OpenAI's $39 Billion Loss: Market Prices In Chaos as IPO Looms

OpenAI hemorrhaged $39 billion last year. Let that number sit for a moment, the way it presumably sat in the offices of potential institutional investors before they collectively shrugged and moved on to other matters.

The ChatGPT maker's audited financial figures, reported by the Financial Times, show a net loss that widened sevenfold from the $5 billion the company lost in 2024. Excluding restructuring charges and other non-cash accounting adjustments—the financial equivalent of sweeping broken glass under the sofa—the operational loss was $8 billion. Still catastrophic by traditional metrics. Apparently not by the metrics that matter anymore.

OpenAI's spending tells you everything about where the generative AI industry has decided to place its bets. The company spent $34 billion last year. Research and development consumed $19 billion of that. Sales, marketing, and general corporate overhead accounted for another $5.8 billion. The remaining money went to the sort of infrastructure spending that makes spreadsheets look like abstract art.

Against this, OpenAI generated approximately $24 billion in annual revenue—$2 billion per month as the company announced in March. That's a revenue-to-spending ratio that would have disqualified the company from polite financial society somewhere around 2019. Today it is a rounding error in the conversation about AI dominance.

The company confidentially filed for a U.S. IPO last week. OpenAI carries an $852 billion valuation into those public markets discussions, a figure that exists in a separate dimension from traditional valuation frameworks. Timing remains uncertain. Reuters reported in late June that OpenAI is now considering delaying its public debut until 2027, which is another way of saying: we will go public when we feel like it, and the market will price our shares based on what we might be worth in five years, adjusted for factors that haven't been invented yet.

Here's what's remarkable about this particular moment in market history: the $39 billion loss, disclosed ahead of an IPO filing, generated the market equivalent of a slight nod. Wall Street has collectively agreed that profitability in AI companies is a nice-to-have, like cup holders on a spaceship. The real asset is dominance in a technology race whose endpoint nobody can quite articulate but everyone agrees matters enormously.

The scale of OpenAI's losses reflects the capital intensity of large language model development. Training state-of-the-art AI systems requires data centers, specialized chips, and the kind of electricity consumption that makes power utility operators nervous. Scaling these systems to serve millions of users requires more of the same. The company is not losing money on mismanagement or operational incompetence. It is losing money by design, as a competitive strategy.

Investors understand this, or believe they do. The valuation discipline that would normally apply to any company burning through tens of billions in cash annually has been suspended in the case of generative AI firms. This suspension is not unique to OpenAI. Anthropic, another large language model company, has raised billions at multibillion-dollar valuations on similar economics. The market is betting that whoever wins the AI race gets to write the rules about profitability afterward.

The adjusted $8 billion loss—the figure that excludes the non-cash items and makes investors feel slightly better about attending the IPO roadshow—still represents a company spending roughly three times its annual revenue. That arithmetic would normally trigger immediate skepticism from institutional investors. Instead, it has triggered a different question: how much more will it cost to maintain competitive advantage, and can we afford not to pay it?

OpenAI's IPO will face scrutiny, particularly if Anthropic launches alongside it. Investors will demand clarity on the path to profitability, on unit economics, on the sustainability of current spending patterns. OpenAI will provide answers that acknowledge these concerns while largely sidestepping them. Wall Street will accept this because the alternative—betting against AI market dominance—has become the riskier trade.

The $39 billion loss is the number. The $24 billion in annual revenue is the context. The $852 billion valuation is the market's verdict. Markets are pricing in neither profitability nor near-term financial discipline. They are pricing in the belief that OpenAI has won something worth winning, and that the cost of the victory is secondary to the spoils that follow.