AI-written

Topic

NVIDIA

20 stories, newest first.

  1. California Attorney General Rob Bonta served NVIDIA with an investigative subpoena, a legal demand for information, on Nov. 19, a week after Jensen Huang resigned as NVIDIA's chief executive. His office is asking whether NVIDIA's or Huang's actions let advanced chips or confidential information reach China, and whether the company misled investors and the public about that risk. It says the answers will help it judge whether NVIDIA broke California law, and it has not said which documents it demanded or when they are due.

  2. The House Foreign Affairs Committee on Nov. 19 made public the U.S. Commerce Department's decisions on requests to sell NVIDIA's H200 and AMD's MI325X AI chips to Chinese companies. By May 11, the department had approved sales of 879,868 of about 3.4 million chips requested, including for Alibaba and Tencent, which the committee notes are on the Pentagon's list of Chinese military companies. The records mark every approval as not yet shipped at that date and do not show what has shipped since.

  3. NVIDIA says its sales reached $115.8 billion in the three months to Oct. 25, up 103% from a year earlier, with $107.6 billion of it from the AI chips and other hardware it sells for data centers. It expects $130 billion, give or take 3%, in the quarter that began Oct. 26, without counting on sales of its data center chips in China. These are its first results since Jensen Huang resigned as chief executive on Nov. 12, and the board has not yet named a permanent successor.

  4. Will DePue, who says he helped create OpenAI's Sora video tool while he worked there, started a small nonprofit research group called Unemployment Labs. Its report, published Nov. 12, says six AI models released since April by DeepSeek, Moonshot AI, MiniMax and Ant Group carry a pattern that matches NVIDIA's Blackwell chips, which need a U.S. license to be exported to China. The pattern comes from the files the companies published themselves. The report says the pattern shows which kind of chip set the models up, which supports but does not prove that the same chips did the training. It does not show where those chips were, how the companies reached them or whether any rule was broken.

  5. NVIDIA has released Alpamayo 2 Edge, an AI for self-driving cars that is built to run on NVIDIA's computer for cars and writes out why it makes its moves, such as yielding or making room, when driving stops being routine. Companies and researchers can download it for free and adapt it. It is a starting point for developers, not a finished self-driving system, and NVIDIA says builders must test it further before relying on it.

  6. Jensen Huang, who co-founded NVIDIA in 1993 and had run it ever since, resigned as chief executive and left its board, NVIDIA's board said on Nov. 12. Finance chief Colette Kress will run the AI chipmaker until the board picks a permanent successor. The board gave no reason; his exit comes four days after the board opened an investigation into reports that he was the target of a foreign intelligence effort.

  7. NVIDIA unveiled GR00T 2, an AI model for robots trained mostly on hundreds of thousands of hours of video of people at work, which NVIDIA says lets a robot carry out skills it was shown only by people or on other robots. In NVIDIA's tests on 50 tasks its robot had never practiced, the model succeeded 40% of the time with no practice and 95% after five demonstrations of each task. NVIDIA calls the model open, but its project page says the code arrives in December, and so far the results come only from NVIDIA.

  8. NVIDIA's board has opened an independent investigation after news reports that chief executive Jensen Huang was the target of a foreign intelligence effort, and has hired a former federal prosecutor to run it. The board says it has no reason to believe national security or NVIDIA's secrets were compromised, and Huang says he will cooperate. On Nov. 12, the board said Huang had resigned as chief executive and left the board, and named finance chief Colette Kress interim chief executive.

  9. NVIDIA on Wednesday released Nemotron 3.5 Transcribe, a free-to-download AI that turns recordings of up to an hour into text in 40 languages, labels who is speaking in 13 of them and translates between English and 31. Developers tell it in everyday language what kind of transcript they want. It works only on recordings, not live audio, and so far its accuracy figures come from NVIDIA's own tests and from rankings NVIDIA cites.

  10. Two German ministries signed a declaration of intent with NVIDIA to develop AI models that run factory machines and robots, starting from NVIDIA's open models and drawing on German industrial data. NVIDIA also plans to invest in four German startups. Neither side has said how much money the joint work involves or when the first models will be ready.

  11. NVIDIA released two free-to-download AI models on Thursday, Oct. 15, that are built to carry out tasks on their own, and a smaller third one on Oct. 23. Super VL is the large model, aimed at companies' servers, and Bolt is a cut-down version squeezed into 24 gigabytes of memory for the RTX Spark laptops NVIDIA said would arrive on Oct. 16. The comparisons are NVIDIA's own, and its own speed test shows Bolt writing code more slowly than two of three smaller rivals.

  12. NVIDIA says it will give British researchers access to computing power it values at $250 million over five years, or $50 million a year, for research on the safety and security of AI agents, AI that can carry out tasks on its own. Most of it is meant for the wider UK research community, with a share for labs led by Oxford and UCL and for Imperial College London. The cloud providers that will supply it, and the terms, have not yet been agreed.

  13. Microsoft and five other PC makers opened preorders on Wednesday for Windows laptops built around NVIDIA's new RTX Spark chip, which the companies said would ship Oct. 16, with Microsoft's own starting at $2,599. The companies say more AI work can then run on your own computer instead of a paid cloud service, without sending your data off the device. The speed claims come from the companies' own tests on early units, and outside testing has yet to show how the laptops hold up.

  14. NVIDIA's DGX Spark, a desktop computer for running AI models, now comes in a version with 64GB of memory, half as much as the original, which NVIDIA said would go on sale from $4,999 on Friday, Oct. 23. Less memory means it can't run AI models as large as the 128GB version, which NVIDIA's store lists from $5,999. Most partners hadn't announced prices at launch, and NVIDIA's speed figures come from its own tests.

  15. NVIDIA and CoreWeave said on Wednesday, Sept. 30, that NVIDIA's Vera Rubin NVL72 systems are now available on CoreWeave's cloud, with Cognition, the lab behind the Devin coding agent, the first customer running production work on them. CoreWeave will also offer NVIDIA Vera, which the companies call the first CPU built for AI agents. The speed gains they cite come from early tests.

  16. NVIDIA announced its Open Agent Safety Platform on Monday, Sept. 28, pairing OpenShell, free open-source software that sets limits on what AI agents can do, with Sentry, a watchdog on its BlueField-4 chip that NVIDIA says can isolate a runaway agent in milliseconds. NVIDIA presents it as a response to recent cases of agents slipping past the controls built into their software. OpenShell is broadly available now, while Sentry is a design for others to build, so how widely it will be used is not yet known.

  17. In an interview published Sept. 20, 2026, NVIDIA CEO Jensen Huang said there is "0% chance" that 2030 is "the end of the world" and called AI extinction warnings doomsday narratives that are "not grounded in science." He argued that existing U.S. product-liability and cybersecurity laws should be applied before any new AI rules. Whether lawmakers or other AI leaders accept that case is not yet clear.

  18. NVIDIA and Perplexity on Sept. 14, 2026, brought Portable Computer, Perplexity's AI agent that carries out multistep tasks on the user's own PC, to the Perplexity app for Windows. It needs a GeForce RTX or RTX PRO graphics card with at least 24GB of VRAM, the memory on the card, plus a Pro or Max subscription. The companies say work it finishes on the PC keeps sensitive files local and does not use up Perplexity's cloud credits.