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

Former Google CEO Eric Schmidt warned that, because open-source AI models are free, most governments — especially those with limited budgets — may default to using Chinese open-source models rather than pricier Western closed-source alternatives. On the Moonshots podcast he contrasted “the biggest models in the United States [being] closed source and the biggest models in China [being] open-source,” naming the recent popularity of Chinese offerings like DeepSeek and Alibaba’s Qwen3 as examples fueling adoption. The upshot: cost-driven standardization on foreign models could erode competitive advantage and create geopolitical dependencies. The warning amplifies a growing push for “sovereign AI,” where nations build and govern their own models and data infrastructure. Tech leaders such as Nvidia’s Jensen Huang and Mistral’s Arthur Mensch have urged countries to codify local language and cultural data into national LLMs to prevent money, talent, and sensitive information from flowing abroad. The debate hinges on trade-offs: open-source accelerates innovation and auditability but may raise privacy and security concerns if foreign-controlled, while closed-source models can be monetized and restricted but are less transparent. Schmidt’s comments crystallize the strategic choice facing policymakers and technologists: adopt free foreign models and accept dependency risks, or invest in domestic AI capacity to retain control and safeguard data.