“I must repeat myself: AI is more expensive today than it was three years ago, and it is not getting cheaper.”
Our reading, the sentence that was checked:
AI costs more today than it did three years ago and is not getting cheaper.
Assessment: Unresolved Evidence as of
Where and when Ed Zitron wrote it
Where's Your Ed At, AI Doesn't Have ROI, paragraph 29.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment wyea-ai-doesnt-have-roi-e16d98-a2
We recorded 18 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 2026-06-02. This page does not claim the wording was captured then.
What the claim commits to
- Kind
- Observation
- 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
- AI is more expensive today
- than it was three years ago
- it is not getting cheaper
Our interpretation
- Today means June 2, 2026; three years ago means approximately June 2023.
- In the surrounding discussion of inference and customer bills, the plain reading is the cost of obtaining comparable useful work from generative AI.
- The comparison must account for total token consumption and repeated calls, not just token prices.
- Not getting cheaper is read as an ongoing trend at the post date, not an unlimited forecast about all future prices.
- The claim concerns service costs broadly, not a specified company's financial metric.
Counts as supported if: Comparable task-cost measurements show higher costs around June 2026 than around June 2023, with no meaningful decline in the period immediately preceding the post.
Counts as contradicted if: Comparable measurements show lower task costs around June 2026 than around June 2023, or a clear decline in comparable task costs immediately before the post.
Retrospective: these criteria were written after the outcome was public.
Current assessment
Was it true for the period described?
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
Epoch's September 22, 2026 study estimates that the cost of achieving a fixed level of AI performance fell about 47% per quarter since 2023 (e1). That estimate includes models released after June 2, 2026, so it cannot establish the change over the claim's specified period (e1). Without matched costs around June 2023 and June 2, 2026, or a separate measurement of the immediate prepublication trend, this evidence leaves the claim unresolved (e1).
Limitations
- The packet does not isolate matched total task costs at the required endpoints or the immediate prepublication trend.
- Fixed-performance benchmark costs do not establish representative customer bills, maximum-performance usage costs, or provider production costs.
Evidence (1 item)
Context only: Epoch AI; Luke Emberson and David Roodman,
47% per quarter
Why it matters here: The study counts task-level expense and estimates substantial declines at fixed performance. Its aggregate period includes models released after June 2, so the headline estimate is material contrary context, not a decisive measurement of the claim's exact period.
- Publisher
- Epoch AI; Luke Emberson and David Roodman
- Kind of source
- Independent measurement
- Describes events of
- Public from
What would settle it
Independent, reproducible historical comparisons using unchanged tasks and completion standards around June 2023 and June 2, 2026, plus a comparable series immediately before publication. Count input, output, reasoning, caching, retries, and agent calls, and distinguish customer charges from provider production costs. Lower endpoint costs or a clear immediate prepublication decline would contradict the claim; higher endpoint costs with no meaningful immediate prepublication decline would support it.
Where it was said (1)
Said once in the sources this register covers.
: Where's Your Ed At, AI Doesn't Have ROI
I must repeat myself: AI is more expensive today than it was three years ago, and it is not getting cheaper.
- Where
- paragraph 29
- Published
- 2026-06-02
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment wyea-ai-doesnt-have-roi-e16d98-a2: 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.