Jensen Huang Posts First X Post, Joins Meta, Microsoft and Other Tech Giants to Call for Open-Source AI Models

TradingKey - Nvidia ( NVDA) CEO Jensen Huang posted his first message on social media platform X on Friday, calling for open-source AI models.

The article points out that early open-source pioneers shattered the traditional concept that 'software progress can only rely on companies strictly controlling code,' establishing a transparent ecosystem that global developers can study, modify, and improve. Today, open-source software supports most of the internet, the core systems of the world's largest tech companies, and even mission-critical systems of the U.S. military and federal agencies.

Huang and the co-signatories believe that the U.S. currently faces a similar choice in the AI field. The open letter explicitly states that the measure of U.S. AI leadership is not how powerful a single frontier model is, but whether a robust, open ecosystem that permeates all industries can be built. Open weights models are the very foundation of this ecosystem.

[Source: X]

Notably, the signatories span chip giants (NVIDIA, Dell), foundation model developers (Meta, Microsoft, Mistral, Hugging Face), application-layer companies (Perplexity, Replit, Box, ServiceNow, Palantir, CrowdStrike), top investment firms (Andreessen Horowitz, Y Combinator, Emergence Capital), and industry organizations (The Linux Foundation, Mozilla, IBM), indicating that the open weights issue has forged a broad consensus across all segments of the industry chain.

Three Major Strategic Values of Open Weights

Expanding economic access to AI, allowing frontier capabilities to penetrate thousands of industries. The article points out that open weights allow startups, established enterprises, universities, and public institutions to build applications on top of existing advanced models, eliminating the need to train from scratch or pay the high prices of frontier models for every task. The key to this logic lies in a "division of labor"—letting frontier-class models focus on truly cutting-edge problems while other tasks are run by efficient, small, specialized models, thereby ensuring the economic sustainability of AI across billions of daily calls. The article emphasizes that "the way the US wins the AI era" is by diffusing AI into the workflows of factories, hospitals, farms, classrooms, and small shops.

Strengthening full-stack competition to prevent innovation from concentrating in the hands of a few giants. By allowing diverse entities to build, adapt, and deploy advanced models, competition will not only occur between model developers but will also extend across the entire industry chain, including cloud chips, applications, and services. The open letter notes that competition is the fundamental driving force for driving innovation, lowering costs, and widely distributing the benefits of AI across the entire economy.

Giving users control over data and models. Open weights enable organizations to maintain control over their data, evaluate and adjust models on demand, and deploy them anytime, anywhere. The value generated by a model's self-improvement capabilities, specialized functions, and accumulated knowledge is retained by the organization itself. The article elevates this to the national level—highlighting it as a critical pathway to strengthening U.S. sovereignty and prosperity.

Closed Source Is Not Inherently Secure

The open letter does not shy away from the real risks of open weights: once released, the weights escape the control of the original developers, and modified versions are difficult to trace and reverse. However, the article explicitly opposes the regulatory approach of 'banning due to risk' and offers three counterarguments:

Asymmetry of attack and defense: Cybersecurity attackers are already widely using advanced AI, and defenders must have models of equal capability to detect, simulate, and respond to threats;

Openness means transparency: Open models expand defensive capabilities and improve system transparency, allowing vulnerabilities to be discovered and patched in a timely manner by multiple teams.

Community oversight is superior to closed-source assumptions: Security inherently comes from scrutiny by more people—openness allows researchers to inspect model behavior, identify vulnerabilities, establish safety guardrails, and continuously improve, making it safer and more reliable than simply assuming closed-source systems are secure.

The letter specifically emphasizes that policymakers should not confuse legitimate model development technologies with 'misappropriation.' The article points out that distillation is a widely used technique in model improvement, evaluation, and verification, continuing the tradition of 'learning from and improving existing technologies' since the open-source software era; illegal value extraction targeting closed-source models should be resolved through targeted legal and commercial frameworks, rather than one-size-fits-all restrictions on these technologies themselves.

The open letter concludes: The AI era can be an era of prosperity—provided the right choices are made, open-weight AI can expand opportunities, strengthen competition, solidify America's technological leadership, mitigate risks, and allow the technological dividends to be widely shared across the entire economy.