Companies

AI Start-Up Takes on Nvidia's Dominance with Beijing's Backing

Published March 16, 2025

A new artificial intelligence (AI) framework, developed by teams connected to China’s Tsinghua University, aims to reduce dependence on Nvidia chips for AI model inference. This initiative represents a significant step by China in its quest for greater technological self-sufficiency.

The framework, named Chitu, is designed for high-performance inference of large language models (LLMs) and is compatible with domestically manufactured chips. It poses a challenge to Nvidia’s Hopper series graphics processing units (GPUs), particularly for models like DeepSeek-R1. This announcement was made in a joint statement by a start-up called Qingcheng.AI and a research group led by computer science professor Zhai Jidong at Tsinghua University.

AI frameworks are crucial tools that help in the development of sophisticated AI models. They provide developers with a set of libraries and tools needed to design, train, and validate complex models effectively.

The Chitu framework, which has been open-sourced as of last Friday, supports several mainstream AI models, including those from DeepSeek and Meta Platforms' Llama series. According to the company, tests conducted using the full-strength version of DeepSeek-R1 on Nvidia’s A800 GPUs showed remarkable results, achieving a 315% increase in model inference speed while also cutting GPU usage by 50% when compared to existing foreign open-source frameworks.

What This Means for Nvidia

This move is part of a larger strategy by Chinese AI companies to reduce their dependence on Nvidia. The US government has imposed export controls that prevent Nvidia from selling its advanced H100 and H800 chips to clients in China. This has created an opportunity for domestic alternatives like Chitu to gain traction.

Conclusion

The emergence of affordable AI solutions such as DeepSeek could signal the beginning of a shift away from Nvidia’s current dominance in the AI hardware market. As China continues to invest in its homegrown technologies, developments like the Chitu framework could play significant roles in the future landscape of AI.

AI, Nvidia, Chitu