‘Sputnik
moment’: $1tn wiped off US stocks after Chinese firm unveils AI chatbot
Trump calls
emergence of DeepSeek a ‘wake-up call’ amid doubts about sustainability of
western artificial intelligence boom
Dan Milmo,
Amy Hawkins, Robert Booth and Julia Kollewe
Tue 28 Jan
2025 05.56 CET
https://www.theguardian.com/business/2025/jan/27/tech-shares-asia-europe-fall-china-ai-deepseek
The race for
domination in artificial intelligence was blown wide open on Monday after the
launch of a Chinese chatbot wiped $1tn from the leading US tech index, with one
investor calling it a “Sputnik moment” for the world’s AI superpowers.
Investors
punished global tech stocks on Monday after the emergence of DeepSeek, a
competitor to OpenAI and its ChatGPT tool, shook faith in the US artificial
intelligence boom by appearing to deliver the same performance with fewer
resources.
The
tech-heavy Nasdaq Composite closed down 3.1%, with the drop at one point wiping
more than $1tn off the index from its closing value of $32.5tn last week, as
investors digested the implications of the latest AI model developed by
DeepSeek.
Nvidia, a
leading maker of the computer chips that power AI models, was overtaken by
Apple as the most valuable listed company in the US after its shares fell 17%,
wiping nearly $600bn off its market value. Google’s parent company lost $100bn
and Microsoft $7bn.
Nvidia’s
fall was the biggest in US stock market history.
The DeepSeek
AI assistant also topped the Apple app store in the US and UK over the weekend,
above OpenAI’s ChatGPT.
US President
Donald Trump said DeepSeek should be a “wake-up call for our industries that we
need to be laser-focused on competing to win”.
He said he
had been “reading about China” and its companies, in particular one that had
come up with a “faster method of AI and [a] much less expensive method”.
“That’s good
because you don’t have to spend as much money. I view that as a positive, as an
asset,” Trump said.
OpenAI chief
executive Sam Altman praised DeepSeek’s launch, saying that it was
“invigorating to have a new competitor”.
In a social
media post, Altman called it “an impressive model, particularly around what
they’re able to deliver for the price”.
Japanese
tech firms linked to the AI sector tanked for a second straight day on Tuesday
as investors tracked the rout on Wall Street. Advantest plunged more than 9%,
while tech investor SoftBank, a key investor in Trump’s Stargate AI project,
tumbled more than 5%, having lost 8% the day before.
Most other
Asian markets rose in limited trade ahead of the lunar new year break.
DeepSeek was
hit with a cyber-attack on Monday, forcing it to temporarily limit
registrations. On its status page, DeepSeek said it started to investigate the
issue late Monday night Beijing time. After about two hours of monitoring, the
company said it was the victim of a “large-scale malicious attack”. While
DeekSeek limited registrations, existing users were still able to log on as
usual.
DeepSeek
claims to have used fewer chips than its rivals to develop its models, making
them cheaper to produce and raising questions over a multibillion-dollar AI
spending spree by US companies that has boosted markets in recent years.
The company
developed bespoke algorithms to build its models using reduced-capability H800
chips produced by Nvidia, according to a research paper published in December.
Nvidia’s
most advanced chips, H100s, have been banned from export to China since
September 2022 by US sanctions. Nvidia then developed the less powerful H800
chips for the Chinese market, although they were also banned from export to
China last October.
DeepSeek’s
success at building an advanced AI model without access to the most
cutting-edge US technology has raised concerns about the efficacy of
Washington’s attempts to stymie China’s hi-tech sector.
Marc
Andreessen, a leading US venture capitalist, compared the launch of DeepSeek’s
R1 model last Monday to a pivotal moment in the US-USSR space race, posting on
X that it was AI’s “Sputnik moment” – referring to when the Soviet Union
astounded its cold war rival by launching a satellite into orbit.
According to
DeepSeek, its R1 model outperforms OpenAI’s o1-mini model across “various
benchmarks”, while research by Artificial Analysis puts it above models
developed by Google, Meta and Anthropic in terms of overall quality.
The company
was founded by the entrepreneur Liang Wenfeng, who runs a hedge fund,
High-Flyer Capital, that uses AI to identify patterns in stock prices. Liang
reportedly started buying Nvidia chips in 2021 to develop AI models as a hobby,
bankrolled by his hedge fund. In 2023, he founded DeepSeek, which is based in
the eastern Chinese city of Hangzhou.
The company
is purely focused on research rather than commercial products – the DeepSeek
assistant and underlying code can be downloaded for free, while DeepSeek’s
models are also cheaper to operate than OpenAI’s o1.
In an
interview with Chinese media, Liang said “AI should be affordable and
accessible to everyone”. Liang also said that the gap between US and Chinese AI
was only one to two years.
The DeepSeek
development raises doubts over the necessity for hefty investment in AI
infrastructure such as chips and the market-leading role of US tech companies
in AI, which in turn threatens to put American tech sector valuations under
pressure.
DeepSeek
claims R1 cost $5.6m to develop, compared with much higher estimates for
western-developed models, although experts have cautioned that may be an
underestimate. Last year Dario Amodei, the co-founder of leading AI firm
Anthropic, put the current cost of training advanced models at between $100m
and $1bn.
Analysts at
US investment bank Goldman Sachs raised the alarm over AI spending last year by
publishing a note in June with the title “Gen AI: too much spend, too little
benefit?”
It asked if
a $1tn investment in AI over the next few years will “ever pay off”, voicing
concerns about a return on spending that may have been crystalised by DeepSeek.
The
pan-European Stoxx 600 fell on Monday, and major European technology stocks
were down. The Dutch chipmaker ASML slid by 7%, while Germany’s Siemens Energy,
which provides hardware for AI infrastructure, was down nearly 20%, and
France’s digital automation company Schneider Electric fell by 9.5%.
Richard
Hunter, the head of markets at the platform Interactive Investor, said: “It
will almost certainly put the cat among the pigeons as investors scramble to
assess the potential damage it could have on a burgeoning industry, which has
powered much of the gain seen in the main indices over the last couple of
years.
“The larger
question has suddenly become whether the hundreds of billions of dollar
investment in AI needs re-evaluation.”
Dr Andrew
Duncan, the director of science & innovation at the UK’s Alan Turing
Institute, said the DeepSeek development was “really exciting” because it
“democratised access” to advanced AI models by being an open source developer,
meaning it makes its models freely available – a path also followed by Mark
Zuckerberg’s Meta with its Llama model.
“Academia
and the private sector will be able to play around and explore with it and use
it as a launching,” he said.
Duncan
added: “It demonstrates that you can do amazing things with relatively small
models and resources. It shows that you can innovate without having the massive
resources, say, of OpenAI.”
Sem comentários:
Enviar um comentário