Nvidia’s board said on 12 November that it expects “the company’s operations, customer commitments and product roadmap to continue without interruption”, in the release filed with its Form 8-K naming CFO Colette Kress interim chief executive. The 8-K and the release do not mention open models, Nemotron or the Open Secure AI Alliance.
On record are a July letter “@NVIDIA signed” with 234 other organisations, an alliance of 122 named partners, and 1,040 public models under Nvidia’s Hugging Face account, 128 of them created since the letter, by our count at 14:46 UTC.
Business Pill · Open Weights Explained
An evergreen Business Engineer explainer on open weights: renting a model through an API versus running the weights yourself. It teaches the concept only and does not discuss this story.
The key insight: As we read the documents, Nvidia’s open-model program is on record as company commitments: the company signed the July letter, its blog launched the alliance, and its Hugging Face account added 128 public models after the letter, by our count, while the 12 November 8-K does not address the program.
The letter, “Open Weights and American AI Leadership”, is dated 24 July 2026. Its signatory list names 235 organisations by our count, NVIDIA included, among them OpenAI, Google, Meta, Microsoft, Amazon, AMD, Intel, Hugging Face and Mistral.
The letter is addressed to policy, not to any one company’s releases. It asks for “expanding access to compute for startups and researchers, investing in shared training assets (datasets, tools, evaluation frameworks), and keeping the frontier plural by avoiding premature restrictions on open models that stifle competition or drive innovation overseas”.
It also says that “open weights carry real and distinct risks” and argues that “the right response to this risk is not to prohibit open weights”. The letter was shared on X on 24 July from the account of Nvidia’s then chief executive, in a post that reads: “For my first post, I’m sharing a letter @NVIDIA signed on why open models matter”. We covered the letter at the time in our piece on the open-weights coalition.
Public Hugging Face repositories under the organisation “nvidia” by month created, January to October 2026: models and datasets. Our count from the Hugging Face API at 14:46 UTC on 12 November 2026; repositories since deleted or made private are not included.
What Was Posted on X
For my first post, I’m sharing a letter @NVIDIA signed on why open models matter.
AI will transform every industry, power every company, and be built by every country.
Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty.
The world needs both frontier closed models and frontier open models.
Posted on 24 July 2026 from the account of Nvidia’s then chief executive, this shares the ‘Open Weights and American AI Leadership’ letter that @NVIDIA signed. The piece counts the letter’s 235 named organisations from the PDF linked in the post.
Rip. Now I have to explain why. Lots of work ahead.
Nathan Lambert, who writes the Interconnects newsletter, posted this at 14:06 UTC on 12 November, quoting a post from 11 November on the open-model cause in the US. It is Lambert’s view; Nvidia’s 8-K and release do not address open models.
What the Open Secure AI Alliance Set Out
Three days later, on 27 July, an NVIDIA blog post launched the Open Secure AI Alliance. The post names 122 inaugural partners by our count, NVIDIA included, from Amazon, Microsoft and IBM to CrowdStrike, Palo Alto Networks and Hugging Face. Of those names, 51 also appear on the letter, by our count of exact matches.
The post states Nvidia’s part: “NVIDIA is contributing open models, model weights, data and new agent harness research to the Open Secure AI Alliance to speed the development of new cybersecurity tools and techniques”. It released the NOOA agent-harness framework on GitHub the same day, and asked regulators to “recognize open models, harnesses and security tooling as defensive assets, not liabilities”.
Organisations named on each Nvidia-backed open-model commitment in 2026: 235 on the open-weights letter of 24 July and 122 inaugural partners of the Open Secure AI Alliance in NVIDIA’s 27 July blog post, NVIDIA included, with 51 exact names on both; and the eight inaugural members named in the March Nemotron Coalition release. Our count of the documents as served on 12 November.
What Nvidia Has Shipped Since July
Nvidia’s Hugging Face account listed 1,040 public model repositories and 357 dataset repositories when we read the public API at 14:46 UTC on 12 November. Of those, 128 models and 56 datasets were created on or after 24 July, and 45 of those models carry the Nemotron name, by our count.
The models were downloaded about 34.8 million times in the last 30 days, the window Hugging Face counts, by our sum. October was the busiest month of the year, with 69 new public model repositories, by our count.
On 11 August Nvidia released Nemotron 3.5 Lightning, which its blog calls “a 30-billion-parameter mixture-of-experts model”. On 15 October it released Nemotron 3.5 Super VL, “an open 120-billion-parameter model”, with two edge models; the blog said “Super VL and Bolt 74B-A9B are available now, with Bolt E4B coming Oct. 23”, as set out in our piece on Nemotron 3.5 Bolt.
The October post says Super VL was “built in collaboration with the Nemotron Coalition”, the group of labs Nvidia launched in March with eight named inaugural members, including Mistral AI, Perplexity and Thinking Machines Lab.
What Today’s Filing Says
The 8-K says Kress, Nvidia’s CFO since 2013, “will continue in that role while she serves as the interim Chief Executive Officer”. The release says “The Board is initiating a search for a permanent Chief Executive Officer and will consider internal and external candidates”.
In the release, Chair Harvey Jones said Kress “is supported by an experienced leadership team that has been instrumental in executing NVIDIA’s strategy and advancing its technology roadmap”. Kress said: “We will continue advancing the work that has made NVIDIA a leader in accelerated computing and AI, and execute on the opportunities ahead”. The order of events is set out in our timeline of the CEO exit and our report on the 8-K.
How Open-Model Researchers Reacted
Nathan Lambert, who writes the Interconnects newsletter and previously co-led Ai2’s Olmo models, according to Lambert’s X profile, posted at 14:06 UTC: “Rip. Now I have to explain why. Lots of work ahead.” The post quotes the same account’s post of 11 November: “Jensen no longer being CEO would be very very bad for the open model cause in the US.” Interconnects AI is one of the letter’s signatories.
That is Lambert’s view. Nvidia’s filing and release do not address it.
The Structural Read
As we read it, the board’s roadmap sentence is the only line in the 12 November documents that bears on the program, and it speaks of the product roadmap in general, not of open models.
As we read Lambert’s post, it is a view about the open-model cause in the US, not a report of any change at Nvidia; the 12 November documents report none.
NVIDIA blog, 27 July 2026
“NVIDIA is contributing open models, model weights, data and new agent harness research to the Open Secure AI Alliance to speed the development of new cybersecurity tools and techniques.”
Three Implications
FOR DEVELOPERS Nvidia’s Hugging Face account listed 1,040 public models and 357 datasets at 14:46 UTC on 12 November, by our count; Nemotron 3.5 Lightning, Super VL and Bolt were released between August and October, per Nvidia’s blog.
FOR THE ALLIANCE’S PARTNERS The 27 July post lists Nvidia’s contribution as open models, model weights, data and agent harness research, and names 122 partners, by our count.
FOR SHAREHOLDERS The board said it expects the product roadmap to continue without interruption, and its search for a permanent chief executive will consider internal and external candidates, per the release.
The framework states the pattern it sees in platform shifts: “Close the SCARCE layer (keep proprietary) + Open the ABUNDANT layer (commoditize)”.
As we read it, Nvidia’s July documents put the company on the open side of the model layer. The framework predates the filing and does not discuss it.
What Is Not Established
Neither Kress nor Jones mentioned open models in the release, and neither the letter nor the alliance post lists dated release commitments by Nvidia. We did not contact Nvidia or the alliance’s members for this piece.
This Week In The AI Supercycle
The Great Convergence of Capital, Power, and Geopolitics
This week’s issue: for two years the AI build-out was priced as a supply story, and last week as a credit story. This week it was priced as a permission story, and the answer arrived as a cost, not a ban.
On record are a 24 July letter “@NVIDIA signed” with 234 other organisations, an alliance post naming 122 partners, and 1,040 public models on Hugging Face, 128 of them created since the letter, by our count. Today’s 8-K and release do not mention it; the board said it expects the product roadmap to continue without interruption.
A note on sourcing. The board and interim chief executive quotes come from Nvidia’s Form 8-K and its Exhibit 99.1 release of 12 November 2026, which we read in full. The letter, its signatory list and the alliance post were read in full as served on 12 November; the Nemotron details come from NVIDIA’s own blog and newsroom. Signatory counts, overlaps and the Hugging Face figures are our own counts from the public documents and the Hugging Face API at 14:46 UTC. The X posts were read in full. Nvidia was not contacted for this piece.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
