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?
72 67 78 -
Predictions
Statements about events outside the person's control, checked once they fell due: did they come true?
Insufficient record: 1 prediction resolved, 5 needed for a score.
-
Promises
Commitments to the person's or organisation's own future action, with something to deliver: was it delivered?
Insufficient record: 4 promises resolved, 5 needed for a score.
-
Guidance
A company's own figures for its results over a period, checked against what it then reported.
Insufficient record: 1 guidance claim resolved, 5 needed for a score.
All types combined: 72 (likely 67 to 78, high confidence), from 101 claims: 101 factual claims. It mixes kinds of claim in different proportions for different people, so compare people kind by kind.
Jensen Huang's claims
49 claims
“It is trained on 20 million hours of video.”
Jensen Huang, NVIDIA CEO 7 January 2025 [Full transcript] NVIDIA CEO Jensen Huang Keynote at CES 2025 (y2doc)
NVIDIA Cosmos is trained on 20 million hours of video.
Mostly true True in substance, wrong in a detail.
The second judge said resolved true; the verdict shown is the first judge's.
The evidence is the person's or their organisation's own records.
Said for an organisation they lead or work for; it counts on both records.
What would count as true or false
Judged as of 2025-01-07. True: NVIDIA's official Cosmos announcement, technical paper or model documentation states that the Cosmos world foundation models were trained on about 20 million hours of video. Mostly true: official sources give a figure close to 20 million hours (within about 10%), or attach the 20 million hours to the Cosmos data curation pipeline or training dataset in a way that still supports the claim. Mostly false: the 20 million hours is real but describes something else (for example video that was processed or curated rather than used for training) and the training figure is substantially different, or the figure applies only to a subset of Cosmos models presented as the whole. False: official documentation shows a training figure that is materially different (more than 25% off) or unrelated to 20 million hours.
Evidence
- NVIDIA Makes Cosmos World Foundation Models Openly Available to Physical AI Developer Community NVIDIA Blog · 7 January 2025 Own record
“The models have been trained on 9,000 trillion tokens from 20 million hours of real-world human interactions, environment, industrial, robotics and driving data.”
- Advancing Physical AI with NVIDIA Cosmos World Foundation Model Platform | NVIDIA Technical Blog NVIDIA Technical Blog · 9 January 2025 Own record
“Cosmos world foundation models are pretrained large generative AI models trained on 9,000 trillion tokens—including 20 million hours of data from autonomous driving, robotics, synthetic environments, and other related domains.”
- NVIDIA Launches Cosmos World Foundation Model Platform to Accelerate Physical AI Development NVIDIA Corporation (GLOBE NEWSWIRE release) · 6 January 2025 Own record
“An NVIDIA AI and CUDA®-accelerated data processing pipeline, powered by NVIDIA NeMo™ Curator, that enables developers to process, curate and label 20 million hours of videos in 14 days using the NVIDIA Blackwell platform, instead of over three years using a CPU-only pipeline.”
Reasoning
NVIDIA's launch blog of 2025-01-07 (E1) says the Cosmos models "have been trained on 9,000 trillion tokens from 20 million hours" of real-world data. That supports the training figure, although the sentence does not use the word "video". NVIDIA's CES press release (E3) attaches the same 20 million hours figure to video that the Cosmos data pipeline can process, curate and label, but it does not state that this video was used to train the models. Both sources come from NVIDIA itself, and they support the claim closely but not exactly in its stated form. E2 is a later restatement and was not needed for the verdict.
Cite this claim
Jensen Huang, “NVIDIA Cosmos is trained on 20 million hours of video.” (said 7 January 2025). Verdict: mostly true. Reliability API, claim c_3fddd666d8e8, method 0.1.0+03d23c0e. https://epistacklabs.com/claims/c_3fddd666d8e8
Claims without a verdict
Is a verdict here wrong, or do you speak for Jensen Huang? 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.