Case Study: WeRide maximizes GPU utilization and cuts costs with Weka

WeRide Runs Its Driverless Robotaxis in the Hybrid Cloud with the WEKA® Data Platform

WeRide, a smart mobility company developing Level 4 autonomous driving technologies, needed to keep its GPU farm fully utilized while managing petabyte-scale AI data, mixed file sizes, and tight cost constraints across its edge-to-core-to-cloud pipeline. The company also wanted a hybrid storage approach that could reduce datacenter footprint, support metadata-heavy workloads, and run on commodity hardware with strong technical support.

Weka addressed this with WekaFS, deployed in a hybrid model across on-premises commodity Intel x86 servers and AWS China to deliver high-bandwidth I/O to GPUs. The solution helped WeRide maximize GPU utilization, reduce idle time, lower datacenter footprint, and improve cost efficiency, while giving engineers a storage utility that was transparent to end users and scalable for future growth.

WeRide

Paul Liu

Engineering Operations Lead