By Daniela L. Ginsburg
Boston University News Service
On Jan. 27, the China-based company Deepseek sent shock waves through Wall Street.
In its stunning latest version, the Chinese-made artificial intelligence platform rose the ranks in the Apple Store as it offered a cutting-edge model for a portion of the budget. United States President Donald Trump called it a “wake-up call” for US companies who need to be “competing to win.”
Founded in 2023 by Liang Wenfeng, the private company claims to utilize fewer chips and a cheaper AI training model than its rivals. Deepseek trained its AI model for around $6 million, compared to ChatGBT which reported to have spent around $100 million.
Thomas Gardos, an Associate Professor of Computing and Data Science at Boston University, said that these key differences caused a panic and resulted in many major private companies pulling and selling off stock, creating a stock-market selloff. Gardos also said that besides these two major selling points, the model remained relatively the same as its rival competitors.
“At the consumer level, there’s really no differences,” Gardos said. “That’s why there was a sell off, they seem to be competitive with open AI at a fraction of the cost.”
Deepseek’s success in overriding the belief that AI can only be advanced through outrageous budgets and first-rate chips is what resulted in the rapid decline of stock prices in companies, such as Nvidia. Nvidia, the American company that creates hardware chips called GPUs which are used to improve AI models such as ChatGBT, had been the most valuable company in the world until Jan. 27 when Deepseek stormed its throne.
Nvidia appeared to have been affected the most with stocks plunging by 17% that same day. The company lost around $600, the largest single-day drop in history.
Gardos said that another key difference between Deepseek and its competitors is open sourcing. Open-source intelligence is when the public has the ability to freely download the AI model. The Chinese company follows this system as users are able to run it on personal servers through the Deepseek website.
Mixed with cheaper costs, this allows users to experiment with the code and customize it to their needs, further accelerating the AI model’s benefits.
Even though U.S. companies were hit hard by Deepseek’s breakthrough, the company is being globally celebrated for its testament to China’s innovation.
“The positive view is that they’re contributing,” Gardos said. “This will accelerate science even more.”
