Filecoin and Arweave are the two biggest names here. Filecoin pays providers per-epoch to store data they commit to holding; Arweave takes a one-time endowment payment and guarantees permanent storage. Both models have real usage — Filecoin has petabytes of committed storage, Arweave hosts permanent copies of NFT metadata for most of the major collections — but neither has displaced AWS.
For the broader compute and storage narrative see the DePIN sector.
Two fundamentally different approaches
Filecoin (FIL) uses a “proof of replication” and “proof of spacetime” model. Storage providers commit disk capacity, take user data, and periodically prove they still hold it. Users pay per-epoch for the duration they want the data stored. If a provider fails the proof, they get slashed. The model is closer to decentralised AWS — you rent storage for a time window, renew or migrate before expiry.
Arweave (AR) uses a “pay once, store forever” endowment model. Users pay an upfront fee calculated to fund storage for 200+ years under conservative cost-decline assumptions. The endowment grows through inflation and transaction fees. Once data is on Arweave, it’s there permanently (or at least as long as the network operates). That makes Arweave the default choice for NFT metadata, research archives, and anything where durability matters more than cost.
Other storage projects worth knowing
Sia (SC) and Storj (STORJ) predate Filecoin and Arweave and operate at smaller scale. Both use distributed encrypted sharding — files split across many providers, reconstructed client-side. Sia’s token economics revolve around host collateral and contract pricing; Storj operates more like a drop-in S3 replacement with encrypted client-side storage.
BNB Greenfield is BNB Chain’s storage layer. Effectively a centrally-governed competitor to Filecoin, with deeper integration with BNB Chain applications but less of the decentralisation story.
Walrus (from Mysten Labs, the Sui team) launched in 2024 as a Sui-native decentralised storage network, targeting low-latency, high-throughput use cases. It’s the newest entrant but has generated meaningful attention given Sui’s backing.
The fundamental thesis (and where it hasn’t played out)
The long-running storage thesis is that decentralised networks can undercut AWS S3 on price while matching durability and accessibility. In practice:
- Price. Filecoin and Arweave are genuinely cheaper per TB than AWS S3 for archival workloads. Not for hot-storage workloads with high retrieval latency requirements.
- Durability. Both networks have survived their launch periods without major data loss events. Arweave’s endowment model is harder to stress-test because the real test is 50+ year durability.
- Accessibility. This is the weak point. Retrieval latency on Filecoin and Arweave remains higher than AWS; pinning services (Pinata, Fleek) exist to smooth the UX gap but add a trust layer.
The bear case: enterprise storage buyers optimise for SLAs, integration, and compliance — not cost per TB. Decentralised storage excels on cost but is still behind on the other dimensions. That’s why FIL and AR have real users without capturing mainstream enterprise share.
Data below is live from CoinGecko.