China’s DeepSeek AI hits Nvidia where it hurts

A chatbot created by Chinese artificial intelligence startup DeepSeek climbed to the top of the Apple App Store charts in the US this week, dethroning OpenAI’s ChatGPT as the most downloaded free app. The titular AI assistant is based on open source DeepSeek models that the company claims you can train at a fraction of the cost using significantly fewer chips than the world’s leading models. The claim enraged financial markets, and Nvidia’s share price fell more than 12 percent in pre-market trading.

DeepSeek also claims that it only needed about 2,000 specialized Nvidia chips to train the V3 version, compared to the 16,000 or more required to train the leading models, according to the New York Times. These unverified claims are leading developers and investors to question the computationally intensive approach favored by the world’s leading artificial intelligence companies. And if that’s true, it means DeepSeek engineers had to get imaginative in the face of trade restrictions designed to ensure U.S. dominance in artificial intelligence.

Nvidia, Microsoft, OpenAI and Meta are investing billions in AI data centers – $500 billion alone for the Stargate project, of which $100 billion is believed to be earmarked for Nvidia. Investors and analysts are now wondering whether it’s money well spent, as Nvidia, Microsoft and other companies with a significant stake in maintaining the AI ​​status quo are trending downward in pre-market trading.