1 min read
Add as a preferred source on Google

Pony.ai Turns to Chinese Chips as AI Training Demand Exceeds Supply

Chief Financial Officer Leo Wang said computing demand exceeds supply and described domestic chips as a hedge against possible tighter U.S. restrictions on Nvidia.

Unmarked processor chips sit beside a compact computing module / TokenPost.ai
Unmarked processor chips sit beside a compact computing module / TokenPost.ai

Pony.ai is increasing its use of China-made chips to address an AI training capacity gap as computing demand outstrips supply, Chief Financial Officer Leo Wang said.

The autonomous-driving company is also working to make its models more efficient and using midtier computing resources to address the shortfall, Wang said.

Pony.ai is treating domestic chips as a hedge against possible tighter U.S. restrictions on Nvidia. Wang’s comments describe preparation for a potential policy shift, not a response to an announced change in U.S. rules.

Wang said Chinese companies have spent years preparing alternatives, though their cost-effectiveness is not yet ideal. He said the options give Pony.ai more flexibility.

Simon Yoon

Reporter

Simon Yoon reports on blockchain technology for TokenPost. Send corrections or tips to info@tokenpost.com.

Loading…