OpenAI Chief Financial Officer Sarah Friar said the company may postpone its initial public offering from late 2026 to 2027 [1].
The shift reflects the financial strain of the current AI arms race, where massive infrastructure costs are outpacing immediate revenue growth. A delay would allow the company to stabilize its balance sheet before facing the scrutiny of public markets.
Speaking on the All-In Podcast, Friar said the company is grappling with challenges of maintaining leadership in a field defined by intense competition. The company is grappling with immense fixed-infrastructure costs, which some reports indicate exceed $1.15 trillion [2]. Other accounts of the spending suggest a figure of more than 100 billion on compute [3].
Friar said the company missed certain revenue targets, necessitating a more cautious approach to its public debut. The capital-intensive nature of developing next-generation models requires constant investment in hardware, and energy—costs that create significant pressure on the company's current financial structure.
Beyond the financial outlook, Friar provided a preview of a new OpenAI device. While specific details remained limited, the hardware represents a strategic move to integrate AI more deeply into physical user interfaces, reducing reliance on third-party platforms.
This strategy comes as OpenAI continues to scale its operations. The tension between the need for rapid innovation and the requirement for fiscal discipline has become a central theme for the organization as it transitions from a research-focused entity to a global commercial powerhouse.
“OpenAI may postpone its initial public offering from late 2026 to 2027.”
The potential delay of the IPO signals that the 'compute bubble' is creating a genuine liquidity challenge for even the most prominent AI firms. By pushing the public debut to 2027, OpenAI is attempting to bridge the gap between its astronomical capital expenditures and its actual revenue generation, suggesting that the path to profitability for generative AI remains more expensive and slower than initial market projections suggested.
