Scores by kind of claim
Each kind of claim is scored on its own, out of 100: the share that held up, with the range it likely falls in. Fewer resolved claims give a wider range.
-
Factual claims
Statements about the present or the past, checked against the record: did they hold up?
79 72 85 -
Predictions
Statements about events outside the person's control, checked once they fell due: did they come true?
80 72 88 -
Promises
Commitments to the person's or organisation's own future action, with something to deliver: was it delivered?
Insufficient record: 1 promise resolved, 5 needed for a score.
All types combined: 79 (likely 74 to 84, high confidence), from 102 claims: 64 factual claims, 38 predictions. It mixes kinds of claim in different proportions for different people, so compare people kind by kind.
Gary Marcus's claims
52 claims
“The company has not returned any profits, and is not expecting to do so until at least 2029, but has a burn rate in the billions, and still lacks a clear technical moat.”
Gary Marcus, Substack author 1 January 2025 25 AI Predictions for 2025, from Marcus on AI (Marcus on AI)
OpenAI has returned no profits, does not expect profits until at least 2029, has a burn rate in the billions, and lacks a clear technical moat.
Mostly true True in substance, wrong in a detail.
What would count as true or false
This is a factual claim judged as of 2025-01-01. It has four parts about OpenAI: (1) it had made no profit to date; (2) it projected no profit until 2029 or later; (3) its annual cash burn or losses were in the billions of dollars; (4) it lacked a clear technical moat. Parts 1-3 are checked against reports from before 2025-01-01 (for example The Information, NYT or WSJ reporting OpenAI's internal projections of about $5B in losses for 2024 and profitability around 2029) or company statements. Part 4 is a judgment and holds unless there is broad expert consensus at the time that OpenAI had a durable technical lead that competitors could not match. True: parts 1-3 are backed by reporting and part 4 is reasonably supported, for example competitors such as Anthropic, Google and Meta had models of comparable capability. Mostly true: one detail is a little off, such as profit projected for 2029 rather than after it, or burn reported as losses rather than cash burn. Mixed: one of parts 1-3 is clearly wrong. Mostly false: two or more of parts 1-3 are wrong. False: OpenAI was reported profitable, or was projecting profit well before 2029, and its burn was far below $1B.
Evidence
- OpenAI Said to See Losses Triple to $14 Billion in 2026, Unprofitable Until 2029 | TMTPOST TMTPost · 3 October 2024
“OpenAI may not achieve profitability until 2029, according to a report cited the internal documents.”
- OpenAI Said to See Losses Triple to $14 Billion in 2026, Unprofitable Until 2029 | TMTPOST TMTPost · 3 October 2024
“OpenAI will remain unprofitable until 2029 and its could incur losses up to $14 billion in 2016, nearly tripling this year’s expected loss”
- Burning Billions: OpenAI’s High-Stakes Financial Gamble — Can It Pay Off? CMSWire · 25 October 2024
“add in $2 billion for inference, $700 million for salaries, $500 million for data and various other costs, and OpenAI is projecting a $5 billion loss this year.”
- OpenAI Said to See Losses Triple to $14 Billion in 2026, Unprofitable Until 2029 | TMTPOST TMTPost · 3 October 2024
“OpenAI calculated net losses of $3 billion for the first half of the year, while its steep net losses mainly result from expenses like stock compensation and computing costs, which don’t flow through its cash statement.”
Reasoning
Reporting from before 2025-01-01 says OpenAI was projecting a $5 billion loss for 2024 (E3), had net losses of $3 billion in the first half of the year (E4), and did not expect to be profitable until 2029 (E1, E2). Together these support parts 1 and 2 and show losses in the billions. However, E4 notes that much of the net loss came from stock compensation and computing costs that don't pass through the cash statement, so the "burn rate in the billions" is backed by reported losses rather than cash burn figures. That matches the "mostly true" detail in the criteria. No evidence points to a consensus that OpenAI had a durable technical lead, so under the criteria part 4 holds by default.
Cite this claim
Gary Marcus, “OpenAI has returned no profits, does not expect profits until at least 2029, has a burn rate in the billions, and lacks a clear technical moat.” (said 1 January 2025). Verdict: mostly true. Reliability API, claim c_40d96612ad85, method 0.1.0+c876030c. https://epistacklabs.com/claims/c_40d96612ad85
Claims without a verdict
Is a verdict here wrong, or do you speak for Gary Marcus? Reply on the claim's own page, with the form at its foot (how replies work). Verdicts are made automatically; how verdicts and scores are made.