Filecoin Cold Storage Service (FCSS)

Archive large datasets at low cost, sealed and proven with Proof of Replication (PoRep) across a decentralized network of storage providers. Need fast, active access instead? See Filecoin Warm Storage.

  • PoRep-Sealed Durability

    Data is sealed at write time and proven onchain with Proof of Replication (PoRep) — built for maximum durability over years, not fast reads.

  • Built for Scale

    Designed for datasets of 1TiB and up, where cost per TiB matters more than retrieval speed.

  • Marketplace Pricing

    Storage providers compete on price and durability in an open marketplace — no single vendor controls your archive.

Who it's for

FCSS is built for datasets of 1TiB and up, where cost per TiB and durability matter more than retrieval speed.

  • AI/ML data teams

    Training datasets, model checkpoints, and inference logs pile up fast and are read rarely. Cost per TiB at scale matters more than any single retrieval time.

  • Media, research & archival institutions

    Media libraries, scientific and genomic repositories, and public archives with long-retention, latency-tolerant datasets that need durability over decades.

FCSS vs FWSS

Both are part of the same Filecoin Onchain Cloud marketplace and settle through Filecoin Pay.

Comparison of Filecoin Warm Storage and Filecoin Cold Storage
DimensionFilecoin Warm Storage (FWSS)Filecoin Cold Storage (FCSS)
Proof mechanismProof of Data Possession (PDP) — continuous, lightweight proofsProof of Replication (PoRep) — sealed at write time, onchain-verified
Access patternHot — fast retrieval for actively-used dataCold — archival, for infrequently-accessed data
Typical dataset sizeAny size, optimized for frequent reads1 TiB and up
Best forApp data, media serving, agent workloadsAI/ML training data, model checkpoints, media and research archives
Payment railFilecoin PayFilecoin Pay

Frequently asked questions

Ready to archive your data?

Cold Storage deals are set up directly with the FCSS team, where you can request a quote based on your dataset size and retention needs.