Unpacking OpenAI's Billions: Questioning Revenue Claims Amidst Massive Cash Burn
An in-depth analysis challenges Sam Altman's OpenAI revenue claims, contrasting them with Microsoft's disclosures of $12bn quarterly losses and an unreported $4.1bn cash burn.
Soundtrack: Queens of the Stone Age - Song For The Dead
Editor's Note: The original piece contained a mathematical error concerning the burn rate, which has now been corrected. Welcome to another premium issue! Please consider subscribing; this extensive, approximately 7000-word piece exemplifies the depth of analysis you receive every week with your subscription.
A few days ago, Sam Altman stated that OpenAI's revenues would be "well more" than $13 billion in 2025. This assertion raises questions, particularly given other outlets' reports indicating OpenAI generated only $4.3 billion through the first half of 2025. Estimating approximately a billion dollars per month, this suggests the company likely made around $8 billion by the end of September. This is an estimate; should I receive information to the contrary, I will promptly report it.
Nevertheless, OpenAI is also experiencing a significant cash burn. Recent public disclosures (as reported by The Register) from Microsoft revealed funding commitments to OpenAI totaling $13 billion, of which $11.6 billion had been funded by September 30, 2025. These disclosures further revealed that OpenAI incurred a $12 billion loss in the last quarter—Microsoft's Fiscal Year Q1 2026, encompassing July through September 2025. To be clear, these are actual, verifiable accounting figures, distinct from numbers leaked to reporters. While leaks aren't inherently problematic, any information appearing in SEC filings typically meets an exceptionally high standard of scrutiny.
These numbers absolutely do not support Sam Altman's claim to a group of reporters at a dinner in mid-August that OpenAI is "profitable on inference."
Let's delve into the specifics.
The Information reported that through the first half of 2025, OpenAI spent $6.7 billion on research and development, which likely includes servers for developing new artificial intelligence. A common explanation is that OpenAI is "spending so much on training that it's eating the rest of its margins." However, if this were the sole cause, it would imply OpenAI spent the equivalent of six months' training costs within just three months.
I believe a more plausible explanation is that OpenAI is allocating substantial funds to staff, sales and marketing (a staggering $2 billion alone in the first half of the year), real estate, lobbying, data, and, of course, inference. According to The Information, OpenAI held $9.6 billion in cash at the end of June 2025.
Assuming OpenAI lost $12 billion by the end of calendar year Q3 2025 and, generously, generated around $3.3 billion (or $1.1 billion a month) within that quarter, this would suggest OpenAI's operations cost them over $15 billion in just three months. Where, precisely, is this money going? And how do the published figures reconcile with Microsoft's disclosures?
Within three months, OpenAI's costs—if we are to believe the information leaked to The Information (whose reporting I respect)—transitioned from a net loss of $13.5 billion in six months to, I presume, a net loss of $12 billion in three months. While stock-based compensation likely contributes to losses, it represented only $2.5 billion in costs in the first half of 2025. The Information also reported that OpenAI "spent more than $2.5 billion on its cost of revenue," suggesting inference costs around that figure?
I genuinely don't know the full picture, but something is amiss, and today, I'm going to investigate it further. In this newsletter, I will reveal how OpenAI's reported revenues and costs do not align, pointing to an additional $4.1 billion in cash burn that has yet to be reported elsewhere.
Read the full story
Sign up now to read the full story and get access to all posts for paying subscribers only.
Subscribe
Already have an account? Sign in