At the World Artificial Intelligence Conference (WAIC) held in Shanghai from July 17 to 20, China unveiled a new milestone in its AI ambitions with the launch of the Kimi K3, an open-weight AI model developed by Beijing-based start-up Moonshot AI. Positioned as a competitor to leading U.S. models from Anthropic and OpenAI, Kimi K3 has captured global attention by offering competitive performance despite limitations on access to advanced semiconductor technology.
The Kimi K3’s debut was met with significant interest from international media and industry delegates, signaling China’s accelerating presence in the global AI landscape. Unlike proprietary, closed-source models predominant among U.S. firms, Kimi K3 belongs to the category of open-weight models, which publicly release the parameters needed to run and modify the AI locally. This contrasts with the closed systems accessed exclusively through their developers’ servers. While not explicitly a geopolitical strategy, China’s emphasis on open-weight and open-source AI aligns with President Xi Jinping’s vision, expressed during his keynote address, of AI as an “international public good” and a tool to promote inclusive development, especially among Global South countries.
Xi’s attendance marked his first at the annual conference since its inception in 2018, underscoring the political significance China assigns to WAIC and AI diplomacy. During the event, China launched the World Artificial Intelligence Cooperation Organization (WAICO), aiming to foster international collaboration on AI governance and equitable technology access. Analysts note that WAICO may serve as a counterweight to the U.S.-led Pax Silica coalition, focused on securing key components of the AI supply chain, including semiconductors and critical minerals.
Chinese state media have characterized U.S. technology restrictions as attempts to establish an “AI Iron Curtain,” framing Beijing’s open approach as a more inclusive alternative. The Chinese initiative further promises substantial support to developing countries, including 5,000 AI training opportunities over five years and access to China’s AI-powered meteorological system, MAZU, intended to aid disaster preparedness and align with China’s broader global development and diplomatic strategies.
The divergence between China’s and the U.S.’s approaches also reflects differing views on digital sovereignty. Open-weight models like Kimi K3 enable countries to host, modify, and control AI infrastructure domestically, reducing dependency on foreign cloud-based services. This appeals to governments seeking greater control over their data and technology ecosystems. Experts observe that reliance on proprietary U.S. models, which are occasionally subject to export restrictions—as demonstrated by the U.S. Department of Commerce’s temporary suspension of foreign access to Anthropic’s AI models—raises concerns about vulnerability to shifting policies in Washington.
The U.S. government’s own AI strategy recognizes the value of open-source and open-weight models for fostering research and competition. However, the rising prominence of Chinese open-weight models like Kimi K3 has led to heightened scrutiny. Reports indicate that U.S. officials have considered restrictive measures against the use of Chinese open-weight AI on grounds of national security, partly fearing that outputs from U.S. models might be co-opted through techniques like “industrial-scale distillation,” though such claims are disputed.
This evolving technological rivalry occurs amid escalating public debate over the pace and risks of AI development. High-profile resignations and calls by AI leaders for a developmental slowdown contrast with political figures, including former U.S. President Donald Trump, who emphasize maintaining American leadership to prevent ceding technological dominance to China.
As China intensifies its efforts to position AI as a global common good and challenge U.S. technological primacy, the contest over AI governance, technological openness, and international alliances is increasingly shaping the geopolitical landscape surrounding the future of artificial intelligence.