Taipei:Nvidia Corp. and Hon Hai Precision Industry Co., also known as Foxconn, have successfully improved the assembly success rates for Nvidia's GB300 NVL72 rack-scale AI platform by integrating robotic automation into the manufacturing process.

According to Focus Taiwan, The collaboration between the American AI chip designer and the Taiwanese manufacturing giant focused on the critical assembly steps of the GB300's tester tray, a component essential for verifying the functionality and performance of the AI computing modules. The use of robots in tasks such as busbar assembly and the insertion of electrical connectors has resulted in success rates for key assembly steps exceeding 90 percent.

Nvidia highlighted that the automation system achieved a more than 95 percent success rate in busbar assembly, which involves securing a metal bar with screws at 16 locations. The goal was to complete this process in under 124 seconds. For multi-connector insertion, which involves four cable connectors of varying sizes, success rates ranged from 90 to 95 percent, targeting completion within 72 seconds.

The need for flexible automation systems was emphasized due to the rapid design cycles and relatively low production volumes of the GB300, making fixed automation impractical. Therefore, the robots must be adaptable and robust to manage variations in the assembly process.

Hon Hai, a major supplier to Nvidia, previously announced plans to collaborate with the company to build a smart manufacturing plant in Houston, Texas, using Nvidia's humanoid robots. These efforts are part of Hon Hai's broader strategy to expand its use of robotics in other U.S. locations and to diversify its business into electric vehicles, robots, and digital healthcare, leveraging AI, semiconductor, and communications technologies.