China’s top AI executives caution about a growing divide with the US following a week of $1 billion IPOs

China’s Leading AI Experts Express Doubt Over Surpassing US in Near Future

At a recent summit in Beijing, several of China’s top minds in generative artificial intelligence voiced skepticism about the country’s prospects of overtaking the United States in the global AI arena any time soon.

Justin Lin, who leads Alibaba’s Qwen open-source AI models, estimated that there is less than a one-in-five chance that a Chinese company will outpace industry giants like OpenAI or Anthropic with groundbreaking innovations within the next three to five years. This cautious outlook was echoed by representatives from Tencent and Zhipu AI, both of which have recently played significant roles in bringing large-language models to the public market.

Highlights from Bloomberg

“OpenAI dedicates a vast amount of computing power to pioneering research, while we are often stretched thin just to meet current demands,” Lin remarked during a panel at the AGI-Next summit. “It raises the age-old question: does true innovation come from those with abundant resources, or from those with less?”

The summit, jointly organized by Zhipu and Tsinghua University, came on the heels of major fundraising successes. Zhipu and MiniMax Group, based in Shanghai, together secured over $1 billion in recent market debuts. MiniMax’s shares more than doubled on their first day of trading, and Zhipu’s stock has risen by 36% since its launch.

Despite these achievements, Chinese AI leaders remained cautious about surpassing the US in developing advanced AI models. The event, held in Zhongguancun—often called Beijing’s Silicon Valley—featured similar sentiments from Tang Jie, Zhipu’s founder and chief AI scientist, and Yao Shunyu, who recently joined Tencent from OpenAI to spearhead its AI initiatives.

“We’ve released some open-source models, and while there’s excitement that China may have caught up with the US, the reality is the gap could actually be growing,” Tang observed.

The impressive performance of DeepSeek’s R1 model at the beginning of 2025 inspired a surge of Chinese companies, from established players like Alibaba to startups such as Zhipu, to make their latest AI models open-source. These efforts have helped narrow the distance with proprietary models from US leaders like OpenAI, Anthropic, and Google.

Major Challenges Facing China’s AI Industry

While acknowledging significant progress, speakers pointed to limited resources and US restrictions on the export of advanced chips and lithography equipment as major obstacles.

Industry Leaders Outline Future Priorities

Yao encouraged his colleagues to address the main challenges facing next-generation AI models, such as improving long-term memory and self-learning capabilities.

Looking ahead, the panelists shared their strategic focuses for the coming year. Yao is working to help Tencent harness AI to deliver more value to its vast user base, for instance by integrating the company’s Yuanbao assistant with WeChat chat histories.

Lin emphasized Alibaba’s commitment to advancing multimodal AI and real-world agents, while Tang and Moonshot AI founder Yang Zhilin highlighted the latest versions of their foundational models.

“Internal competition that leads nowhere is pointless,” Tang stated. “Our goal should be to represent China and advance artificial general intelligence for the benefit of the world.”

Most Popular from Bloomberg Businessweek

©2026 Bloomberg L.P.