In This Article:
Feb 27 - Nvidia (NVDA, Financial) Chief Executive Officer Jensen Huang said the next wave of artificial intelligence will demand 100 times more computing power than previous models, citing emerging reasoning techniques that process questions step by step. He made the remarks during an interview with CNBC following the company's fiscal fourth-quarter earnings release on Wednesday.
Nvidia reported a 78% year-over-year jump in total revenue, reaching US$39.33 billion, and a 93% surge in data center sales to US$35.6 billion. Despite topping analyst estimates, the stock has not fully recovered from its 17% slide on Jan. 27, which was triggered by concerns that Chinese AI lab DeepSeek might achieve higher AI performance at lower infrastructure costs.
Huang disputed that view, explaining that DeepSeek's open-source approach popularized a new generation of AI models requiring substantially greater chip capacity. He highlighted models like R1, GPT-4, and Grok 3 as examples of advanced reasoning systems.
In China, revenue dropped roughly in half from before export controls, but Nvidia is also restricted in that market. With the GB200 chip able to generate AI content 60 times faster than its export restricted version, Huang said developers often work around hardware limitations to create software as it is. Nvidia has held the lead in AI hardware for several quarters, but analysts see potential challenges ahead if China's permissive tech environment shifts or if competition increases.
This article first appeared on GuruFocus.