China’s DeepSeek Rises as a Strong AI Contender, Undermining the Impact of US Export Restrictions
The imposition of US export controls on advanced semiconductor chips has unintentionally paved the way for the rapid rise of DeepSeek, a Chinese start-up based in Hangzhou. This emerging AI company, despite facing strict US regulations that prevent Chinese firms from accessing cutting-edge chips, has successfully developed its R1 AI programme, now considered a formidable rival to US-developed models, all while operating at a much lower cost.
Founder Liang Wenfeng has openly acknowledged the challenges posed by the US embargo, yet analysts argue that these very restrictions have propelled DeepSeek to innovate. By focusing on more efficient AI models that don’t rely on high-end computing power, the company has managed to compete at a global level. Professor Jeffrey Ding from George Washington University has pointed out that DeepSeek’s success illustrates how US export controls are ultimately ineffective in curbing China’s AI progress.
Instead of relying on the most advanced chips, DeepSeek used the less powerful H800 chips—previously allowed for export to China until late 2023—to train its massive AI models. This strategy has allowed the company to keep costs low while remaining competitive. The impact of DeepSeek’s achievements has raised eyebrows in Washington, with some experts warning that China’s advancements in AI might now be unstoppable. This new approach is seen as groundbreaking, defying the earlier belief that superior computational power was necessary for AI breakthroughs.
Venture capitalist Marc Andreessen compared DeepSeek’s success to a “Sputnik moment,” a reference to the 1957 Soviet Union satellite launch that revealed the technological gap between the US and the USSR. For years, the US was believed to have an unchallenged lead in AI, with companies such as OpenAI and Meta at the forefront. DeepSeek’s progress, however, questions this assumption, showing that AI innovation can be achieved in more cost-effective and efficient ways.
Developed for just US$5.6 million, DeepSeek’s R1 chatbot has forced experts to reassess what is possible in AI development. While some dismiss DeepSeek’s cost-saving measures as unremarkable, others, such as Samm Sacks from Yale, argue that the company’s approach has disruptive potential, particularly in terms of cost-efficiency.
In light of DeepSeek’s success, there is growing speculation that Washington may need to adjust its strategy. While expanding export controls to limit China’s access to critical technologies remains a possibility, experts are also calling for a stronger focus on boosting US domestic AI capabilities and forming stronger international partnerships to counter China’s increasing influence in this pivotal sector.
Rebecca Arcesati, an analyst at the Mercator Institute for China Studies, has suggested that DeepSeek’s rise could prompt the US to intensify efforts to secure its leadership in the global AI race, out of fear of falling behind China in this vital technological domain.
