Companies

AI Start-up Challenges Nvidia's Dominance with Beijing Support

Published March 16, 2025

A team associated with Tsinghua University in China has introduced a new artificial intelligence (AI) framework that aims to reduce the reliance on Nvidia chips for AI model inference. This initiative showcases China’s ongoing efforts to boost its technological independence.

The framework, known as Chitu, is designed for high-performance inference of large language models (LLMs). It operates using chips produced in China, positioning itself as a direct competitor to Nvidia’s Hopper series graphics processing units (GPUs). This information was shared by Qingcheng.AI, a start-up, together with Professor Zhai Jidong’s team from Tsinghua University.

AI frameworks are critical components that facilitate the development of sophisticated AI models. They consist of libraries and tools that help developers efficiently design, train, and validate complex models.

The Chitu framework, which has been made open-source recently, supports various mainstream models, including those from DeepSeek and Meta Platforms' Llama series. According to the company, when tested with the complete version of DeepSeek-R1 running on Nvidia’s A800 GPUs, Chitu demonstrated impressive results. It achieved a 315% increase in model inference speed while simultaneously cutting GPU usage by 50% compared to some foreign open-source frameworks.

Impact on Nvidia

This development raises questions about the future of Nvidia's grip on the market. The company has faced restrictions on exporting its advanced GPUs to China due to U.S. government regulations. In particular, Nvidia is prohibited from selling its high-end H100 and H800 chips from the Hopper series to clients in China.

Conclusion

The introduction of Chitu represents a significant step for Chinese AI companies seeking to minimize reliance on foreign technology. As China continues to enhance its capabilities in AI, the competition with established players like Nvidia is likely to intensify, potentially altering the landscape of the AI technology market.

AI, Technology, China