OpenAI's CEO Wants to Go Public This Year. Its CFO Says That's a $600 Billion Mistake.

OpenAI, currently valued at $852 billion after closing a $122 billion funding round in March, is preparing to go public in Q4 2026 — except its own CFO reportedly does not think that is a good idea. Sam Altman is pushing for a 2026 listing. Sarah Friar has told colleagues the company may not be ready, flagging $600 billion in future infrastructure commitments against an annualized revenue base of $25 billion. The number she keeps returning to is not the revenue. It is the gap between what OpenAI has promised to spend and what it needs to earn to justify those promises to public market investors.

This is not a routine CFO caution. When Friar spoke publicly to CNBC in April, she said demand from retail investors is "strong" and committed to setting aside IPO shares for them. In private, she has reportedly argued for pushing the listing to 2027. A CFO simultaneously cultivating retail investor enthusiasm and warning internally against the timeline is not a contradiction — it is a description of the specific pressure she is under.

What Happened

The Wall Street Journal reported in late April that OpenAI had missed internal targets for both weekly active users and monthly revenue at multiple points in early 2026. The company's internal goal was to reach one billion weekly active ChatGPT users by the end of 2025. It reached approximately 900 million by February 2026 — 125 percent year-over-year growth that would be the envy of almost any consumer internet company, but still short of the internal benchmark.

On revenue, OpenAI's annualized run rate sits at roughly $25 billion as of February 2026. That represents extraordinary growth from a standing start. But OpenAI had set internal projections higher, and the shortfall compounded: Anthropic surpassed OpenAI in enterprise revenue, particularly in coding and agentic workflows, while Google Gemini began recapturing consumer market share through search and Android integration.

Sam Altman's public response was that the company is "firing on all cylinders." Sarah Friar's internal response was to ask whether the company was ready to defend those cylinders to public market analysts who would compare every metric against OpenAI's own projections.

The OpenAI IPO debate is not about whether to go public. It is about whether going public before 2027 creates legal and reputational exposure that a delay would avoid. If OpenAI lists in Q4 2026 at a $1 trillion valuation and then misses a Q1 2027 earnings target, the resulting correction would damage the company in ways that a private funding round does not.

Why the $600 Billion Number Is the Real Story

The $25 billion in current revenue gets most of the coverage. The $600 billion commitment is what matters.

OpenAI has contracted for hundreds of billions of dollars in cloud infrastructure from Oracle, CoreWeave, Microsoft, and others. These are not options or soft commitments. They are structured agreements that assume OpenAI's revenue will scale from $25 billion today to $280 billion by 2030 — an 11-fold increase over four years at a compound annual growth rate above 80 percent. No technology company of OpenAI's size has sustained growth at that rate. Microsoft, which grew faster than any peer in its early years, never approached 80 percent CAGR at $25 billion in revenue.

Friar's concern, according to reporting on the internal debate, is that public market investors will do this math. Private investors at $852 billion accepted the narrative that frontier AI is winner-take-most, that ChatGPT's 900 million weekly users are a durable moat, and that the infrastructure commitments are a feature rather than a liability. Public market analysts — particularly the short-sellers who materially shaped how Meta, Uber, and Snap were priced in their first years as public companies — will scrutinize every missed monthly target against a $600 billion denominator.

The question is not whether OpenAI will eventually justify a $1 trillion valuation. The question is whether it can maintain that valuation during the eighteen months immediately following the IPO, when the gap between the infrastructure commitment and the revenue trajectory will be most visible.

Altman's case for proceeding is structural: OpenAI is converting from a nonprofit-controlled entity to a fully commercial corporation, and that conversion requires capital that the private market has effectively maxed out. The $122 billion round was notable not just for its size but for how unusual the terms were — essentially a one-of-a-kind arrangement assembled by SoftBank, Amazon, NVIDIA, and Microsoft that would not easily repeat. The IPO is partly a necessity of the restructuring, not purely a financing choice.

The Comparison Every Analyst Will Make

The openai ipo will be measured against two reference points that do not flatter it.

The first is Uber. Uber went public in 2019 at a $82 billion valuation while losing $8.5 billion annually. The stock dropped 37 percent in its first year. It eventually recovered as the company reduced losses and reached profitability in 2023. OpenAI's situation is structurally similar but at 10× the scale: roughly $14 billion in projected 2026 losses against a valuation that has now exceeded $852 billion. Uber at least had a clear path to unit-level profitability through driver take-rate optimization. OpenAI's path requires inference costs to fall faster than they currently are, and competition to not materially worsen.

The second reference is the timing. SpaceX is targeting a roadshow as early as June 2026. Anthropic is expected to follow in October. All three companies are expected to pursue public listings within a nine-month window, with combined valuations exceeding $3 trillion. Analysis firm TradingKey estimates this IPO wave could absorb or displace a significant share of 2026 institutional and retail liquidity — meaning each company's reception will partly depend on how the others are received.

If SpaceX's IPO outperforms expectations in June, it likely helps OpenAI's September or October window. If Anthropic, which has cleaner financials and a stronger enterprise revenue story, prices well in October, it makes OpenAI's pitch harder. Sarah Friar's 2027 preference is not just about OpenAI's own numbers — it is about not competing for the same capital pool as two companies with arguably simpler IPO stories.

What Happens Next

The immediate question is whether Altman or Friar's preference wins internally. Based on current reporting, the Q4 2026 timeline remains the operating assumption, with the most likely window being October or November to avoid holiday-period low liquidity.

The filings will be the first moment of real transparency. OpenAI has never published audited financials publicly. When the S-1 drops, the $600 billion commitment number, the revenue trajectory, and the projected loss for 2026 will all become verifiable public record. Every projection that appeared in private pitch decks will be tested against actual quarterly figures.

Friar's retail investor commitment — promising allocations to individual investors — adds another layer. Retail participation in tech IPOs has historically been a leading indicator of post-IPO volatility. The stocks that allocated heavily to retail in 2021 (Robinhood, Coinbase, Rivian) experienced some of the most dramatic post-IPO declines. Friar's public promise was probably made to build goodwill and democratize access. The execution will matter more than the intention.

The openai ipo will be the most closely watched public market event of 2026. Whether the CEO's timeline or the CFO's caution prevails, the filing itself will answer questions that $122 billion in private funding never needed to address. That transparency — forced, not voluntary — is the actual milestone.

The bet implicit in Altman's timeline is that the narrative of AI inevitability is durable enough to carry a $1 trillion valuation through the scrutiny of public markets. Friar's argument is that one more year of operational data would make that narrative much easier to defend. They are both probably right. The question is which risk is larger: the cost of waiting, or the cost of listing before the gap closes.

For anyone tracking AI infrastructure and the companies building on top of it, the OpenAI IPO signals something important about where the industry thinks value will concentrate — in frontier model providers, not just applications. Whether those providers can convert their user scale into the revenue velocity required to support $600 billion in commitments will shape what gets built on top of them for the next decade. The tools that let knowledge workers move faster in AI-native workflows — from real-time information synthesis to AI-powered second brain systems — depend on this infrastructure existing and remaining accessible. That makes the CFO's spreadsheet everyone's problem.