Former Tencent AI chief says LLMs cannot beat U.S., points to AI video as next battleground
Former Tencent AI chief says LLMs cannot beat U.S., points to AI video as next battleground
Publish Date: 2026-05-22 02:14:00
Source Domain: www.digitaltoday.co.kr
- A growing sense of crisis is present in China’s AI industry regarding the competitiveness of large language models (LLMs) compared to the United States.
- Liu Wei from Tencent emphasized that although benchmark scores suggest a narrowing gap with the U.S., China still struggles with actual technological innovation and “paradigm innovation.”
- The U.S.-China AI technology gap is estimated to be around 3 to 6 months, potentially widening if the U.S. continues to lead in new technological directions.
- Liu highlighted the need for Chinese companies to shift focus from LLMs to AI video generation, where China appears to have a more secure competitive position.
- Companies like ByteDance and Kuaishou are gaining global recognition for their AI video models, with Video Reverse, led by Liu, also performing well.
- Resource constraints due to U.S. semiconductor export controls make video-generation models more accessible for Chinese firms compared to LLMs.
- Video Reverse plans to expand into developing “world models” and is working on next-generation AI video and simulation models.
- The trend indicates that China’s AI industry is seeking new competitive fields to move beyond merely following U.S. innovation.
