“And right now, as I’ve covered, there’s not enough compute being built to keep up with Anthropic or OpenAI’s voracious demands, meaning that they will both be bartering to buy whatever’s available at whatever price it’s available at. This naturally will savage their already-negative margins…”
Our reading:
Because not enough compute is being built, Anthropic and OpenAI will pay whatever prices they must for compute, further damaging their already-negative margins.
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
InfoWorld reported Anthropic's capacity restrictions on March 27, 2026, and CNAS described chip supply constraints on May 7 [e1, e2]. OpenAI acknowledged demand growing faster than capacity on July 29, after the May 19 forecast [e3]. These items support compute supply pressure, but do not show that shortages forced both companies to purchase compute on materially worse terms [e1, e2, e3]. They also do not establish already-negative overall profit margins or subsequent material deterioration caused by those purchases [e1, e2, e3]. Anthropic announced some usage-limit relief on May 6, before the forecast [e4]. That relief establishes neither subsequent margin effects nor a permanent end to shortages [e4]. The evidence therefore does not establish or permanently foreclose the predicted financial sequence for both companies [e1, e2, e3, e4].
Caveats
- No matched post-publication records connect shortages, worse comparable purchase terms and resulting margin deterioration for both companies.
- The evidence does not establish the claimed negative starting margins on a consistent accounting basis.
- The claim specifies neither a deadline nor a margin measure or numerical threshold for material damage.
- Temporary capacity relief cannot settle this forecast with no stated deadline.
Evidence (4 items)
Supports the claim: Center for a New American Security,
binding constraint
Why it matters: The report identifies a supply constraint on compute construction. Together with the company-specific items, it supports the antecedent, not the predicted financial damage.
- Publisher
- Center for a New American Security
- Kind of source
- Secondary summary
- About events on
- Published
Supports the claim: InfoWorld,
compute capacity
Why it matters: Anthropic faced demand pressure sufficient to restrict usage. This establishes a capacity constraint, not premium procurement or a subsequent margin decline.
- Publisher
- InfoWorld
- Kind of source
- News report
- About events on
- Published
Supports the claim: OpenAI,
compute-constrained world
Why it matters: OpenAI acknowledges the capacity constraint while discussing its inference infrastructure. The disclosure does not quantify procurement premiums or the resulting effect on profit margins.
- Publisher
- OpenAI
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- About events on
- Published
Context only: Anthropic,
removing the peak hours limit reduction
Why it matters: Anthropic reports easing a restriction alongside new compute agreements. This qualifies any assumption that shortages cannot ease, but neither establishes adequate total supply nor settles future margins.
- Publisher
- Anthropic
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- About events on
- Published
How we judged it
- Kind of claim
- Forecast
- Deadline
- None given. We don’t add one.
Conditions
- because compute supply is insufficient (their words)
Where we had to interpret
- The asserted shortage is the antecedent: available compute cannot keep pace with the labs' demand.
- Whatever price means materially worse purchasing terms forced by scarcity, not literally an unlimited willingness to pay.
- Savage means a material deterioration attributable to procurement costs, not an immaterial fluctuation.
- Already-negative margins most naturally refers to overall profitability because the preceding gross-margin figures are positive. Net profit margin is a provisional operationalization; the speaker does not specify it.
- No deadline is supplied.
Right if: After publication, both companies face constrained compute supply, purchase capacity on materially worse comparable terms because of that constraint, and experience a material resulting deterioration in consistently measured profit margins.
Wrong if: A decisive completed episode or permanent change establishes that the asserted shortage did not produce the predicted procurement and margin effects. Temporary improvement or mere nonoccurrence cannot defeat an unlimited-horizon forecast.
We wrote this test after the outcome was known.
What would settle it
Dated evidence after May 19, 2026 showing that shortages forced both companies to buy comparable compute on materially worse terms and materially worsened their already-negative profit margins. The records must use consistent margin measures and separate procurement effects from other costs. Alternatively, a decisive event must permanently foreclose that sequence.
We found 44 claims in this post. See them all.
Full record
Source and dates
Where's Your Ed At, AI Is Too Expensive, paragraph 78.
- 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 2026-05-19.
Where it was said (1)
Said once in the posts we read.
: Where's Your Ed At, AI Is Too Expensive
And right now, as I’ve covered, there’s not enough compute being built to keep up with Anthropic or OpenAI’s voracious demands, meaning that they will both be bartering to buy whatever’s available at whatever price it’s available at. This naturally will savage their already-negative margins…
- Where
- paragraph 78
- Published
- 2026-05-19
- 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: our summary didn’t match their words.