nvidia (NVDA 8.93 %) We created the history of the stock market on Monday, January 27, but it was not a good kind. Chip manufacturers have fallen by 17 % due to concerns about the artificial intelligence (AI) model of Chinese emerging companies. The steep descent erased $ 58.9 billion market value. This is the largest day for all companies.
What did meltdown cause? Despite the regulations from the US government that NVIDIA’s most advanced AI chips to China, DeepSeek has created a large -scale language model comparable to the performance of a more sophisticated model created in the United States. Is told. There is little money, no cutting -edge NVIDIA chip.
The news was miserable for NVIDIA shareholders. However, many warrier analysts regard this sale as an overreaction for investors to create a long -awaited opportunity.
Why DeepSeek’s breakthrough AI model crashed NVIDIA stock
Last week, DeepSeek published a research paper claiming that the R1 inference model comparable to the performance of O1 problem solving models on Openai on a specific benchmark. China’s emerging companies have also claimed that they spent less than $ 6 million in large language model training, and said they had completed training only with the 2,048 NVIDIA H800 Graphics Proseshin (GPU). Importantly, the H800 GPU was specially designed to comply with export restrictions.
In comparison, OPENAI spent more than $ 100 million training of GPT-4 models and used a more powerful NVIDIA H100 GPU. Although the company does not clarify the exact number, analysts estimate Openai using over 10,000 processors to train GPT-4. Considering that the meta platform uses 16,000 NVIDIA H100 GPUs to train LLAMA 3 models and spend $ 60 million, the estimate seems to be plausible.
The meaning is surprising to NVIDIA. If DeepSeek trains R1 using less powerful chips, US companies can theoretically reduce spending by imitating the training technology adopted by Chinese AI. 。 Next, hyper -scale companies such as Amazon, Alphabet, Meta Platforms, and Microsoft may spend less expenses than expected in the next few years.
Image source: Getty image.
DeepSeek’s breakthrough may have silver lining for NVIDIA shareholders.
DeepSeek trained the R1 model with impressive efficiency, but many analysts consider it as positive development. They believe that it accelerates the pace where artificial intelligence is adopted and promotes more demand for the NVIDIA GPU. In some industries, experts in the industry are questioning the validity of DeepSeek’s claims on cost and infrastructure.
Alexandr WANG, CEO of Scale AI, believes that DeepSeek has 50,000 NVIDIA H100 GPUs, but has not discussed research papers because it violates US export management. The H100 is the second most powerful NVIDIA GPU today and is very expensive. If DeepSeek had so many H100, they probably underestimated model training costs. Morningstar Brian Colero has maintained a target price of $ 130 per share of NVIDIA shares, following Deepseek News. “I doubt that the major cloud vendors and AI builders will pause the plan,” he wrote to the client. “We believe that high -tech companies will continue to purchase all GPUs as part of this AI Gold Rush.” Dan Ives of Wedbush Securities is “wrong in response to the news market.” I have stated. Based on about 25 technology companies, IVES believes that AI spending in the United States is not affected by recent DeepSeek. He argues that cheaper training has accelerated AI recruitment and sells it as one of the best purchase opportunities in the past 10 years. Wedbush kept NVIDIA’s target price of $ 175 per share. MOOR Insights & Strategy’s PATRICK MOORHEAD believes that demand for NVIDIA GPU will continue to increase until companies achieve artificial general information (AGI). He also believes that the new training method used by DeepSeek will accelerate AI adoption by reducing costs. Moore Head believes it is positive to create a new use case in a shorter time. Bernstein’s Stacy Rasgon believes that DeepSeek is underestimating training costs related to the R1 model. But he believes that the efficient training method is good. In an interview with CNBC, Rasgon stated that the situation was an example of Jebon’s Paradox. This is an economic principle more than the decrease in the cost of new technology is offset by the subsequent increase in demand. Bernstein has maintained a target price of $ 175 per share. Fundstrat Global Advisors Tom Lee, called the stock market drawdown, compared the sale and situation following the COVID-19 start. “This pullback in NVIDIA is proved to be an opportunity to buy,” he told CNBC.
Investors need to keep in mind that the situation of DeepSeek has evolved rapidly and analysts can change their opinions. Meta Platform and Microsoft will report their financial results on January 29, and Alphabets and Amazon will continue on February 4. If the management team of these companies holds a quarterly revenue call, the market may be more clear.
However, several analysts now believe that DeepSeek’s breakthroughs hardly affect the long -term demand of the NVIDIA GPU. In that sense, the news was miserable for NVIDIA shareholders, but there is a silver lining: NVIDIA’s stock is much cheaper than last week, creating attractive opportunities for future investors. 。
In fact, eight of the 39 analysts following the company have updated their predictions in the past two days, but the average target price of NVIDIA is $ 177 per share. This means the current stock price of $ 118 to 50 %.
SUZANNE FREY, an executive of Alphabet, is a member of the Motley Fool’s Board of Directors. John Mackey, a former CEO of Amazon subsidiary WHOLE FOODS MARKET, is a member of the Motley Fool’s Board of Directors. Randi Zuckerberg, a former director of market development, Facebook spokeswoman, and sister of Mark Zuckerberg, the CEO of Meta Platform, is a member of the Motley Fool’s Board of Directors. TREVOR JENNEWINE is the role of Amazon and NVIDIA. Motley Fool has a position in Alphabet, Amazon, Meta Platforms, Microsoft, and NVIDIA. Motley Fool recommends the following options: A $ 395 phone call at Microsoft in January 2026 and a $ 405 call to the short -term Microsoft in January 2026. Motley fools have a disclosure policy.
