Most teams store their models and datasets in one region of a single cloud, whereas available GPUs for development, training, or serving are increasingly located in a different cloud. This leads to additional costs for transferring data between clouds, which can be expensive.

Hugging Face Blog announces a new solution enabling AI workloads to run on any cloud while storing data on Hugging Face with zero egress costs. This could reduce costs associated with cross-cloud data transfer and simplify working with GPUs located in different cloud environments.