Huawei Chips Set to Power DeepSeek AI Soon

Updated On:
Huawei Chips Set to Power DeepSeek AI Soon

A new player has entered the AI race. DeepSeek, a Chinese AI model, is gaining attention for its cost-effectiveness, claiming to surpass US competitors. It has already made waves in the AI space, especially with its DeepSeek R1 LLM (Large Language Model), a powerful tool for generating responses from trained AI models.

Transition to Huawei’s Chips

Initially, DeepSeek used Nvidia’s H100 GPUs for training. However, it now relies on Huawei’s Ascend 910C for inference, the process of generating AI responses. This shift marks a significant move in China’s push for self-reliant AI infrastructure, reducing its reliance on US-based chipmakers.

While Ascend chips are not used for training, they require less GPU power, making them more efficient. Despite their advantages, the 910C still falls short compared to Nvidia’s H100 in performance. This limits its ability to handle large-scale AI training, a crucial factor for industry leaders.

Huawei’s Upcoming AI Chips

Huawei plans to address this gap with the upcoming Ascend 920C chip, which will compete with Nvidia’s Blackwell B200, a top-tier AI chipset. This new chip aims to provide better performance, helping DeepSeek reach its full potential.

China’s Growing AI Ambitions

DeepSeek’s rapid rise signals China’s increasing ambition in the AI space. As US restrictions limit access to cutting-edge Nvidia hardware, China has been accelerating its AI development, focusing on domestic chip solutions. DeepSeek’s success in the App Store and its open-source approach suggest it is already gaining traction in the market.

While Huawei’s AI chips are not yet on par with global competitors, DeepSeek’s growing presence suggests China is quickly catching up. Whether the new Ascend chips can compete with global leaders remains to be seen. However, the company’s rise in the AI field proves it is a contender to watch.

Leave a Comment