2 min read
DeepSeek, an AI company from China, has made waves in the global tech arena with its cost-efficient chatbot, which now sits at the top of app stores in 51 countries. However, the latest news has revealed a significant shift in the company’s technology strategy: DeepSeek’s R1 large-language model, initially developed using Nvidia’s H100 processors, now operates on Huawei’s Ascend 910C chips.
The inference-focused Ascend 910C differs significantly from Nvidia’s existing high-performance GPU ecosystem. The Ascend 910C provides lower energy costs and superior operational efficiency compared to the H100 although it performs inferiorly in basic learning tasks.
The analysts predict these developments expose Chinese firms’ adaptation strategies due to U.S. export controls which restricted their access to advanced Nvidia chips. Chinese technology firms like DeepSeek along with others adapt to trade limitations by creating their own tech platforms and deploying alternative solutions for success.
The Ascend 910C chip from Huawei delivers notable AI deployment capabilities although its performance does not match the H100 models during intensive training operations. Huawei has launched development of the next-gen Ascend 920C processor which aims to duel against Nvidia’s forthcoming Blackwell B200 processor as China seeks parity with worldwide AI forefront companies.
The rise of DeepSeek, powered by Huawei’s semiconductor technology, occurs at a time when American enterprises face increasing competition from Chinese technology rivals. DeepSeek attained significant success in the realm of open-source chatbots by utilizing Huawei’s semiconductor technology to develop a cost-effective system that competes favorably with American platforms, thereby causing unease among Wall Street investors. The achievements of Chinese enterprises amidst geopolitical pressures have yielded significant outcomes that contest the supremacy of American corporations.
