OpenAI Lost $34B in the Year Before Its IPO

The OpenAI burn rate just hit $34 billion in twelve months for $13 billion in revenue. The official net loss climbs to $39 billion. All of it as the company prepares an IPO targeting more than $1 trillion in valuation.

Key Takeaways

• The OpenAI burn rate hits $34 billion in spending over one year, with $19 billion in R&D and $6 billion in marketing.
• Net loss jumps from $5 billion to $39 billion, but $30 billion is a non-cash accounting charge.
• The upcoming IPO targets a valuation north of $1 trillion.

How the cash is burning

The numbers land at the worst possible moment for OpenAI. Over the past fiscal year, the company spent $34 billion for only $13 billion in revenue. The gap between cash going out and cash coming in has never been more visible.

In detail, R&D swallows $19 billion. Sales and marketing costs approach $6 billion. The rest splits between infrastructure, long-term compute supplier contracts, and operating costs. Across nearly every product segment, reconstructed gross margin stays in the red.

The top line is still climbing fast. Monthly revenue reaches $2 billion at year end, compared to $1 billion per quarter at the end of 2024. The curve goes up sharply, but the spend-to-revenue ratio remains massively unfavorable.

This financial snapshot drops in the middle of the API price war OpenAI launched against Anthropic in recent weeks. Cutting prices while showing an OpenAI burn rate of this scale is a deliberate strategic choice, not a comfortable margin of maneuver.

The accounting trap and the real loss

The official net loss is spectacular. It moves from $5 billion to $39 billion in one year. Multiplied by almost eight. At first glance, this looks like a collapse.

Except about $30 billion of this loss is a non-cash accounting charge, tied to the previously executed corporate restructuring. This is not money leaving the books. It is an entry that reflects the new legal architecture of the company.

Strip out that charge, and the real cash loss lands around $8 billion. Mechanically less violent. But $8 billion in cash lost over twelve months remains a colossal hole for a company that has never been profitable and must keep funding an open-ended GPU race.

The gap between the two numbers will become a sensitive topic in coming weeks. Institutional investors preparing the IPO will want a clean read. Analysts will want to know whether this charge truly isolates a one-off restructuring, or whether it hides other adjustments likely to return next year.

This opacity lands at the wrong moment for the OpenAI burn rate story. The state attorneys general probe already scrutinizes legal structures and Microsoft partnership terms. A public listing builds on transparency, not on two parallel sets of figures.

Also on Horizon:

• Google Appeals Munich AI Overviews Ruling
• Claude Code Artifacts: Anthropic Opens Sharing
• Amazon Shelves the Artificial Film on Sam Altman

What the trillion-dollar IPO changes for rivals

OpenAI is preparing a public listing targeting a valuation above $1 trillion. The timeline is firming up. At that level, the company joins the very small club of listed global giants and turns its spend-to-revenue ratio into the new reference standard for the sector.

For Anthropic, which already filed a confidential draft S-1 with the SEC, the arbitrage becomes strategic. If OpenAI prices first at one trillion, the bar is set. Every rival will be compared to that benchmark, including on gross margin trajectory and on speed of converting revenue into cash.

For Microsoft, the operation is more ambivalent. A successful IPO mechanically revalues Redmond’s stake. It also raises questions on dilution, on the governance of the new entity, and on Microsoft’s exact post-listing role inside that governance.

In the short term, these numbers weigh on private market sentiment. AI startups raising right now will have to show user cohorts and gross margins, not just revenue curves. The “we lose money but we take share” narrative gets harder to sell when the leader posts $8 billion in cash losses over twelve months.

In the medium term, the question gets brutal. Either OpenAI converts its revenue growth into real margin, and the IPO legitimizes the generative AI business model. Or the public market discovers that the OpenAI burn rate isn’t shrinking fast enough, and the punishment will be heavy for the whole sector.

For end users, these figures partly explain the pressure on API prices and the billing changes seen across every provider. When a leader burns $34 billion in one year, the margins of the entire market end up moving.

Follow the story on Horizon.