Tuesday, January 28, 2025 - US tech giant Nvidia saw its stock plummet nearly 17% on Monday, leading a sell-off across AI-related stocks after the Chinese AI chatbot DeepSeek gained sudden prominence. Nvidia’s market value dropped by $589 billion in a single day, marking the largest one-day loss in stock market history.
DeepSeek, a Chinese-developed AI model, launched its latest version,
DeepSeek-R1, on January 20. The app has since become the most downloaded free
app in the US. Developed at a fraction of the cost of its American
counterparts, the emergence of DeepSeek has raised concerns about the future of
AI investments and the dominance of US firms.
DeepSeek's researchers claim the model was trained at a cost of just $6
million—significantly less than the estimated $100 million required to train
OpenAI's GPT-4. "If DeepSeek’s innovations are adopted broadly, training
costs could come down significantly," said Raymond James analyst Srini
Pajjuri, though he cautioned that this may accelerate efforts by US firms to
leverage their access to advanced chips to maintain a competitive edge.
Marc Andreessen, a Silicon Valley venture capitalist and adviser to
President Donald Trump, called DeepSeek-R1 "AI's Sputnik moment,"
referencing the shock caused by the Soviet Union's 1957 satellite launch.
The ripple effects of DeepSeek’s rise were evident across the tech
sector. Nvidia's rival Broadcom saw its stock tumble 17.4%, while Micron fell
12% and Advanced Micro Devices dropped more than 6%. Major tech players like
Microsoft and Alphabet, Google's parent company, also experienced declines of
2.14% and over 4%, respectively.
In Europe, Dutch chip equipment maker ASML ended Monday with a 7% drop
in share price, while Siemens Energy, which produces AI-related hardware, saw a
staggering 20% decline.
DeepSeek’s sudden popularity has sparked debates about the efficiency of
AI models and their implications for the industry. Singapore-based tech equity
adviser Vey-Sern Ling suggested that the app’s emergence "could
potentially derail the investment case for the entire AI supply chain."
However, some analysts remain skeptical of DeepSeek’s cost claims. Stacy
Rasgon, an analyst at Bernstein, pointed out that the reported $6 million
training cost likely does not account for associated expenses like prior
research and algorithm development. "This announcement is not really
worthy of the hysteria that has taken over the Twitterverse," Rasgon
added.
Meanwhile, Nvidia issued a statement describing DeepSeek-R1 as "an
excellent AI advancement" but appeared unfazed by the competition.
DeepSeek's launch comes at a time of heightened restrictions on the
export of advanced US chip technology to China, further complicating the AI
landscape. Former President Joe Biden's administration had tightened these
rules, limiting China’s access to Nvidia chips and advanced chipmaking
technology.
Despite these challenges, Chinese developers like DeepSeek have found
ways to innovate, using open-source software and exploring lower-cost
approaches to AI. Liang Wenfeng, DeepSeek's founder, reportedly built up a
stockpile of Nvidia A100 chips before the US imposed export restrictions. Liang
combined these chips with less advanced ones to train his model
cost-effectively.
The timing of DeepSeek's rise coincides with a major US push to bolster
its own AI infrastructure. Last week, President Trump announced the Stargate AI
project, which involves an immediate $100 billion investment in data centers
and infrastructure, with plans to allocate an additional $400 billion over the
next four years. The initiative aims to secure America's dominance in the AI
space, with backing from SoftBank, Oracle, OpenAI, and UAE-based MGX.
As the AI sector grapples with DeepSeek’s disruptive potential, the
company revealed on Monday that it had been targeted by cyberattacks. "Due
to large-scale malicious attacks on DeepSeek's services, we are temporarily
limiting registrations to ensure continued service," the company said in a
statement.
DeepSeek, founded in 2023 by Liang Wenfeng in Hangzhou, China, continues
to challenge the industry with its low-cost AI model. However, questions remain
about the long-term implications of its innovations and whether Chinese AI
firms can sustain their progress amidst US export restrictions.
0 Comments