Nvidia reveals AI robots that taught themselves to install GPUs into motherboards — video shows robot ‘solve high-precision tasks like… installing GPUs all by itself’

Nvidia has demonstrated robots that can teach themselves to perform high-precision physical tasks, including installing graphics cards into motherboards, organizing pins, and cutting zipties. The ENPIRE project gave coding agents control of a fleet of robots with computing resources and instructed them to complete tasks as quickly as possible without errors, after which the robots learned to look for visual clues, practice novel skills, and troubleshoot failures through repeated attempts on actual hardware. The framework uses four core modules that handle automatic environment reset, policy improvement, parallel robot evaluation, and analysis of failure modes to guide the robots toward better performance. The project will be made open source, allowing researchers to run similar self-improving robot systems outside of corporate labs.