When 25 of the biggest names in tech agree on something and put it in writing, Washington tends to notice. On July 24, Nvidia, Meta, Microsoft, and 22 other companies signed an open letter advocating for open-source AI models, effectively drawing a line in the sand against a growing chorus of policymakers who want to lock down American AI development in the name of national security.
The letter lands at a moment when the Trump administration is wrestling with a thorny question: how do you compete with China’s rapidly advancing AI capabilities without accidentally handing Beijing the keys to your own technology?
The open-source argument, translated
Here’s the thing about open-source AI models. They let anyone inspect, modify, and build on the underlying code. Think of it like the difference between a recipe you can tweak at home and a meal you can only order from one restaurant. The tech industry’s heavyweights are arguing that the recipe-sharing approach is actually safer, not more dangerous.
Nvidia CEO Jensen Huang put it plainly in his support of the letter.
“Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty.”
That word “sovereignty” is doing a lot of heavy lifting. It’s a deliberate pitch to policymakers who are worried about depending on a handful of closed-model companies for critical AI infrastructure. If the models are open, the argument goes, more companies and governments can build on top of them without being locked into a single vendor’s ecosystem.
The China problem
The backdrop here is impossible to ignore. Chinese AI labs, particularly Moonshot AI and DeepSeek, have been releasing increasingly sophisticated open-source models that have caught Washington’s attention. Some policymakers see these Chinese models as a dual threat: they could siphon American intellectual property while simultaneously offering alternatives to US-built systems.
The letter directly addresses these concerns. Rather than retreating behind closed walls, the signatories argue that the US should lean into openness as a competitive advantage. Their logic is straightforward. If Chinese labs are already releasing open models, restricting American ones doesn’t eliminate the technology from the global stage. It just ensures that the Chinese versions become the default.
The companies also point to practical constraints that limit Chinese AI development. Nvidia’s chips remain essential hardware for training advanced AI models, and US export restrictions have already curtailed Beijing’s access to the most powerful processors. The signatories highlight this as evidence that the US retains significant leverage even in an open-source environment.
The closed-model counterargument
Not everyone in Silicon Valley is on board with the open-source gospel. Companies like OpenAI and Anthropic have built their businesses around closed, proprietary models. Their argument is that keeping the most powerful AI systems behind access controls prevents misuse by bad actors, whether that’s generating bioweapons instructions or creating sophisticated disinformation campaigns.
The open letter represents a strategic pivot by its signatories, positioning openness not as a liability but as a feature. For Nvidia, the calculus is particularly interesting. As the dominant provider of AI training hardware, the company benefits regardless of whether models are open or closed. But open models tend to democratize demand for computing power, which means more customers buying Nvidia chips.
Meta has been the most vocal corporate champion of open-source AI, having released its Llama family of models to broad adoption. Microsoft straddles both worlds, investing heavily in OpenAI’s closed models while simultaneously supporting open-source initiatives through its Azure cloud platform.