Why China's DeepSeek is redefining AI landscape, disrupting US stocks
Japanese technology shares fell on Tuesday as a global market rout sparked by the emergence of a low-cost Chinese artificial intelligence model entered day two.
Japanese technology shares fell on Tuesday as a global market rout sparked by the emergence of a low-cost Chinese artificial intelligence model entered day two.
Japanese technology shares fell on Tuesday as a global market rout sparked by the emergence of a low-cost Chinese artificial intelligence model entered day two.
Beijing: Chinese startup DeepSeek has captured global attention with its groundbreaking AI models, claiming they rival or surpass leading US-based AI systems at a fraction of the cost. This disruptive innovation is challenging the status quo in the global technology landscape.
DeepSeek gained prominence after publishing a paper last month stating that its latest AI model, DeepSeek-V3, required less than $6 million in computing power, leveraging Nvidia’s H800 chips for training.
The company’s AI assistant, powered by DeepSeek-V3, has already overtaken ChatGPT to become the top-rated free app on Apple’s US App Store, raising questions about the enormous investments being made by US tech giants in AI. The development has impacted the shares of several major tech firms, including Nvidia.
Japanese technology shares fell on Tuesday as a global market rout sparked by the emergence of a low-cost Chinese artificial intelligence model entered day two, with investors questioning the sky-high valuation and dominance of AI bellwethers.
Shares of Nvidia, the poster child of the AI boom in recent years, dragged US stocks lower, sinking 17% on Monday and wiping $593 billion from the chipmaker's market value, a record one-day loss for any company.
Here’s a closer look at DeepSeek and its rapid rise:
Why is DeepSeek making waves?
The launch of OpenAI’s ChatGPT in late 2022 set off a competitive race in China to develop AI chatbots. Initial efforts, such as Baidu’s ChatGPT equivalent, were criticised for lagging behind their US counterparts, reinforcing perceptions of a significant gap in AI capabilities.
DeepSeek has flipped this narrative. Its flagship models, DeepSeek-V3 and DeepSeek-R1, have been hailed by Silicon Valley executives and engineers as being on par with advanced systems from OpenAI and Meta. The key difference? DeepSeek’s solutions are far more cost-effective.
For instance, the DeepSeek-R1 model, launched last week, is reported to be 20 to 50 times cheaper to operate than OpenAI’s models, depending on the task, according to the company’s WeChat account.
However, scepticism about DeepSeek’s claims remains. Alexandr Wang, CEO of Scale AI, alleged during a CNBC interview that DeepSeek possesses 50,000 Nvidia H100 chips, potentially breaching US export controls that restrict the sale of advanced AI chips to Chinese companies. DeepSeek has not responded to these claims.
Moreover, analysts at Bernstein raised concerns about the reported $5.58 million training cost for DeepSeek-V3, suggesting the actual figure might be significantly higher. The training costs for the R1 model remain undisclosed.
Who is behind DeepSeek?
DeepSeek is based in Hangzhou and is controlled by Liang Wenfeng, co-founder of the quantitative hedge fund High-Flyer. In March 2023, High-Flyer announced its shift towards artificial general intelligence (AGI) research, leading to DeepSeek’s creation.
AGI refers to autonomous systems capable of outperforming humans in most economically valuable tasks—a goal that DeepSeek appears to be pursuing.
While High-Flyer’s exact investment in DeepSeek remains unclear, corporate records show that High-Flyer operates a cluster of 10,000 A100 chips and holds patents related to AI training infrastructure. The hedge fund shares an office building with DeepSeek, suggesting a close operational link.
How is Beijing responding?
DeepSeek’s success has garnered attention from China’s top political leaders. On January 20, the day DeepSeek-R1 was launched, founder Liang Wenfeng attended a closed-door symposium with Chinese premier Li Qiang, according to state news agency Xinhua.
Liang’s inclusion in such a high-level meeting signals that Beijing views DeepSeek as a critical player in its strategy to counter US export controls and achieve self-sufficiency in strategic technologies like AI.
Last year, a similar symposium was attended by Baidu CEO Robin Li, indicating that DeepSeek is now part of China’s elite tech circle, with significant implications for the global AI landscape.
The keywords 'deepseek ai stocks', 'Chinese ai', 'deepseek r1 ai', and 'nvidia market cap' have been among the most searched on Google Trends since Tuesday morning.
(With Reuters inputs.)