The above button links to Coinbase. Yahoo Finance is not a broker-dealer or investment adviser and does not offer securities or cryptocurrencies for sale or facilitate trading. Coinbase pays us for certain activity generated through this link. Prices displayed are informational.

Our analysts just identified a stock with the potential to be the next Nvidia. Tell us how you invest and we'll show you why it's our #1 pick. Tap here.

OpenAI generated net revenue of $13.1 billion last year but incurred a net loss of $38.5 billion that same year, according to tech writer and independent journalist Ed Zitron, who got his hands on the figures. The numbers were independently verified by the Financial Times .

For context on how big that loss is, Meta’s VR and hardware division, Reality Labs, lost $17.7 billion in 2024, and more than $70 billion cumulatively since 2020. WeWork's peak annual losses reached roughly $4.6 billion before its collapse.

That’s obviously not what you’d want to hear if you’re Softbank’s Masayoshi Son — who also backed WeWork — or Oracle’s Larry Ellison, who are all-in on OpenAI. If it can't survive on its own, the federal government may end up taking an equity stake to bail them out.

OpenAI’s cash burn is almost impressive. CEO Sam Altman dumped $20 billion into research and development and $5.7 billion on sales and marketing, with a total operating loss of about $20.1 billion, validating Zitron’s long-held bear case on AI. SoftBank also paid $867 million to OpenAI while Microsoft paid $303 million last year, or about 9% of OpenAI’s reported $13.07 billion in revenue, per Zitron’s report.

The numbers are bad, with some caveats: OpenAI converted from a non-profit to a for-profit entity last year. That led to a $41.6 billion loss due to changes in fair value of convertible interests and warrant liability, Zitron writes. This means all early investor IOUs that later converted into equity became debt on the books. That number spiked because OpenAI's value did.

"As OpenAI's worth rose, the increased value of those investor rights created a roughly $30 billion charge," a person familiar with the matter told FT. No money actually left the building, but a one-time clerical ding that, to be fair to OpenAI, inflated its "losses."

Zitron then factors in interest income and interest expense, leaving OpenAI even deeper in the hole with a "net loss of $60.4 billion." The ChatGPT-maker managed to whittle that number down only because OpenAI doesn't control 100% of its own house. The company was able to peel about $21.2 billion of the total loss off its own books by assigning it to non-controlling interests, blunting the impact. But it's unclear what those interests even are.

The final damage? "A nearly eightfold increase in the net loss attributable to OpenAI," FT writes, "which soared from $5 billion in 2024 to around $39 billion in 2025." Ouch.

One stock. Nvidia-level potential. 30M+ investors trust Moby to find it first. Get the pick. Tap here.

The numbers are bad, but not as bad as Zitron makes them out to be. At least according to the person talking to the FT.

After stripping away non-cash accounting, non-recurring charges, and other "non-cash expenses" like stock-based compensation of staff or computing credits from Microsoft, OpenAI's true "out-of-pocket" cash burn for the year was around $8 billion, according to the anonymous source. This figure is supported by revenue of $13.1 billion and a $25 billion cash reserve, backed, of course, by its insane funding numbers.

The biggest red flag for OpenAI is obviously the loss from operations at $20.1 billion, which is a universal thorn for every AI firm. xAI burned $6.4 billion last year, a 307% explosion from its $1.6 billion loss in 2024. Anthropic was the winner, kind of, as the company turned a profit in Q2 this year, though Zitron called that “engineered.”

But none of this — the $38.5 billion paper loss, the billions invested into data centers, and selling equity to the federal government makes sense unless the engine underneath is highly profitable. Right now, based on these numbers, it isn’t.

Let's not forget that even the $8 billion cash burn figure understates the real cost of compute. Much of OpenAI's compute has come from Microsoft as in-kind Azure credits rather than cash OpenAI actually paid out, and the scale isn’t small. It ran up Azure inference costs between $8.7 and $9 billion in the first three quarters of 2025, according to leaked documents. Microsoft is willing to do this because it owns roughly 27% of OpenAI.

When those prepaid credits run out, or any other curveball comes OpenAI's way, it's going to get a lot worse for its cash burn, because that cost is going to actually start leaving its books.