Le Monde: China's New AI Models Unsettle Silicon Valley

An article published by Le Monde on Monday emphasized that China's advancements in artificial intelligence are unsettling Silicon Valley. The piece notes that China has recently launched multiple highly powerful AI models with open-source approaches, among which Kimi3, the latest large model unveiled by Moonshot AI, has become a focal point of market attention.

On July 16, Moonshot AI released its newest large model, Kimi3. Following the announcement, the Nasdaq index in New York was impacted, triggering significant market turbulence. Industry insiders were astonished to discover that, in certain performance metrics, this model approaches or even surpasses some leading American models, ranking among the top globally in comprehensive intelligence evaluations.

Additionally, about a month prior to Kimi3’s release, another Beijing-based AI startup—Zhipu AI—launched its new model in mid-June.

Le Monde states that for years, the United States has believed that due to Washington’s ban on exporting NVIDIA’s most advanced AI chips to China, Chinese companies would lag behind their American counterparts by several months—or even years. However, this perception is now beginning to be challenged.

Le Monde points out that while neither Moonshot AI nor Zhipu AI disclosed the specific chips used to train their models, an important feature of China’s rapid AI development lies in how Chinese engineers, under chip restrictions, place greater emphasis on algorithm optimization and efficient resource utilization. They extensively adopt advanced architectures such as "Mixture of Experts (MoE)," activating only relevant parts of the model when handling different tasks, significantly reducing computational costs and continuously improving efficiency in using computing resources. Le Monde notes that most industry experts believe Chinese engineers have successfully turned “chip shortages” into advantages.

Open Source: China’s New Flagship in AI

Le Monde adds that more importantly, Chinese AI models—including Kimi3—are adopting open-source models, allowing developers to freely download, deploy, and modify them. In contrast, major U.S. AI companies still predominantly rely on paid, closed-source models. China’s open strategy not only lowers barriers to AI application but also facilitates rapid dissemination of AI technology and supports the building of a vibrant AI ecosystem.

Open source has become one of the most distinctive characteristics of China’s AI development. “Open source” is not merely a technical choice—it is gradually emerging as a core philosophy guiding China’s AI growth, seen as a key competitive advantage distinguishing it from the U.S.’s closed business model.

Highly Competitive, Diverse Landscape in China’s AI Sector

Le Monde believes that China’s rapid rise in the AI industry is also driven by a highly competitive AI environment within the country.

The Chinese government regards artificial intelligence as one of the key domains in future international competition and has prioritized catching up with—and even surpassing—global AI leaders as a national strategic goal.

At the same time, major Chinese internet giants are engaged in intense competition. In cities like Shenzhen, Hangzhou, and Beijing, new AI startups continue to emerge, making the entire industry dynamic and full of vitality.

Beyond Moonshot AI, Zhipu AI, and DeepSeek, China already hosts numerous mature AI enterprises—such as MiniMax, Douyin’s Dobo, Alibaba’s Tongyi Qianwen, Tencent’s Hunyuan—each developing large-scale models that collectively drive continuous improvement across China’s overall AI capabilities.

China Possesses Abundant Talented AI Personnel

Le Monde further highlights that China boasts a vast pool of outstanding AI talent. Many founders of leading AI companies graduated from top-tier universities like Tsinghua University and previously studied or worked at renowned U.S. universities or international tech firms. For example, Yang Zhilin, founder of Moonshot AI, earned his Ph.D. at Carnegie Mellon University in the United States and worked at Meta (formerly Facebook) and Google Brain before returning to China to establish his company, becoming a representative figure of China’s new generation in AI.

Original article: toutiao.com/article/1871310497737735/

Disclaimer: The views expressed in this article are those of the author(s) and do not necessarily reflect the opinions of the publisher.