SkyPilot and Hugging Face have integrated to let users store models and datasets on Hugging Face Hub while running compute workloads on any cloud through SkyPilot. Users can mount a Hugging Face Bucket (read-write) or any model, dataset, or Space repo (read-only) into a SkyPilot task with one hf:// URL and their existing HF_TOKEN, using MOUNT or COPY.
Hugging Face Storage charges no egress or CDN fees, so wherever SkyPilot lands the job, it reads models and datasets straight from the same bucket with no per-cloud copies and no egress bill. Buckets are built on Xet for deduplication, so incremental checkpoints and model variants only store and transfer chunks that changed.
SkyPilot finds compute across 20+ clouds, Kubernetes, Slurm, and on-prem, allowing users to leverage reserved or on-demand GPUs from any vendor. The Hugging Face team upstreamed the hf-mount FUSE fixes that make this work in unprivileged containers.