Bright Info Get your data valued

AI lab

NVIDIA: what they buy and what they pay

NVIDIA is profiled here as an AI lab[1]. Data types cited in this profile include: Text, Images, Video, Audio[1].

Last verified November 3, 2026

Quick facts

TypeAI lab[1]
StatusActive[1]
Websitehttps://www.nvidia.com/[2]
HQSanta Clara, California, USA[2]
Founded1993[2]
OwnershipNot published
ProgramNot published
PriceNot published
Pricing basisNot published
Payment timingNot published
How to startNot published

What they buy

Data typesText, Images, Video, Audio[1]Translation memories; Exam/reasoning items; Competitive-programming problems and solutions; Speech; Agentic task, tool-use and dialogue datasets[1]
ExcludedNot published
Intended useTraining (Nemotron 3 / 3.5 model family)[1]

Track record

Named data sources (from NVIDIA's EU AI Act training summary)

  • Named private datasets NVIDIA lists as obtained from third parties for Nemotron 3/3.5 (Section 2.2.2, separate from rightsholder licences; the summary does not say whether they were bought): Global Regulation; TAUS Translation Memory; Scale HLE; HackerRank Coding; Granary; SIFT-50M; Mercor (agentic terminal-task and agentic-search tool-use data); Turing Math Data Pack; Patronus (dialog memory data); "Droid Harness Pivot" data; "Chemistry Ether0" data[1]

Coverage

  • TorrentFreak, Jan 2026: amended complaint alleges a member of NVIDIA's data strategy team contacted Anna's Archive about acquiring its millions of pirated materials (reported)[3]

Watch for

  • Lawsuit: Nazemian et al. v. NVIDIA Corporation, No. 4:24-cv-01454 (N.D. Cal.), authors' proposed class action alleging use of their copyrighted works to train NeMo Megatron and other LLMs; court denied NVIDIA's motion to dismiss in part (per plaintiffs' law firm, May 2026) (reported)[4]
  • Allegation in amended complaint (reported Jan 2026): a member of NVIDIA's data strategy team contacted Anna's Archive, and NVIDIA sought to find out what 'high-speed access' to its data would look like; the complaint does not say whether NVIDIA paid. Anna's Archive later said on Reddit it never dealt with NVIDIA directly.[3]
  • Report (Proof News, Aug 2024): according to company Slack and internal documents, NVIDIA staff gathered YouTube videos for a new video model by downloading datasets of previously scraped videos and scraping their own; NVIDIA said it is confident its models and research comply with copyright law (reported)[5]

Sources

  1. Public Summary of Training Content for NVIDIA Nemotron (PDF; section 'Public Summary of Training Content for NVIDIA Nemotron 3 and 3.5 Family', Version 1.0, Last update: 10/08/2026), developer.download.nvidia.com. Accessed October 30, 2026. (company's own site)
  2. NVIDIA Corporation Form 10-K for fiscal year ended January 25, 2026, sec.gov. Accessed October 30, 2026. (filing)
  3. NVIDIA Contacted Anna's Archive to Secure Access to Millions of Pirated Books - TorrentFreak, torrentfreak.com. Published January 19, 2026. Accessed November 3, 2026. (press)
  4. Susman Godfrey Advances AI Copyright Case as Court Denies NVIDIA Motion to Dismiss | Susman Godfrey L.L.P., susmangodfrey.com. Published May 8, 2026. Accessed November 3, 2026. (press)
  5. Nvidia Scrapes YouTube, Eyes Netflix, Discovery to Train New Video Model - Proof News, proofnews.org. Published August 9, 2024. Accessed November 3, 2026. (press)

Last verified November 3, 2026