Investors attacked technological values on Monday, picked up companies such as Nvidia and Oracle, since the appearance of a Chinese low -cost artificial intelligence model called Depseek, questioned the domain of Western companies in the sector in the sector .
The Emerging Deepseek company launched a free assistant last week that, he says, uses less data and costs a fraction of the cost of traditional companies models, which possibly marks an inflection point at the level of investment necessary for AI .
Nasdaq 100 futures lost almost 4%, suggesting that the index could later suffer their greatest daily drop since September 2022 if these losses are maintained. Those of the S&P 500 gave up 2%.
The actions of the Nvidia chips manufacturer lost 10%; Your rival Oracle, 8%; and the Palantir data analysis company, 7% in operations prior to the opening of markets.
Deepseek, who had already surpassed his American rival Chatgpt on the number of downloads in the Apple Store, offers the perspective of a viable and cheaper alternative to AI, which has raised doubts about the sustainability of the level of spending and investment In the western companies, including Apple and Microsoft.
From Tokyo to Amsterdam, the actions of the AI actors collapsed.
“We still do not know the details and nothing has been confirmed 100% in regards to the statements, but if there really has an advance in the cost of training models of more than 100 million dollars to this alleged figure of 6 million , it is very positive for productivity and end users, since the cost is obviously much lower, which implies a lower access cost, ”said Jon Withaar, of Pictet Asset Management.
Advertising around AI has promoted a huge influx of capital to the stock market .
You may be interested: EU companies that have abandoned the objectives of diversity, equity and inclusion
Deepseek’s ‘Sputnik’ causes strong falls in the big companies of AI
Little is known about Hangzhou’s small business behind Depseek. His researchers wrote in an article last month that the Deepseek-V3 model, launched on January 10, used NVIDIA H800 chips for training, spending less than 6 million dollars, the figure referred to Withaar de Pictet .
The H800 chips are not high range. Initially developed as a reduced capacity product to avoid sales restrictions to China, they were subsequently prohibited by the sanctions of the United States.
Silicon Valley Risk capitalist Marc Andreessen said in a publication in social network X on Sunday that Deepseek’s R1 model is the “Sputnik Moment” of AI, in reference to the launch by the former Soviet Union of a satellite that marked The beginning of the space race in the late 1950s.
“Deepseek R1 is one of the most amazing and impressive advances that I have ever seen and, as an open source, it is a deep gift for the world,” he said in another publication.
In Europe, ASML, which has among its customers the Taiwanese TSMC, Intel and Samsung, collapsed almost 11%.
In Japan, the SoftBank Group startup investor lost more than 8%. Last week he announced a commitment of 19,000 million dollars to finance Stargate, a joint company of data centers with OpenAI.
Large technological ones have increased the expense in the development of AI capacities and optimism on possible benefits has triggered the assessments of actions.
Only Nvidia has risen more than 200% in about 18 months and quotes 56 times the value of its benefits, compared to a 53% rise in Nasdaq, which quotes a multiple of 16 with respect to the value of the profits of its Components, according to LSE data.
With Reuters information.
Subscribe to our YouTube channel and do not miss our content
