“These economics are also load-bearing for OpenAI and Anthropic, representing around 80% of their revenues, meaning that once they die, the AI labs’ underlying revenues begin to decay.”
Our reading:
AI startups account for about 80% of OpenAI's and Anthropic's revenues, so when they die the labs' revenues will decay.
Assessment: Unresolved
What happened
Did the forecast come true?
Outcome: Unresolved
We checked, and the evidence so far doesn’t settle it.
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
Anthropic's filing, available October 26, 2026, includes both AI-native startups and large organizations in its enterprise customer category [e2]. YipitData's October 27 release warns that spending shares in its sample are not companywide revenue shares [e3]. The supplied sources do not establish startups' share of each lab's total realized revenue for April–June 2026 [e1] [e2] [e3]. They also do not document failure of the relevant startup population followed by attributable declines in each lab's revenue [e1] [e2] [e3]. Both the 80% premise and the conditional forecast remain unresolved [e1] [e2] [e3].
Caveats
- Enterprise concentration and sampled spending do not establish startup-only shares of each lab's total revenue.
- The evidence does not establish that the relevant startup population failed or show subsequent revenue effects.
- The startup population, accounting basis and post-failure observation interval remain unspecified.
Evidence (3 items)
Context only: PYMNTS,
share of total revenue
Why it matters: The article's 80% figure concerns enterprise-revenue concentration among large customers. It does not establish that AI startups provide 80% of either lab's total revenue.
- Publisher
- PYMNTS
- Kind of source
- News report
- About events on
- Published
Context only: Anthropic, PBC / SEC,
Our customers fall into two primary categories: enterprises and prosumers.
Why it matters: The filing describes enterprises as including both startups and large established organizations. That classification cannot be converted into a startup-only revenue share or evidence that the failure antecedent occurred.
- Publisher
- Anthropic, PBC / SEC
- Kind of source
- Regulatory filing
- About events on
- Published
Context only: YipitData,
Panel spending shares should not be interpreted as companywide revenue shares.
Why it matters: The measurement provider explicitly limits what its spending sample can establish. Such concentration data cannot settle the claimed total-revenue dependence.
- Where
- Data and Methodology
- Publisher
- YipitData
- Kind of source
- Independent measurement
- About events on
- Published
How we judged it
- Kind of claim
- Conditional forecast
- Deadline
- None given. We don’t add one.
Conditions
- once AI startups die (their words)
Where we had to interpret
- Their revenues ordinarily means each lab's total revenue. Enterprise revenue, API revenue and spending observed in a payment panel are different denominators.
- The antecedent is failure across the revenue-significant startup customer population described in the surrounding passage, not the failure of any isolated startup.
- Decay means an absolute decline in realized revenue attributable to those customer failures, rather than merely slower growth.
- For the current 80% premise, use April through June 2026, the latest completed calendar quarter before publication. This is not a deadline for the forecast.
- The passage supplies no accounting basis or source for the revenue share.
Right if: Evidence establishes the startup-failure antecedent, approximately 80% startup dependence in comparable baseline accounts, and a subsequent decline in each lab's realized revenue attributable to those failures.
Wrong if: After the specified customer population has failed and its spending has ceased, comparable accounts establish that the predicted revenue decline did not occur. An incorrect baseline revenue share would separately contradict the premise, but would not establish that the antecedent occurred.
We wrote this test after the outcome was known.
What would settle it
Define the relevant startup population and measure its April–June 2026 share of each lab's total realized revenue. Document failures and cessation of customer spending, then compare subsequent realized revenue while accounting for other customers and price changes. This would test the revenue-share premise and whether the predicted decline followed the failures.
We found 54 claims in this post. See them all.
Full record
Source and dates
Where's Your Ed At, Dead Money, paragraph 153.
- 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 2026-09-29.
Where it was said (1)
Said once in the posts we read.
: Where's Your Ed At, Dead Money
These economics are also load-bearing for OpenAI and Anthropic, representing around 80% of their revenues, meaning that once they die, the AI labs’ underlying revenues begin to decay.
- Where
- paragraph 153
- Published
- 2026-09-29
- We read it
- , live page
- Access
- full text, public
History
Versions
- Current: Unresolved. 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
The final check failed: a cited source didn’t say what the write-up said, and the write-up said more than the evidence showed.