“OpenAI’s models and products — and we'll get into their utility in a bit — are deeply unprofitable to operate”
Our reading:
OpenAI's models and products are deeply unprofitable to operate.
Assessment: Supported
What happened
Was it true for the period described?
Outcome: Supported
The evidence matches what they said.
Judged on a plain reading. They named no measure, so we used the reading a typical informed reader would take. It is spelled out under “How we judged it”.
Evidence up to . Checked .
Why
The 2024 estimates independently confirmed by CNBC put operating-related losses at about $5 billion against $3.7 billion in revenue. That supports substantial aggregate unprofitability under the reading supplied by the surrounding discussion of compute, staffing and training. Estimated inference expenses alone were below revenue, so the evidence does not support a stronger claim that serving every model or product was itself loss-making.
Caveats
- Support is for aggregate profitability on the annual-estimate basis used in the post.
- CNBC confirmed the figures through a source but did not inspect the underlying financial statements.
- The result depends on including attributable development and operating costs.
Evidence (2 items)
Supports the claim: CNBC,
$5 billion in losses
Why it matters: The report confirms a substantial estimated annual loss relative to revenue. It attributes the loss to running services, salaries and rent, while noting that equity compensation and some other expenses were excluded.
- Publisher
- CNBC
- Kind of source
- News report
- Measures
- OpenAI, Aggregate operating loss after attributable model and product costs, USD billions, estimate, 2024-01-01 to 2024-12-31
- About events on
- Published
Context only: Epoch AI,
compute in 2024
Why it matters: The analysis reports estimated inference expenses of $1.8 billion against approximately $3.7 billion in revenue. This qualifies an inference-only reading, but inference expenses exclude the training and other operating costs covered by the contract.
- Publisher
- Epoch AI
- Kind of source
- Secondary summary
- Measures
- OpenAI, Inference compute expenses, USD billions, estimate, 2024-01-01 to 2024-12-31
- About events on
- Published
How we judged it
- Kind of claim
- Observation
- Deadline
- None given. We don’t add one.
Where we had to interpret
- The context expressly uses annual 2024 estimates, so the target is calendar 2024 rather than an invented first-half reporting period.
- The ordinary reading is substantial aggregate operating losses after costs attributable to developing and delivering the products, including compute and staffing.
- The claim does not require every product or individual inference request to have a negative gross margin.
- Deeply means losses substantial relative to revenue. The speaker supplies no numerical margin threshold.
- The accounting basis is estimate because the argument relies on press estimates. Full-year estimates concern annual activity, not an annualized revenue run rate.
Right if: Comparable 2024 estimates show that revenue from OpenAI's model and product business falls substantially short of its attributable operating costs.
Wrong if: Comparable 2024 estimates show aggregate operating profitability, break-even performance, or only a small operating deficit relative to revenue.
We wrote this test after the outcome was known.
We found 56 claims in this post. See them all.
Full record
Source and dates
Where's Your Ed At, The Subprime AI Crisis, paragraph 22.
- Said
- We read the source
- live page, full text, public
- Evidence up to
- Checked
We read this wording on 2026-10-28. The post may have been edited after 2024-09-16.
Where it was said (1)
Said once in the posts we read.
: Where's Your Ed At, The Subprime AI Crisis
OpenAI’s models and products — and we'll get into their utility in a bit — are deeply unprofitable to operate
- Where
- paragraph 22
- Published
- 2024-09-16
- We read it
- , live page
- Access
- full text, public
History
Versions
- Current: Supported. Evidence up to 2026-10-28.
What happened to this record
PublishedBy us
Published after both AI reviews and the final check passed.
Held backBy us
A reviewer said our test wasn’t fair to what was actually said.