“Generative AI is extremely expensive because even the simplest prompts require GPT-4 to run highly-complex mathematical equations on GPUs that cost $10,000 apiece.”
Our reading, the sentence that was checked:
Generative AI is extremely expensive because even simple GPT-4 prompts require complex computation on $10,000 GPUs.
Assessment: Unresolved Evidence as of
Where and when Ed Zitron wrote it
Where's Your Ed At, Expectations Versus Reality, paragraph 18.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment wyea-expectations-versus-reality-474ae0-a1
We recorded 24 claims from this post. See every claim from this post.
The wording above is the version this site observed on 2026-10-28. The source may have been edited after 2024-05-13. This page does not claim the wording was captured then.
What the claim commits to
- Kind
- Causal claim
- Deadline, as the speaker wrote it
- None stated. This site does not add one.
- Deadline state when assessed
- No horizon
- Deadline state on 2026-10-29
- No horizon (the release’s reference date)
From the speaker
- Generative AI is extremely expensive
- even the simplest prompts
- GPT-4
- highly-complex mathematical equations
- GPUs that cost $10,000 apiece
Our interpretation
- Assess GPT-4 serving infrastructure and costs as of May 13, 2024.
- Read $10,000 as an approximate acquisition price per GPU, not a cost per prompt.
- Read 'extremely expensive' as substantial serving expense caused materially by computation and hardware requirements; no numerical expense threshold is supplied.
- The sentence names no source or accounting basis for the price. The contract therefore records unknown rather than assuming a reported estimate.
- The monetary scope records the specified unit acquisition price as a stock valuation snapshot; the unquantified serving expense remains a separate measurement gap.
Counts as supported if: Matching evidence identifies the GPUs actually serving GPT-4, their approximately $10,000 acquisition price, and measurements showing that the required computation materially produces the claimed high serving expense even for simple prompts.
Counts as contradicted if: Matching infrastructure and cost evidence establishes that the stated GPU requirement, approximate unit price or causal explanation is materially false.
Retrospective: these criteria were written after the outcome was public.
Current assessment
Does the evidence support the cause?
Outcome: Unresolved
The record was checked, and the evidence available by the cutoff does not settle it.
Scored on a plain reading. The speaker named no measure for this claim, so it was judged on the reading an informed reader would take, stated below under our interpretation. Both reviews had to find that reading fair.
- Evidence as of
- Deadline state when assessed
- No horizon
- Assessment recorded
- Policy version
- 2026-10-28.7
Rationale
The broad explanation is plausible: Microsoft describes GPU-based inference infrastructure, and CNBC reported roughly $10,000 A100 prices. Neither source establishes the complete claim about GPT-4's May 2024 hardware and serving expense. The unspecified financial basis also prevents a supported or contradicted score under the money-claim rules.
Limitations
- Market prices are not OpenAI purchase records.
- Hardware purchase cost does not directly measure expense per request.
- The claim provides no threshold for 'extremely expensive'.
Evidence (2 items)
Context only: CNBC,
These GPUs aren't cheap.
Why it matters here: The article reports an approximate $10,000 market price for NVIDIA A100 GPUs. That makes the price plausible as background, but does not establish OpenAI's purchase price or GPT-4 serving configuration in May 2024.
- Publisher
- CNBC
- Kind of source
- News report
- Measures
- NVIDIA, Reported market acquisition price of an A100 GPU, USD USD per GPU, estimate, 2023-02-23 to 2023-02-23
- Describes events of
- Public from
Context only: Microsoft,
optimized for inferencing
Why it matters here: Microsoft describes GPU-based serving and explains how networking and optimization affect cost. It does not disclose GPT-4's cost per simple prompt or a matching GPU acquisition price.
- Publisher
- Microsoft
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Describes events of
- Public from
What would settle it
A specified financial basis and dated records identifying GPT-4's serving hardware, acquisition prices, utilization and measured serving costs for simple prompts.
Where it was said (1)
Said once in the sources this register covers.
: Where's Your Ed At, Expectations Versus Reality
Generative AI is extremely expensive because even the simplest prompts require GPT-4 to run highly-complex mathematical equations on GPUs that cost $10,000 apiece.
- Where
- paragraph 18
- Published
- 2024-05-13
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment wyea-expectations-versus-reality-474ae0-a1: Current. Outcome: Unresolved. Evidence as of 2026-10-28. Public date not recorded.
Recorded events
Assessment publishedBy this site
First assessment published after both reviews and the final-text check passed.
About assessment wyea-expectations-versus-reality-474ae0-a1.