DeepSeek: The Chinese AI Company Shaking Up the Stock Market
A surge of interest in an artificial intelligence chatbot developed by the Chinese technology startup DeepSeek has caused significant volatility in stock markets. The development is also fueling debates about the global race between the United States and China to lead in AI technology.
DeepSeek’s AI assistant quickly became the top downloaded free app on Apple’s App Store, driven by curiosity about this competitor to OpenAI’s ChatGPT. Some U.S. tech industry observers are concerned that DeepSeek may have matched the advancements of American companies at the forefront of generative AI, but at a far lower cost. If true, this raises questions about the large sums of money U.S. tech firms say they plan to invest in the data centers and computer chips required to drive further AI innovation.
People reflected in a window with a slogan about AI at a representation of a company ahead of the World Economy Forum in Davos, Switzerland, Sunday, Jan. 19, 2025. (AP Photo/Markus Schreiber)
However, hype and misunderstandings surrounding DeepSeek’s technological innovations have also led to market confusion.
“The models they built are fantastic, but they aren’t miracles either,” said Bernstein analyst Stacy Rasgon, who follows the semiconductor industry. He was one of several stock analysts who characterized Wall Street’s response as overblown. “They’re not using any innovations that are unknown or secret or anything like that,” Rasgon explained. “These are things that everybody’s experimenting with.”
What is DeepSeek?
The startup DeepSeek was established in Hangzhou, China in 2023 and released its first AI large language model later that year. Its CEO, Liang Wenfeng, previously co-founded High-Flyer, one of China’s leading hedge funds, which focuses on AI-driven quantitative trading. By 2022, the fund had acquired a cluster of 10,000 Nvidia A100 graphics processor chips, based in California, that are used to build and run AI systems, according to a post on the Chinese social media platform WeChat. The U.S. soon after restricted sales of those chips to China.
DeepSeek has said its recent models were built with Nvidia’s lower-performing H800 chips, which are not banned in China. This suggests that the most advanced hardware may not be essential for cutting-edge AI research.
DeepSeek began attracting more attention in the AI industry last month when it unveiled a new AI model, which it claimed was comparable to those of U.S. companies such as OpenAI (the creator of ChatGPT). Moreover, DeepSeek stated its model was more cost-effective in its use of expensive Nvidia chips to train the system on large amounts of data. The chatbot became more broadly accessible when it became available on Apple and Google app stores early this year.
However, it was a follow-up research paper, published last week — on the same day as President Donald Trump’s inauguration — that triggered the panic that followed. That paper was about an additional DeepSeek AI model called R1 that showed advanced “reasoning” skills, such as the capability to rethink its process when resolving a math problem, and was substantially cheaper than a similar model sold by OpenAI called o1.
“What their economics look like, I have no idea,” Rasgon said. “But I think the price points freaked people out.”
The ‘Sputnik’ Backdrop
Behind the drama surrounding DeepSeek’s technical capabilities lies a debate within the U.S. over the best way to compete with China on AI. “Deepseek R1 is AI’s Sputnik moment,” said venture capitalist Marc Andreessen in a Sunday post on social platform X, referencing the 1957 satellite launch that sparked a Cold War space exploration race between the Soviets and the U.S.
Andreessen, advisor to Trump on tech policy, has warned that excessive regulation of the AI industry by the U.S. government will hinder U.S. companies and allow China to gain an advantage.
However, the attention on DeepSeek also threatens to undermine a key strategy of U.S. foreign policy in recent years: restricting the sale of U.S.-designed AI semiconductors to China. Some experts on U.S.-China relations believe this is not accidental.
“While the technology innovation is real, the timing of the release is political in nature,” said Gregory Allen, director of the Wadhwani AI Center at the Center for Strategic and International Studies. Allen compared DeepSeek’s announcement last week to the release of a new phone by Huawei, a Chinese company that is under U.S. sanctions, during diplomatic discussions about the Biden administration’s export controls in 2023. “Trying to show that the export controls are futile or counterproductive is a really important goal of Chinese foreign policy right now,” Allen said.
On Monday, Trump commented that DeepSeek’s breakthrough was “good because you don’t have to spend this much money.”
Speaking to House Republicans in Miami on Monday, Trump described the DeepSeek news as “positive” if accurate because “you won’t be spending as much and you’ll get the same result.” He called the development a “wakeup call for our industries that we need to be laser focused on competing to win.” Trump signed an order on his first day in office last week stating that his administration would “identify and eliminate loopholes in existing export controls,” signaling that he is likely to continue and strengthen Biden’s approach.
DeepSeek’s advancement in AI without the same level of spending could possibly undermine the approximately $500 billion AI investment by OpenAI, Oracle, and SoftBank that Trump mentioned at the White House. Nvidia’s stock dropped 17% on Monday, but the company, in a statement, lauded DeepSeek’s work as “an excellent AI advancement” that utilized “widely-available models and compute that is fully export control compliant.”
What Makes DeepSeek Different?
One thing that distinguishes DeepSeek from competitors such as OpenAI is that its models are “open source,” meaning that key components are freely available for anyone to access and change, although the company has not disclosed the data it used for training.
However, what has gained the most admiration about DeepSeek’s R1 model is what Nvidia refers to as a “perfect example of Test Time Scaling” – which is when AI models effectively show their train of thought and then use that for further training without needing to feed them additional data sources.
“It’s just thinking out loud, basically,” said Lennart Heim, a researcher at Rand Corp.
OpenAI’s reasoning models, starting with o1, do the same, and it’s likely that other U.S.-based competitors such as Anthropic and Google have similar capabilities that haven’t been released, Heim said. But “it’s the first time that we see a Chinese company being that close within a relatively short time period. I think that’s why a lot of people pay attention to it,” Heim said. “I used to believe OpenAI was the leader, the king of the hill, and that nobody could catch up. Turns out this is not completely the case.”
