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
42 claims
“Profits from AI models will continue to be modest or nonexistent”
Gary Marcus, Substack author 1 January 2025 25 AI Predictions for 2025, from Marcus on AI (Marcus on AI)
Profits from AI models will remain modest or nonexistent in 2025.
Resolved true The record shows it happened or was so.
The second judge would not resolve this claim; the verdict shown is the first judge's.
What would count as true or false
A prediction for 2025, judged at year end. The claim: profits from AI models (the businesses of selling or serving foundation models, such as OpenAI and Anthropic, or model-provider segments) remain modest or nonexistent. True: the major model developers reported or credibly reported operating losses or negligible profits for 2025, and no major firm reported large profits specifically from AI models. Mostly true: the leading pure-play model makers were unprofitable, but some firm earned meaningful but modest model-specific profit. Mixed: the picture is split, with some major model providers substantially profitable from models and others not. Mostly false: several major providers reported substantial model-specific profits, though the industry overall was still loss-making. False: AI model businesses broadly reported large profits in 2025. Profits from chips (Nvidia) or cloud infrastructure count only if attributed to AI models themselves; the claim concerns models, not hardware.
Evidence
- Leaked financial docs show OpenAI is losing billions of dollars a year Ars Technica · 16 June 2026
“All told, OpenAI’s day-to-day “loss from operations” increased from $8.78 billion in 2024 to $20.92 billion in 2025, a concerning direction for a company that is telling investors it hopes to be profitable by 2030.”
- Anthropic's Revenue Jumped 12-Fold in 2025, but So Did Its Losses Yahoo Finance · 10 October 2026
“Anthropic's revenue surged nearly 12-fold to about $4.6 billion in 2025. But the artificial intelligence (AI) company also posted an operating loss of $8.06 billion and a net loss of roughly $42 billion.”
- Anthropic API Gross Margin Estimated Above 80% as Inference Margin Reaches Mid-60s | TokenPost TokenPost · 11 October 2026
“Anthropic’s blended inference gross margin was estimated at 38% in 2025 and negative 94% in 2024 before reaching the mid-60% range. The API business was estimated to have a gross margin above 80%.”
- Deep Dive: Economics of the AI Build-Out | Contrary Research Contrary Research · 28 August 2025
“Under its 2025 operating model, OpenAI is not expected to turn a profit until 2029, though Microsoft is likely to profit off its arrangement with the company in the meantime.”
Reasoning
E1 reports, based on leaked financial documents, that OpenAI's loss from operations rose to $20.92 billion in 2025. E2 reports that Anthropic posted a 2025 operating loss of $8.06 billion and a net loss of about $42 billion. E3 gives analysts' estimates of Anthropic's gross margins, not reported profits. None of the evidence shows a major firm reporting large profits from AI models. OpenAI and Anthropic were therefore loss-making in 2025, and none of the evidence shows large profits from AI models, which meets the 'true' criterion.
Cite this claim
Gary Marcus, “Profits from AI models will remain modest or nonexistent in 2025.” (said 1 January 2025). Verdict: resolved true. Reliability API, claim c_4c7558b0b806, method 0.1.0+c876030c. https://epistacklabs.com/claims/c_4c7558b0b806
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.