anthropic, for example, functions with approximately one-twelfth of OpenAI’s infrastructure costs, leading to stronger gross margins and greater capital efficiency.
This efficiency is reflected in per-employee metrics. Anthropic generates about $6 million in annualized revenue per employee with a 5,000-person staff. OpenAI generates roughly $5.6 million per employee with a 4,500-person staff, a figure that will be diluted as it plans to nearly double headcount by the end of the year.
PitchBook’s analysis suggests the IPO delay is becoming unavoidable, with mid-to-late 2027 being a more realistic target. Public investors, the firm noted, will require several quarters of stable performance to believe that OpenAI’s more than $1.15 trillion in infrastructure deals can generate meaningful free cash flow. If a rival like Anthropic or Databricks lists first on cleaner economics, OpenAI risks having the market’s valuation terms dictated to it, despite having deployed the most capital in the AI race.
This article is for informational purposes only and does not constitute investment advice.
