Eric Schmidt said he worries that most governments will use Chinese AI models (www.businessinsider.com)

Former Google CEO Eric Schmidt warned on the Moonshots podcast that, because most large Chinese models are open-source (and therefore free to use), many countries—especially those with limited budgets—may standardize on Chinese AI rather than U.S. offerings, which are often closed-source and commercial. He framed this as a geopolitical concern: open-source code can be freely adapted and redistributed, accelerating adoption, while closed models require payment or licensing. Recent popularity of Chinese models such as DeepSeek and Alibaba’s Qwen3 has sharpened worries about data privacy, national security, and erosion of America’s competitive edge. The exchange highlights a growing push for “sovereign AI”—national control over models, data, and infrastructure—to prevent dependence on foreign providers. Tech leaders including Nvidia’s Jensen Huang and Mistral’s Arthur Mensch echo Schmidt’s call for countries to build independent systems (for example, by codifying local language and cultural data into domestic LLMs). The debate exposes trade-offs: open source fosters rapid innovation and democratic access, while closed-source proponents argue for stronger security and control. For the AI/ML community, the outcome will shape model governance, global funding flows, standards for data stewardship, and who sets the technical and ethical norms for future AI deployment.