Why OpenAI's 2025 Losses Mask a Fast-Growing Engine

A single line item in confidential filings landed like a thunderclap: a net loss of about €35.9 billion in 2025. It reads like a courtroom headline, but the story behind the number is messier, louder, and far more revealing about how modern AI firms scale.

Big revenue. Bigger bills.

OpenAI’s top line surged: revenue jumped from roughly €3.40 billion in 2024 to about €12.02 billion in 2025. Monthly receipts climbed toward €1.84 billion by the end of 2025, up from roughly €920 million per quarter only a year earlier. Growth that fast is rare. It also eats cash like a furnace.

Costs exploded alongside revenue. Total expenses rose to about €31.3 billion last year. R&D spending more than doubled to around €17.65 billion, and within that, roughly €9.74 billion flowed to a single partner: Microsoft. Sales and marketing climbed too, from about €1.02 billion to €5.27 billion. Put it together and operating outlays ballooned from about €8.08 billion in 2024 to roughly €19.25 billion in 2025.

So how does a company that multiplies revenue several times over still report such a headline loss? The short answer: accounting quirks and an appetite for spending that matches its ambition.

One-off accounting and the convertible debt shadow

Part of the €35.9 billion loss is an accounting charge tied to the company’s prior structure and investor instruments. Before OpenAI changed into a public benefit corporation, early backers held convertible rights that were treated as debt under US accounting rules. When OpenAI’s valuation climbed, those instruments revalued too, producing a roughly €27.6 billion non-cash accounting hit. Insiders say that expense is unlikely to recur after the corporate restructuring.

That matters. Strip out the revaluation, and the headline loss narrows substantially. But even without the one-off, the business still spends big, and those day-to-day expenses are the hard reality investors will scrutinize.

Worried yet? You should be curious, not alarmed. High burn is part of the playbook when the product is an AI platform that needs constant retraining, new models, and global scale to stay relevant.

Funding the runway

Who bankrolls this run? So far, investors have. In March, OpenAI raised about €112.24 billion in new capital, pushing the company’s private valuation to roughly €783.84 billion. Executives then filed confidential paperwork with the US securities regulator, signaling an IPO bid that would aim to push the valuation past the equivalent of €920 billion.

Competitors are racing too. Anthropic closed a large fundraising round that translated to about €59.8 billion and a valuation near €828 billion, and it has also filed confidentially to go public. The result: a market where eye-popping private valuations become the baseline competitors must meet or beat.

But private capital isn't eternal. Turning those paper valuations into durable public-market businesses requires a path to profit. OpenAI’s management has told investors they expect the company to reach full profitability by 2030. That timeline is a promise, not a guarantee.

So what levers can be pulled to get there? The obvious one is cost control. The heaviest line items are model development and training, plus third-party compute and cloud partnerships. Taming these without throttling product innovation is the central balancing act.

Another lever is monetization: converting fast user growth and enterprise adoption into recurring, higher-margin revenue streams. Pricing, partnerships, and product packaging will determine whether growth eventually outpaces spend.

Investors and rivals will watch the margins closely. On one hand the operating loss as a percentage of revenue improved from about 237 percent to 160 percent year over year, showing some operational leverage. On the other hand, the absolute cash burn remains material and demands answers.

In short: the numbers are both alarming and predictable. They show how much it costs to chase leadership in a technology that rewards scale and punishes hesitation. The accounting spike explains some of the drama, but the ongoing test is whether OpenAI can convert its market momentum into steady, profitable operations before the patience of deep-pocketed backers runs thin.

That tension—between breakneck growth and the need for discipline—will define not just OpenAI’s next chapter, but how the AI industry matures on public markets.