In Shanghai on February 5, Chinese companies, ranging from chip manufacturers to cloud service providers, are racing to support artificial intelligence models. Analysts have dubbed this a "watershed moment" for the industry.
Moore Threads and Hygon Information Technology, creators of AI chips aiming to rival Nvidia, announced on Monday that their computing clusters and accelerators would now support DeepSeek's R1 and V3 models.
In a WeChat post, Moore Threads praised DeepSeek for its models' utilization of domestically made graphic processing units (GPU), stating that it could potentially revolutionize China's AI industry.
Huawei Technologies revealed on Saturday that it is collaborating with AI infrastructure start-up SiliconFlow to offer DeepSeek's models to customers via its Ascend cloud platform, with performance comparable to high-end global chips.
According to Bernstein analysts, Huawei's integration of DeepSeek's models into its Ascend chips signifies a significant moment, asserting that competitive large language models (LLM) can be effectively deployed on China's competitive chips, reducing dependence on advanced U.S. hardware.
Alibaba, Baidu, and Tencent's cloud divisions have also made DeepSeek's models available through their services.
DeepSeek recently introduced a cost-effective AI assistant that uses minimal data, swiftly surpassing its U.S. competitor ChatGPT in Apple App Store downloads, causing a tech stock downturn globally.
Previously, DeepSeek gained recognition in global AI communities with a research paper stating that training the DeepSeek-V3 model required significantly less computing power cost compared to tech giants like Meta and Microsoft.
China has celebrated the success of the Hangzhou-based startup and its founder, who have become popular culture icons.
Microsoft and Amazon's cloud services have joined in offering DeepSeek's models, yet several countries have expressed privacy concerns, resulting in some blocking or investigating DeepSeek's AI app.