Chinese Premier Li Qiang: China's AI Models 'Spearheaded Global Open-Source AI Ecosystem' — Beijing Doubles Down on Open-Source as Geopolitical Strategy

Chinese Premier Li Qiang publicly declared that China-made large AI models have 'spearheaded the development of the global open-source AI ecosystem' — the most senior official endorsement of Beijing's strategy to use open-source AI distribution as a geopolitical tool against US AI leadership. Confirmed in reporting on May 1, 2026 coinciding with DeepSeek V4's week-old launch, the statement frames China's open-source posture as a sovereign strategic choice rather than a commercial tactic. By releasing competitive models under permissive licenses (DeepSeek V3, R1, V4 under MIT License; Alibaba Qwen family; Baidu ERNIE), China maximizes global adoption, erodes the revenue premium of US AI companies, and embeds Chinese architectural innovations into the global AI development stack. The strategic calculus is asymmetric: open-source models with Chinese-designed training efficiency become free infrastructure for global developers, while US frontier AI labs — which keep weights proprietary — compete against a rival that gives away comparable capabilities at zero cost. DeepSeek's models have been downloaded millions of times by Western researchers and companies, building global familiarity with Chinese AI architectures. However, Japan Times analysis flags a potential tension: Alibaba's reported pivot toward proprietary enterprise AI products may signal that China's leading commercial AI players see open-source as a growth-phase strategy with diminishing returns as enterprise monetization opportunities mature. If commercially sophisticated Chinese AI companies eventually close their model weights — as US labs did before them — Beijing's open-source narrative advantage could erode. The Premier's public endorsement locks in the open-source strategy at the highest political level, making any commercial retreat from it politically complicated in the near term.