Pony.ai Turns to Local Chips to Address AI Training Compute Capacity Constraints

Written by

in

Pony.ai is shifting toward domestically manufactured processors to ease limited AI training compute capacity while preserving the option to respond to tighter U.S. restrictions on Nvidia. Chief Financial Officer Leo Wang said the company is addressing compute shortages by improving model efficiency and using mid-tier computing resources; locally made chips can add flexibility, but their cost-effectiveness is not yet ideal.

Source:Global Economic Intelligence · Read Original

Comments

Leave a Reply

Your email address will not be published. Required fields are marked *