A young Chinese AI startup, Deepseek, has fueled a massive route In U.S. tech stocks on Monday, as its highly competitive—and potentially shockingly profitable—models cast doubt on the hundreds of billions of dollars America’s biggest tech companies are spending on artificial intelligence.
The appearance of Deepseek has shaken the confidence of investors in Ai The story that has been driving the American bull market for the past two years. It is called into question Hype around Nvidia chips and unzipped all the way to the market to reach stocks electricity producers which were set up to drive AI data center demand.
Here’s how the sell-off on Wall Street in the Markets unfolded:
A new model of reasoning
Deepseek was founded in May 2023 by Liang Wenfeng, who partially funded the company with his hedge fund AI operation. In late December, an AI developer launched a free model of a large open source language that is said It took just two months and less than $6 million to build.
On January 20, the Hangzhou, China-based company released R1, a reasoning model that outperformed Open AI’s latest O1 model in many third-party tests.
Deepseek aims to differentiate itself from its competitors with its reasoning capabilities, which means that before delivering a final answer, the model first generates a “Chain of Thought” to improve the accuracy of his answers.
Top performance model
The buzz around Deepseek’s R1 seemingly picked up steam after Alexandr Wang, CEO of Scale AIit advertised its competitiveness against the best products of US Megacap Tech Giants, who were thought to be leading the AI war. Scale AI provides data to help companies train their AI tools.
“What we found is that DeepSeek, which is China’s leading AI lab, their model is actually a top performer, or about as good as the best American models,” Wang told CNBC from World Economic Forum In Davos, Switzerland, last week.
Wang said Deepseek actually has more H100 chips from Nvidia than expected — about 50,000 of them. Those are chips choice processor for AI companies in the US, such as OpenAnd the US has banned the sale of advanced AI chips to China.
Shares of Nvidia hit 3% on Friday as chatter about Deepseek began to gather.
No. 1 Application
Over the weekend, the hype surrounding Deepseek reached a fever pitch on social media.
Marc Andreessenco-founder and general partner of venture capital firm Andreessen Horowitz, sang Deepseek’s praises at X, saying the R1 model was “one of the most incredible and impressive breakthroughs” he had ever seen. Andreessen’s portfolio includes Airbnb and dozens of AI companies.
Technology investor Chamath Palihapitiya At X, he pointed to Deepseek’s “very good” report, in which he said its R1 model “basically cracked one of the holy grails of AI: getting models to understand one step at a time, without relying on huge surveillance datasets.”
Chinese AI app Deepseek seen in Apple’s US App Store on iPhone 12 on January 27, 2025. In the ranking of free applications, Deepseek was even ahead of chatgpt from OpenAi.
Alliance Image | Getty Images
At this point, Deepseek’s Mobile application increased to the top AppleThe App Store takes over the charts over the weekend in the US, marking a tangible threat to pricing models like Openai’s Chatgpt.
US futures had a big night on Sunday, and investors woke up to a sea of red on Monday morning.
NASDAQ COMPOSITE, one-day
And dear chip producer too Nvidia Monday saw shares up more than 12%, on track for their worst day since March 2020.
Nvidia, 1-day
Other chip makers as well as energy providers were big. Technicolor Nasdaq Composite was by as much as 3.6% on Monday, and megacap names were withdrawn.
