A cyberattack that occurred at the same time that DeepSeek, a Chinese AI startup, saw a sharp increase in the popularity of its AI assistant, prompted the company to temporarily limit new user registrations on Monday.
Following its AI assistant becoming the top-rated free software on Apple’s U.S. software Store earlier in the day, DeepSeek also encountered website problems. The company’s status page states that the problems with its online login and API have been fixed. Monday’s outage, which coincided with a spike in the company’s popularity, was the longest service interruption it has experienced in the previous three months.
Silicon Valley has taken notice of DeepSeek’s ascent, which calls into question accepted notions about American dominance in AI and the efficacy of Washington’s export controls on China’s semiconductor and AI industries.
Monday saw a slump in the technology sector, with big drops in the shares of large businesses like Oracle (ORCL.N) and Nvidia (NVDA.O).
From ChatGPT to DeepSeek, sophisticated chips are necessary for training AI models. In an attempt to slow the advancement of Chinese AI technologies, the Biden administration has increased restrictions to prohibit the transfer of such chips to China till 2021.
In a recent article, however, DeepSeek researchers asserted that their DeepSeek-V3 model was trained with Nvidia’s H800 CPUs, with training expenses totaling less than $6 million. The usage of less powerful chips—in comparison to those the U.S. has sought to keep out of China—as well as the comparatively low training costs have caused American tech leaders to question the overall efficacy of these export regulations, though this claim has been disputed.
The parent company of DeepSeek, a startup based in Hangzhou that was founded in 2023—the same year Baidu unveiled China’s first large-language AI model—is not well known to the general public. Since then, many Chinese companies have created their own AI models, ranging from large enterprises to smaller startups. Nonetheless, DeepSeek has made a name for itself as the first Chinese AI startup to be widely praised by American industry insiders for achieving performance comparable to or even better than top American AI models.