DeepSeek: Setting a New Standard for Chinese Companies

A growing number of Chinese companies are set to emulate DeepSeek in advancing innovation, collaboration, and achieving global milestones, despite numerous sanctions from the United States, according to industry experts.

This statement comes in response to reports that the US National Security Council is examining the possible national security concerns associated with DeepSeek, a private startup from China that has recently disrupted the tech landscape with its remarkably cost-effective AI large language model.

DeepSeek has successfully established a domestic supply chain coalition comprising 278 companies, leading to an increase in the domestic production of essential intellectual properties from 19 percent in 2022 to 64 percent today.

According to Pan Helin, a member of the Expert Committee for Information and Communication Economy under the Ministry of Industry and Information Technology, "In the ongoing AI competition between the US and China, DeepSeek has paved a new path for large language models that does not depend on high-end chips and minimizes computing power usage."

He further stated, "This signifies that the computing power constraints imposed by the US on Chinese firms have been circumvented by DeepSeek. More crucially, it highlights that innovation capability, rather than geopolitical hurdles, will determine a nation's success in the global AI race. "

Despite US restrictions on chip exports to China, DeepSeek has produced its models. Analysts from market consultancy Jefferies have estimated that a recent version of DeepSeek's models cost only $5.6 million to train, which is less than 10 percent of the expense for Metas's Llama.

Pan also predicted that DeepSeek would not be the only Chinese enterprise to achieve such advancements in the foreseeable future.

Since the launch of DeepSeek's new AI model, R1, in January, American tech firms have been striving to integrate it, despite limitations instituted by US authorities.

Recently, Microsoft, a key investor in OpenAI, began supporting DeepSeek's R1 on its Azure cloud computing platform.

In addition, top chip manufacturer Nvidia made available DeepSeek's R1 model to users of its NIM microservice, asserting that it delivers exceptional reasoning abilities, high inference efficiency, and superior task accuracy.

Wang Peng, a researcher at the Beijing Academy of Social Sciences, remarked, "While the US perceives AI as an instrument to sustain its supremacy, DeepSeek aims to benefit humanity by utilizing open-source technology and encouraging global applications."

He highlighted that the trend of AI development in China appears increasingly unstoppable, bolstered by both financial backing and an expanding culture of collaboration and innovation.

Last Saturday, Huawei's cloud division partnered with Chinese AI company SiliconFlow to jointly launch DeepSeek's R1/V3 inference service, leveraging Huawei's Ascend cloud services.

"This will allow the model to function reliably in large-scale environments and fulfill the demands for commercial business implementation," the two companies stated.

Liu Zhiyuan, an associate professor of computer science and technology at Tsinghua University, observed that DeepSeek has illustrated that the technological gap between the US and China in AI development has significantly diminished. Nevertheless, he added that DeepSeek still confronts substantial challenges in expanding its model and service capabilities, as AI technologies continue to evolve rapidly.

AI, Innovation, Technology