China’s Deepseek AI is sick of NVIDIA

The chatbot produced by China’s AI Startup Deepseek has been at the top of the Apple’s App Store chart this week and has made Openai’s Chatgpt the most downloaded free app. The Sijo AI assistant is driven by the open source model of Deepseek, and the company says that it can be trained as part of the cost using a much less chip than the world’s best model. The claim has surged in the financial market, and NVIDIA’s share price has fallen more than 12 % in all market transactions.

DeepSeek also claims that only 2,000 special chips are required from NVIDIA to V3 training. New York Times. This unproven claim leads developers and investors to question the computing -intensive approach preferred by the world’s best AI companies. And if it’s true, it means that DeepSeek engineers must get creativity in the face of trade restrictions to ensure AI’s domination.

NVIDIA, Microsoft, Openai, and Meta invest only $ 500 billion in the AI ​​data center for the Stargate project, and $ 100 billion is thought to be applied to NVIDIA. Investors and analysts are now wondering if they are well -made with other companies with a significant stake in maintaining AI status with NVIDIA, Microsoft and other companies.