Start-Up’s Success in AI Development Raises Doubts Over the Effectiveness of US Export Controls on China

US export controls on advanced semiconductor chips have inadvertently contributed to the rise of DeepSeek, a Chinese start-up based in Hangzhou, whose artificial intelligence (AI) chatbot is now seen as a significant competitor to US counterparts. Despite strict US regulations preventing Chinese companies from accessing high-tech chips, DeepSeek has developed its R1 programme, which rivals American AI models at a fraction of the cost.

The company’s founder, Liang Wenfeng, acknowledged that the US embargo on advanced chips had presented significant challenges. However, analysts argue that these restrictions have spurred DeepSeek to innovate, creating more efficient models that bypass the need for powerful computing resources. According to George Washington University’s Professor Jeffrey Ding, DeepSeek’s success demonstrates the ineffectiveness of US export controls in halting China’s progress in AI development.

DeepSeek utilised less advanced H800 chips—previously allowed for export to China until late 2023—to train its large AI model, reducing costs while maintaining competitiveness. The firm’s achievements have drawn attention in Washington, with some experts warning that China’s AI push could now be unstoppable. The company’s approach is seen as revolutionary, overturning previous assumptions about the computational power required for AI innovation.

The success of DeepSeek has been likened to a “Sputnik moment” by venture capitalist Marc Andreessen, a reference to the 1957 Soviet satellite launch that exposed the technological gap between the US and the Soviet Union. For years, US dominance in AI was assumed, with major companies like OpenAI and Meta leading the field. But the rise of DeepSeek challenges this assumption, showing that cutting-edge AI can be developed more efficiently and at a fraction of the cost.

DeepSeek’s R1 chatbot, developed for just US$5.6 million, has forced experts to reconsider traditional views on AI development. While some critics argue that the cost-saving measures employed by DeepSeek are not groundbreaking, others, like Yale’s Samm Sacks, emphasise the disruptive potential of the firm’s approach to cost-efficiency.

In response, former US representatives and experts are suggesting that Washington may need to reassess its strategy. While export controls could be expanded to limit China’s access to critical technologies, there are also calls for greater investment in US domestic AI capabilities, as well as stronger international alliances to counter China’s growing influence.

Rebecca Arcesati, an analyst at the Mercator Institute for China Studies, stated that DeepSeek’s success could accelerate the US’s efforts to maintain leadership in the global AI race, driven by the fear of falling behind China in this critical technological domain.