China’s AI Models Lag U.S. Counterparts by 6 to 9 Months, but Rapid Adoption Expected, Says Former Google China Head

Chinese artificial intelligence (AI) models are currently trailing their U.S. counterparts by approximately six to nine months, according to Kai-Fu Lee, the former head of Google China. Lee, a prominent commentator on AI and founder of venture capital firm Sinovation Ventures, spoke at the AVCJ Private Equity Forum China on Wednesday.

Lee, who is also the author of “AI Superpowers: China, Silicon Valley, and the New World Order,” highlighted that while the development of large language models (LLMs) in China is progressing, there remains a noticeable lag compared to the United States. He noted that the most advanced Chinese LLMs are roughly six to nine months behind U.S. models, and less developed models could be lagging by up to 15 months.

Despite this gap, Lee forecasted that Chinese AI applications would experience faster growth compared to their American counterparts. He attributed this potential acceleration to the significantly reduced costs of training high-quality AI models. “It’s inevitable that China will develop some of the best AI applications in the world,” Lee asserted, though he remained uncertain whether these innovations would come from large corporations or smaller startups.

Lee also projected that the transition to more sophisticated generative AI consumer applications might take five to eight years, envisioning the emergence of a “super app” capable of performing multiple tasks. He suggested that this future development might necessitate entirely new types of devices, which should be “always on, always listening.”

In the current landscape, major Chinese tech companies such as Alibaba and Tencent have already launched their AI models and business products, with substantial investments in various AI startups. For instance, Beijing-based ShengShu Technology, supported by Ant Group, unveiled a new feature for its text-to-video model, Vidu. This enhancement aims to improve the consistency of main elements or characters in AI-generated clips, facilitating more effective advertising video creation. Vidu, which was released earlier this year, offers basic tools to the public and more advanced features via subscription. Co-founder and CEO Jiayu Tang indicated growing interest from companies seeking ShengShu’s services, reflecting a broader engagement with the technology.