Hugging Face CEO Says China Is Winning AI Race As US Labs Build In Silos

Hugging Face CEO Clément Delangue says China is winning the artificial intelligence race because its developers collaborate through open models, while leading US laboratories develop advanced systems in isolation and share less with the wider AI ecosystem.

 

TL;DR

• Clément Delangue said China is progressing faster because of its open-science and open-model ecosystem.
• He accused leading US AI laboratories of building advanced systems in silos.
• Delangue believes China could dominate frontier AI, not only open models, by the end of 2026 or during 2027.
• His comments represent an industry assessment, not a confirmed measure of global AI leadership.

Speaking on CNBC’s Squawk on the Street, Delangue was asked whether he believed China was winning the AI race.

“I think they are,” Delangue replied.

He attributed China’s progress to greater collaboration around open science and open models. By contrast, Delangue said frontier AI laboratories in the United States were building their systems in silos without sharing enough technology and research with the broader ecosystem.

According to Business Insider’s report on the interview, Delangue said collaboration between Chinese developers was accelerating the country’s rate of progress.

He also predicted that China could move beyond leading the open-model sector and begin dominating frontier AI more broadly by the end of 2026 or during 2027. That forecast remains Delangue’s assessment and does not establish that China has definitively overtaken the United States across every measure of AI development.

 

China’s Open AI Ecosystem Is Expanding

 

Delangue’s argument is supported by activity documented on Hugging Face’s own platform.

Hugging Face’s Spring 2026 open-source report said China’s AI ecosystem moved heavily toward open-source development during 2025, following the release of DeepSeek’s R1 model.

The company said the number of repositories released by prominent Chinese organizations rose sharply. Baidu went from publishing no models on Hugging Face in 2024 to releasing more than 100 in 2025, while ByteDance and Tencent increased their releases by approximately eight to nine times.

A separate Hugging Face analysis of China’s open-source AI ecosystem said more than 113,000 models on the platform used Alibaba’s Qwen as a base by mid-2025.

More than 200,000 model repositories also carried a Qwen tag, compared with around 27,000 for Meta’s Llama and 6,000 for DeepSeek. These figures measure activity on Hugging Face and do not independently determine which country leads the entire AI industry.

Delangue Previously Predicted China’s AI Rise

 

The latest remarks are consistent with a prediction Delangue published before the current interview.

In December 2024, the Hugging Face CEO predicted that China would begin leading the AI race as a consequence of its position in open-source AI.

Delangue has repeatedly argued that open development allows researchers and companies to inspect, modify and deploy models without depending entirely on a closed provider.

He used Hugging Face’s recent cybersecurity incident as another example during the CNBC interview. According to the company’s July 2026 security disclosure, commercial AI services blocked parts of its forensic analysis because their safety systems interpreted malicious commands in the attack logs as prohibited requests.

Hugging Face instead ran Beijing-based Z.ai’s open-weight GLM 5.2 model on its own infrastructure. The company said this allowed its security team to analyze the material without sending sensitive attacker data or referenced credentials outside its environment.

Delangue’s warning ultimately centers on the strategic value of collaboration. His argument is that US companies may possess powerful proprietary models, but China’s expanding open ecosystem could enable faster collective progress and eventually challenge American leadership at the frontier.