Jensen Huang and 25 Tech Heavyweights Rally to Save Open-Source AI
Jensen Huang finally made his debut on X, not to post memes, but to lead a high-stakes rebel alliance of tech titans fighting for the future of downloadable artificial intelligence.
The chief executive of NVIDIA launched his social media presence by dropping an open manifesto signed by twenty-five major tech firms and venture funds. Heavyweights like Microsoft, Meta, IBM, Mistral, and Hugging Face backed the petition. Unsurprisingly, closed-garden heavyweights like OpenAI, Google, Anthropic, and Amazon were nowhere to be found on the list.
The coalition defines open-weight AI as software that anyone can freely download, inspect, tweak, and host on their own servers. Keeping weights public allows startups and universities to skip building billion-dollar models from scratch while dodging subscription fees for locked-down corporate APIs. The document emphasizes that local hosting prevents severe vendor lock-in while giving researchers direct access to run security stress tests.
While acknowledging that leaked weights cannot be recalled once released into the wild, the group contends that proprietary models suffer from catastrophic single points of failure. The signatories specifically urged US lawmakers to fund public datasets and compute grants rather than slamming the brakes on code distribution. They also demanded legal protections for model distillation—the standard practice of using one AI's outputs to train another—asking regulators to penalize actual copyright theft instead of banning the fundamental training technique outright.
The AI landscape has officially split into two warring tribes: those selling black-box subscriptions and those selling the hardware to run the code at home. When chip kings and cloud giants join forces against regulatory lockdowns, corporate self-interest suddenly wears the heroic armor of open science.
Source: NVIDIA
Comments
This is where the magic happens: AI reads your discussion and rewrites the article based on the most interesting comments. Each strong comment adds points to the meter below. Once the meter is full, the article updates live — no page reload needed.