Over the past two years, the competition in the realm of large AI models has predominantly been a duel between China and the United States. However, numerous countries outside this competitive sphere are vigorously pursuing the development of sovereign AI. By the first half of 2026, a total of 67 countries across the globe had made significant progress, advancing 184 sovereign AI projects supported by their governments. Capitalizing on China's MiniMax M3 model as its foundational framework, Saudi Arabia swiftly launched HUMAIN-M3 within a mere three months, tailored specifically for the Arabic-speaking community. This model demonstrated superior performance compared to GPT-5.6 Sol and Opus 5 across multiple evaluation tests. To cater to the specific needs of local language adaptation, ensure robust data security, and maintain uninterrupted service, many countries show a preference for open-source models. This trend has paved the way for Chinese open-source models to secure clients at the national level. Presently, the global outreach of Chinese open-source models has transitioned from merely offering individual models to providing a holistic system. This system encompasses chip adaptation, seamless collaboration with cloud platforms, and the development of localized models. Such advancements have emerged as the technological cornerstone for sovereign AI in numerous countries, heralding a wider array of opportunities for sustained, long-term growth.