China's DeepSeek startup roiled artificial intelligence-driven stocks on Monday as Wall Street mulled huge investments by technology industry giants in AI infrastructure as well as demand for Nvidia chips. DeepSeek's apparent advances have raised questions over the computing power needed to develop AI systems, a key driver for AI stocks.
The Nasdaq fell 3.1% on Monday as Nvidia stock led decliners.
DeepSeek released a large language model (LLM) in late December that had performance comparable to OpenAI and other AI systems despite U.S. government restrictions on exports of advanced chips. On Jan. 20, DeepSeek released another LLM that it said was trained at a fraction of the cost of AI systems from OpenAI, Alphabet's Google, Meta Platforms and other U.S. companies.
The Wall Street Journal reported on the AI phenomenon on Sunday.
While OpenAI and Google have developed proprietary AI models, DeepSeek has relied on super efficient "open source" technology and innovative model-training techniques. Reports have differed over what Nvidia chips DeepSeek has been using to build AI models. While some reports point to less powerful H800 chips, others claim DeepSeek's parent, High-Flyer, has been able to buy more advanced H100 Nvidia AI chips despite U.S. export restrictions.
On the stock market today, Nvidia stock tumbled 17% to close at 118.42 as many other AI stocks also retreated, including software maker Palantir. PLTR stock dipped 4.5% to 75.44. S&P 500 nuclear power giants Constellation Energy and Vistra pulled back as analysts reevaluated the energy needs of AI-centric data centers.
AI Stocks: DeepSeek Raises Questions
"DeepSeek clearly doesn't have access to as much compute as U.S. hyperscalers and somehow managed to develop a model that appears highly competitive," said Raymond James analyst Srini Pajjuri in a Sunday report. "The natural question is, how would DeepSeek's emergence impact compute intensity growth and the demand for hardware/semiconductors?"
Startups OpenAI and Anthropic have been battling Google, Meta and others in developing large, multimodal and small language foundation models.
DeepSeek's AI Assistant has overtaken OpenAI's ChatGPT on Apple's app store in the U.S., according to data research firm Sensor Tower.
Further, DeepSeek has claimed it spent only $5.6 million over two months to develop its latest AI model.
Meanwhile, DeepSeek's success could weigh on U.S. earnings reports, noted economist Ed Yardeni in a report published on Sunday, amid enormous capital spending increases by Google, Meta, Microsoft and others. Microsoft is the biggest investor in OpenAI.
DeepSeek Bad News For Mag-7, Nvidia?
"This might be bad news for the Magnificent 7 that had plans to dominate the AI market with their expensive AI services," Yardeni added. "The question is whether they will disappoint because their capital spending on AI is soaring faster than are their revenues. That could squeeze their collective profit margin."
He says chipmaker Nvidia also could be impacted if customers need fewer and less sophisticated chips. "It might be good news for the Mag-7 that can learn from DeepSeek to design AI systems with cheaper GPUs. That would reduce their capital spending and boost their profits. It might not be a happy development for Nvidia," Yardeni added.
Microsoft and Meta report fourth quarter results on Jan. 29. Google earnings are due Feb. 4. Amazon.com will likely follow on Feb. 6.
Meta last week said it plans to hike capital spending to a range of $60 billion to $65 billion in 2025, topping Wall Street estimates of $51.4 billion.
Meta stock initially fell but closed up 1.9% to 659.88. Microsoft stock retreated 2.1% to 434.56. Google stock fell 4.2% to 191.81.
DeepSeek Targets Artificial General Intelligence
Deepseek CEO Liang Wenfeng also founded High-Flyer, a Chinese quantitative hedge fund.
In a recent interview, Wenfeng said DeepSeek aims to develop human-like Artificial General Intelligence with novel architectures.
"Our destination is AGI, which means we need to study new model structures to realize stronger model capability with limited resources," he said.
Wedbush analyst Daniel Ives, a perma-bull on AI stocks, has a different take.
"It remains to be seen if DeepSeek found a way to work around these chip restrictions rules and what chips they ultimately used as there will be many skeptics around this issue given the information is coming from China," Ives said in a research note.
He added: "Clearly tech stocks are under massive pressure led by Nvidia as Wall Street will view DeepSeek as a major perceived threat to US tech dominance and owning this AI Revolution. While the model is impressive and it will have a ripple impact, the reality is that Mag 7 and U.S. tech is focused on the AGI endgame with all the infrastructure and ecosystem that China and especially DeepSeek cannot come close to in our view."
In cloud computing services, DeepSeek is making its AI models available to software developers at prices significantly lower than OpenAI's models.
AI Stocks: Nvidia Earnings Due Feb. 26
In addition, Nvidia has been a bellwether for AI stocks. Earnings for Nvidia stock are due Feb. 26. Nvidia is among AI stocks to watch.
For Nvidia stock, ramping up production of next-generation Blackwell AI chips in 2025 has been a key issue.
Also, Nvidia has criticized new chip export restrictions announced by the Biden administration. The global AI chip control system gives priority to 20 U.S.-friendly countries and limits the computer processing power available to China and other countries.
It remains to be seen whether President Trump follows through on the new export rules.
Also, at Raymond James, Pajjuri offered a different scenario.
Open Source AI Models Improve
"A logical implication is that DeepSeek will drive even more urgency among U.S. hyperscalers to leverage their key advantage (access to Nvidia GPUs) to distance themselves from cheaper alternatives."
The AI trend has relied mainly on large language models, or LLMs, which require massive amounts of data to be trained. Large language models allow users to interact with AI systems without the need to write algorithms.
The capabilities of open-source models, which are free to developers, have advanced rapidly, with Meta emerging as the leading proponent.
Meanwhile, Google stock has gained about 6% in 2025. Google stock holds an Accumulation/Distribution Rating of A-minus. That institutional ownership rating analyzes price and volume changes in a stock over the past 13 weeks of trading. A+ signifies heavy institutional buying; E means heavy selling.
Follow Reinhardt Krause on Twitter @reinhardtk_tech for updates on artificial intelligence, cybersecurity and cloud computing.
