OpenAI CFO Pushes Back on Altman's 2026 IPO Timeline, Citing Spending Risks

OpenAI chief financial officer Sarah Friar raised serious internal concerns about CEO Sam Altman's push to take the company public as early as the fourth quarter of 2026, warning colleagues that OpenAI is not yet "IPO-ready" and that its five-year infrastructure spending commitments, totaling more than $600 billion in cloud and compute capacity, expose the company to severe financial risk well before it achieves profitability.

The disagreement represents a sharp split inside one of the most consequential private companies in the world. Altman pressed for a faster public listing, hoping to establish a public-market valuation and secure long-term capital access while investor enthusiasm for AI remained elevated. Friar pushed for delaying an IPO until 2027 or later, pointing to the extensive procedural, compliance, and organizational work still required before OpenAI could withstand the disclosure demands that SEC-registered public markets impose.

The financial exposure she flagged is significant on its own terms. OpenAI committed more than $600 billion across the next five years for cloud server capacity, and the company is projected to burn through more than $200 billion before turning cash-flow positive. Friar also questioned whether OpenAI needed to invest so aggressively in AI infrastructure at all given slower-than-expected revenue growth, a challenge that strikes at the core tension between Altman's growth-at-all-costs posture and the kind of fiscal discipline that public shareholders typically demand.

The governance dimensions of the dispute add a further layer of risk. Reports indicate that Friar was sidelined from certain infrastructure and capital-allocation conversations, an unusual arrangement for a CFO at any company, let alone one preparing for a major public offering. For public investors, the credibility of a company's pre-IPO disclosures depends heavily on whether the finance chief has had full visibility into the commitments those filings must describe.

The national-scale stakes of OpenAI's infrastructure buildout make those disclosures particularly consequential. A $600 billion compute commitment, spread across data-center construction, GPU procurement, and networking contracts, implies enormous demands on power grids, water supplies at cooling-intensive facilities, and semiconductor supply chains already stretched by broad AI industry investment. The partnership structure with Microsoft, whose cloud infrastructure underpins much of OpenAI's product delivery, would also require detailed public accounting of how revenue and costs flow between the two companies, an analysis that could materially affect how investors price the business.

If the Altman-Friar disagreement leads to a delayed or restructured offering, the effects will extend beyond OpenAI's own balance sheet. A postponed IPO would recalibrate private-market valuations across the AI sector, affect liquidity timelines for employees and early investors, and signal to competitors such as Anthropic that the public-market path for frontier AI companies remains harder to navigate than growth metrics alone would suggest. The pressure Friar applied may ultimately be the discipline that makes an eventual OpenAI IPO credible rather than simply large.