U.S. export controls on advanced chips may have unintentionally contributed to the success of the Chinese start-up DeepSeek and its AI chatbot, raising concerns in Washington about the possibility of China gaining a dominant position in global AI development.
Based in Hangzhou, DeepSeek has surprised investors and industry experts with its R1 program, which rivals American competitors at a significantly lower cost.
This achievement comes despite stringent U.S. restrictions that prevent Chinese companies from accessing the advanced chips necessary for building the large AI models used in cutting-edge AI development.
DeepSeek’s founder, Liang Wenfeng, acknowledged that the “embargo on high-end chips” posed a major challenge. However, analysts suggest that these restrictions may have pushed the company to develop innovative solutions to overcome the obstacles. The firm used the less advanced H800 chips, which were permitted for export to China until late 2023, to power its large AI model.
Jeffrey Ding from George Washington University noted that these export controls have not stopped other countries from developing competitive AI models. In fact, he argued that “history shows it is impossible to suppress a general-purpose technology like AI.”
DeepSeek’s success echoes the resilience seen in other Chinese firms, such as Huawei, which has rebounded from U.S. sanctions by adjusting its business strategy. However, DeepSeek’s impact has raised alarms in Silicon Valley and Washington.
Venture capitalist Marc Andreessen compared the situation to the “Sputnik moment” — the 1957 Soviet satellite launch that revealed a technological gap between the U.S. and its main geopolitical rival.
For years, many assumed that the U.S. would maintain supremacy in AI, with companies like OpenAI and Meta leading the way. Despite China’s investments and its ambition to lead AI by 2030, its efforts had not sparked major concerns. Early AI products, like Baidu’s Ernie Bot, failed to impress, reinforcing the belief that China’s restrictive regulatory environment would limit innovation.
DeepSeek has challenged that assumption. Samm Sacks, a Senior Fellow at Yale, noted that the company has shown that cutting-edge AI could be developed at a fraction of the cost, overturning the traditional thinking about the computing power required for innovation.
Unlike U.S. policy, which has focused on scaling up computing power, DeepSeek’s success emphasizes efficiency and cost-cutting. The company claims that its R1 chatbot was developed for just $5.6 million, suggesting that AI development can be far more affordable than previously thought.
Some experts, however, caution that DeepSeek’s cost-saving measures are not groundbreaking, as the cost of training AI models has generally decreased over time. They also argue that the full picture of the company’s infrastructure and research costs is still unclear.
Still, the success of DeepSeek has been described as a “wake-up call” for Silicon Valley. Former U.S. Representative Mark Kennedy said it doesn’t mean that export controls will lose their effectiveness but may lead to increased scrutiny on Chinese access to advanced AI technology.
Washington could respond by tightening restrictions on AI chips or increasing its own investment in AI to ensure its leadership. Kennedy suggested that alongside defensive measures, the U.S. may need to strengthen alliances and refine its policies to prevent other nations from gravitating toward China’s AI ecosystem.
Rebecca Arcesati from the Mercator Institute for China Studies warned that fears of falling behind China in AI could now accelerate efforts to catch up.
