A $ 5.6 million of artificial intelligence breaks in Chinese companies wiped nearly $ 600 billion from Nvidia market value on Monday, breaking the Wall Street confidence that the cost of AI companies would continue and deal with an obvious shot by US technological leadership.
Yet many in the Silicon Valley believe Wide sale is an excessive reaction to the latest Deepseek model, which they claim could encourage the wider adoption and usefulness of AI radically reduction of technology costs, maintaining demand for Nvidia’s chips.
Pat Gelsinger, recently forced as the Intel CEO, was among those who bought his former Nvidia rival section on Monday. “The market reaction is wrong: lowering the cost of AI will expand the market,” the LinkedIn post said. “Deepseek is an incredible engineering part that will attract a greater acceptance of AI.”
Nvidia It has become the world’s most respected company last year because investors are betting on a insatiable appetite of large technological companies for its powerful AI processors. Cypsum executive director Jensen Huang predicted that AI data centers worth 1tn will be built in the next few years.
Basically that confidence was the concept of AI “Skalani Law”, which was popularized by older leaders in Ai Starting such as Openi and Anthropic, which suggested that AI models are becoming smarter because more data and computer resources have been fed.
Deepseek’s edition of his highly capable R1 model – and a research work that openly explains how he was made – he seemed to break the magic of the Scaling Act, as his Chatbot skipped to the top of the iPhone -based App Store. Philadelphia semiconductor index has dropped 9.2 percent, which is the worst daily fall since March 2020.
Chinese Technology Champion Huawei appeared as Primary competitor Nvidia in China To conclude the chips. The Financial Times has previously reported that he has collaborated with AI companies, including Deepseek, to adjust models trained on NVIDIA GPU to start conclusions on his Ascend Chips.
“Huawei is getting better. They open because the Government says to large technological companies that they need to buy their chips and use them to conclude,” said one semiconductor investor in Beijing.
The announcement that started Spasm on Monday on the stock market over Nvidia arrived as they move now claim your leadership In AI over China, and how the biggest American technological companies are preparing to report their latest earnings. US President Donald Trump said Deepseek “should be an invitation to wake up for our industry that we have to be a laser focused on a winning competition.”
In December, Deepseek published a V3 model, which he claimed was comparable to those of Openi -and Google, but Trained him on a budget fraction with $ 5.6 million. The Chinese company announced that it only used 2,048 NVIDIA chips, which could be obtained without violating US export control that extinguished the latest American US chips products.
He then introduced his latest R1 model last week, the “reasoning model” that is comparable to Openi’s O1.
Further spooky investors, Deepseek engineers have been able to unlock larger performance by writing code without relying on Nvidia’s miracle software platform, which is widely considered key to the dominance of Chip -chip -a to dominate the AI development AI development.
“Deepsek has straightened the playground,” said Stephen Yiu, Main Investment Director of Blue Whale Growth, an investment fund supported by billionaire Peter Hargreaves, who reduced exposure to the magnificent seven groups of major American technological companies last month because of their concerns about their problems Ai.
The largest American technological company “had a monopoly approach to AI – the price of entry was in billions of dollars, otherwise there was no chance that you could cause the status quo,” Yiu said. Because of this, Deepseek’s arrival “is a very positive development of adoption, development and penetration AI,” he added.
The short sellers, who have set up ranks against NVIDIA’s high stock prices in recent weeks, were cheerful on Monday. Nvidia -I drop a 17 -percent stock price has generated $ 6.75 billion in profit for short sellers, according to Data Group S3 budgets.
“The Chinese subject has outlined open sources before earning all major American technological companies,” said one short seller with interests in numerous major AI companies. “They tell you that there is no value [in those companies’ AI models]Coming is. ”
However, some analysts challenged the idea that Deepseek’s breakthrough AI was so cheap to build.
Dylan Patel of Chip Consultancy Semianalysis estimated that Deepseek and his sister-owned company, HEDGE Fund of Visoko Flyer, have access to tens of thousands of NVIDIA GPU, which were used to train the predecessor of R1.
“Deepseek spent more than $ 500 million on GPU because of the history of the company,” Patel said. “Although their training was very effective, it required significant experimentation and testing.”
G Dan Hutcheson of Techinsights said the market reaction did not reflect who was most exposed to Deepseek’s breakthrough. “I don’t see it as a big hit to Nvidia, I consider it a bigger problem for companies like Openi trying to sell these services,” he said.
Nvidia claimed on Monday that Deepseek’s innovation would benefit and not develop, his business.
“Deepsek is a great progress of AI and the perfect example of scalating the examination time,” Nvidia said, referring to AI systems that consume more computer resources after the user sets up the question or set the task of “explaining” or taking more connected steps to answer. “The conclusion requires a significant number of NVIDIA GPU and the networking of high performance.”
The implication of Nvidia’s statement is that by pushing the boundaries of what is possible with AI models of “open code”, Deepseek has actually increased the demand for chips used for them.
Although Nvidia is best known for providing chips used for “training” or building a new AI system, he said that he now creates just as much revenue from the “conclusion” chips or the processing of user demands using the finished model.
Huang argued at the recent Podcast that the demand for the conclusion “will soon grow in a billion times” because of the new AI models that “understand” or need time to plan and deliver an answer to a complex inquiry.
“There are two use for Nvidia chips, training and concluding, and we are only at the beginning of the conclusion story,” said Jordan Jacobs, co -founder AI Investor Radical Ventures, who bought several Nvidia shares on Monday as lace stocks have fallen 17 percent. “As we see the world switches to AI, requires a huge upgrade of chips. It seems that sale is an excessive reaction and lack of understanding.”
“The market does not realize that this is great for Nvidia,” said Dmitry Shevelenko, the CEO of Perplexity, launching AI searches based in San Francisco, which counts chip manufacturers among their investors. “No matter what, Jensen wins.”
Additional Cristine Criddle Reporting in San Francisco and Eleanor Olcott in Beijing
