Nvidia CEO Jensen Huang has issued a stark warning about China’s accelerating momentum in artificial intelligence, pointing to the country’s rapid infrastructure development and expanding energy capacity as key advantages over the United States.
Speaking at a CSIS event, Huang said building a data center in the U.S. can take around three years, while China can complete massive projects «in a weekend,» highlighting its ability to deploy AI-ready facilities at unprecedented speed. He also noted that China now has twice the energy capacity of the U.S., with its supply «going straight up,» while American growth remains flat — a critical factor given AI’s enormous power requirements.
While stressing that Nvidia remains «generations ahead» of China in chip technology, Huang cautioned that the U.S. cannot rely on its lead indefinitely.
His comments come as a new study shows Chinese open-source large language models (LLMs) have captured nearly 30% of global AI usage, a meteoric rise from just 1% last year. The report, compiled by OpenRouter and Andreessen Horowitz, found Chinese models now rival global competitors in weekly token usage, led by DeepSeek, Alibaba’s Qwen and Moonshot AI’s Kimi K2.
Chinese-language prompts are now the second-largest source of global LLM token volume, trailing only English.
The findings underscore China’s fast-growing influence in the AI ecosystem — a trend Huang says poses serious competitive pressure as Beijing continues to build data centers, expand energy capacity, and lower the cost of AI development. Many analysts expect China’s upward trajectory to continue, setting the stage for an intense race for AI leadership.
Kursiv Uzbekistan also reports that Uzbekistan has taken a major step toward becoming Central Asia’s fintech hub after President Shavkat Mirziyoyev signed a sweeping decree on 28 November approving a regulatory sandbox for stablecoin payments, set to launch on 1 January 2026.
