Nvidia Unveils Open-Sourced cuFile APIs and Industry Initiative for Improved AI Storage Access

Nvidia is open-sourcing its cuFile APIs and software stack to speed access to AI storage and memory systems, promoting industry cooperation in GPU-driven storage advancements.

The company's goal is to foster broader adoption and interoperability across the AI storage industry by making cuFile a more open foundation for AI-native storage. This move aims to increase storage throughput, reduce latency, improve GPU utilization, and speed up AI training and inference workloads.

Nvidia will also lead a new memory and storage industry initiative called Storage-Next, which brings together over 40 vendors to align on how GPU-driven storage should behave and turn advancements into interoperable, open industry standards. The company's SCADA framework allows massively parallel GPUs to pull only the necessary data directly from storage into their high-speed memory.