“And because OpenAI (and the competition) are so deep in the hole with transformer-based models, I believe they will continue to drive billions into them, burning money on training them using data that may or may not have been legally acquired”
Our reading, the sentence that was checked:
OpenAI and its competitors will keep pouring billions into training transformer models on possibly illegally acquired data.
Assessment: Unresolved No deadline stated on 2026-10-29 Evidence as of
Where and when Ed Zitron wrote it
Where's Your Ed At, How Does OpenAI Survive?, paragraph 137.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment wyea-to-serve-altman-08355a-a1
We recorded 38 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 2024-07-29. This page does not claim the wording was captured then.
What the claim commits to
- Kind
- Forecast
- Deadline, as the speaker wrote it
- None stated. This site does not add one.
- Deadline state when assessed
- No deadline stated
- Deadline state on 2026-10-29
- No deadline stated (the release’s reference date)
From the speaker
- OpenAI (and the competition)
- will continue to drive billions into them
- burning money on training them
- data that may or may not have been legally acquired
Our interpretation
- The plain reading is continued multibillion-dollar investment in transformer training by OpenAI and competing developers after publication.
- The article's preceding financial discussion relies on press estimates and includes training data acquisition and preparation within training costs; estimate is therefore the accounting basis.
- Billions means at least two billion dollars, but the claim does not say per model, per company, or per year. I do not import the screens' annual threshold.
- May or may not have been legally acquired makes no definite prediction of illegality.
- The proposed sunk-cost motive is not established merely by observing continued spending.
Counts as supported if: Records establish continued post-publication spending of billions on transformer training by OpenAI and competing developers, with sufficient separation from inference, unrelated research, and infrastructure commitments.
Counts as contradicted if: A decisive event establishes abandonment of the relevant training investment before the predicted continued multibillion-dollar spending occurs.
Retrospective: these criteria were written after the outcome was public.
Current assessment
Did the forecast come true?
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 deadline stated
- Assessment recorded
- Policy version
- 2026-10-28.7
Rationale
The evidence is consistent with continued spending on the scale Zitron described. In particular, a secondary estimate assigns Anthropic $4.1 billion of training expenditure in 2025. But the reviewed OpenAI figures mix periods or broader spending categories, and the competitor figure lacks an architecture breakdown. These records do not meet the stated early-resolution rule across the claim's scope. The uncertainty about data legality is not a prediction that illegality occurred.
Limitations
- Broad R&D and compute totals are not counted as training-only expenditure.
- Secondary estimates lack directly inspected underlying financial records.
- A collective industry reading would be easier to satisfy than a separate-company reading.
Evidence (2 items)
Context only: Epoch AI,
billions of dollars
Why it matters here: The analysis distinguishes estimated training and research compute from inference. Its annual figures include months before the post and projections for the final quarter, so they do not establish realized post-publication training spending.
- Publisher
- Epoch AI
- Kind of source
- Independent measurement
- Measures
- OpenAI, total cloud compute expenditure, USD billions, estimate, 2024-01-01 to 2024-12-31
- Describes events of
- Public from
Context only: Epoch AI,
$4.1B
Why it matters here: This supports the plausibility of continued multibillion-dollar competitor training expenditure. It relies on press reporting, does not cover OpenAI, and does not allocate spending by architecture.
- Publisher
- Epoch AI
- Kind of source
- Secondary summary
- Measures
- Anthropic, model training expenditure interpreted as R&D compute, USD billions, estimate, 2025-01-01 to 2025-12-31
- Describes events of
- Public from
What would settle it
Records identifying incurred post-publication transformer-training expenditure for OpenAI and competitors, with clear company and period aggregation.
Where it was said (1)
Said once in the sources this register covers.
: Where's Your Ed At, How Does OpenAI Survive?
And because OpenAI (and the competition) are so deep in the hole with transformer-based models, I believe they will continue to drive billions into them, burning money on training them using data that may or may not have been legally acquired
- Where
- paragraph 137
- Published
- 2024-07-29
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment wyea-to-serve-altman-08355a-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.