Case Study: a leading North American automobile manufacturer achieves exabyte-scale AI training data performance with MinIO AIStor

a leading north american automobile manufacturer powers exabyte AI training data with MinIO AIStor, saving millions annually

The customer, a leading North American automobile manufacturer, faced the challenge of managing exabyte-scale AI training data for its autonomous vehicle programs. Their existing storage infrastructure was fragmented and unable to deliver the required throughput to GPU clusters, leading to idle compute resources. Additionally, unsustainable public cloud costs for a portion of their data made their program economically challenging. They selected MinIO's AIStor product to build a new, high-performance AI data infrastructure.

By implementing MinIO AIStor, the automaker created a unified, on-premises data lake capable of handling over 600,000 S3 requests per second. The solution enabled them to deploy 300+ node clusters in a single weekend and scale to over one exabyte of usable capacity. This resulted in millions of dollars in annual savings by eliminating cloud fees and ensured GPU clusters were never starved for data, achieving throughputs of up to 21.8 TiB/sec. MinIO's platform now supports all their AI workloads, including autonomous driving model training and factory vision inspection.