Jensen Huang and Tech Giants Urge US to Support Open Weights AI Models

Text

Huang’s X Debut and the Open Weights Letter

Jensen Huang, CEO of Nvidia, joined the social media platform X in June 2026 and used his first post to promote a policy letter titled Open Weights and American AI Leadership, co-signed by 25 technology companies including Nvidia, Microsoft, Meta, IBM, and Hugging Face. The three-page document urges the U.S. government to avoid "premature restrictions on downloadable AI models" and highlights the importance of open-source AI development. The letter comes amid growing regulatory scrutiny of AI, including reports that the Trump administration is considering a ban on Chinese AI models, though the document does not explicitly mention China, Moonshot AI, or DeepSeek.

Text

Signatories and Omissions in the Open Weights Initiative

The letter’s signatories include chipmakers, cloud providers, enterprise software firms, and venture capital firms, such as Dell Technologies, Palantir, and Andreessen Horowitz. Notably absent are OpenAI, Anthropic, and Google, which have not publicly endorsed the initiative. The document emphasizes that open models "strengthen safety and cybersecurity, accelerate innovation, and enable sovereignty," arguing that both open and closed AI systems are necessary for progress. Huang, in his X post, stated, "The world needs both frontier closed models and frontier open models."

Text

Policy Demands and Regulatory Contrasts

The policy letter calls for expanded compute access for startups and researchers, along with public investment in shared datasets and evaluation frameworks. It also addresses the practice of model distillation—training one model on another’s outputs—urging policymakers to differentiate between lawful and unlawful extraction. The document argues that broad restrictions on distillation could stifle innovation, while targeted legal measures should handle cases of intellectual property violations. This stance contrasts with remarks by Treasury Secretary Scott Bessent, who told Fox Business on July 22, 2026, that the administration is investigating Chinese open-source models for potential intellectual property theft and could impose sanctions.

Text

Nemotron 3 Ultra’s Release and Industry Reactions

Huang’s advocacy for open models aligns with Nvidia’s recent actions. In June 2026, the company released the 550-billion-parameter Nemotron 3 Ultra model under the OpenMDW-1.1 license, managed by the Linux Foundation. This license allows users to download and modify the model’s weights, a practice that critics argue could undermine proprietary AI systems. Artificial Analysis, a third-party evaluation firm, scored Nemotron 3 Ultra at 47.7 on its intelligence index, compared to 53.9 for Moonshot’s Kimi K2.6 model. The letter’s emphasis on open weights reflects a broader industry push to democratize AI access, though it faces resistance from companies prioritizing closed, proprietary systems.

Text

Regulatory Shifts and National Security Concerns

The regulatory landscape for AI is rapidly evolving. A separate report from July 2026 revealed that the Trump administration is revisiting plans to restrict Chinese AI models, citing concerns over "unfair competition" and national security. However, Huang countered these claims in a July 22, 2026, interview with Axios, stating that American firms should be allowed to use Chinese models and dismissing allegations of "backdoors" as "a misconception." This tension underscores the global competition for AI leadership, with the U.S. and China vying to shape policies that favor their respective industries.

Text

Decentralization of AI and Governance Challenges

The Open Weights letter also highlights the growing role of open-source models in AI development. At Nvidia’s CES 2026 press conference, Huang noted that one in every four AI-generated tokens now comes from an open model. These models are distributed through enterprise clusters, regional clouds, and on-premises infrastructure, rather than relying on a handful of hyperscaler API endpoints. This decentralization reduces dependency on large cloud providers and allows organizations to tailor AI systems to their specific needs. However, it also raises questions about governance, security, and the long-term sustainability of open-source AI projects.

Text

Industry Divisions and the Future of AI Regulation

The absence of major players like OpenAI and Google from the letter has sparked speculation about industry divisions. OpenAI’s GPT-4 and Google’s Gemini models remain closed-source, emphasizing proprietary control over AI development. In contrast, Meta’s Llama series and Hugging Face’s models have embraced open weights, fostering a vibrant ecosystem of third-party developers. The letter’s authors argue that open models enhance transparency and collaboration, but critics contend that they may hinder the commercial viability of AI startups.