Case Study: Zenuity achieves faster AI training and safer autonomous driving with Pure Storage FlashBlade

Developing Vehicles of the Future With Artificial Intelligence

Zenuity, a joint venture of Volvo Cars and Autoliv, develops deep-learning software for autonomous vehicles and faced a classic AI challenge: massive, safety-critical sensor datasets and the need to iterate models quickly without compromising time-to-market. Their initial GPU-based systems were hampered by legacy storage that couldn’t feed GPUs fast enough, creating a bottleneck that threatened both development speed and model accuracy.

Zenuity deployed NVIDIA DGX-1 servers paired with Pure Storage FlashBlade, selecting FlashBlade after rigorous real-world benchmarks for its bandwidth, scalability and simplicity. The combined platform keeps DGX-1s fully utilized (Zenuity runs two FlashBlade systems with 15×52 TB blades), cutting training times, boosting data scientist productivity, and providing a future-proof, high-performance infrastructure that accelerates model iteration and improves safety outcomes.

Zenuity

Benny Nilsson

Manager of Deep Learning