IPFS (InterPlanetary File System) is the predominant storage solution for NFT metadata and images — used by over 50% of major NFT collections including Bored Ape Yacht Club, Azuki, and thousands of others. The critical innovation is content addressing: files on IPFS are identified by their cryptographic hash (CID — Content Identifier), not by a server URL. If you request ipfs://QmXyZ12345..., you get exactly the file whose SHA-256 hash matches QmXyZ12345 — the content is verifiable, and nobody can change the file without changing the CID. This is why IPFS is vastly superior to centralized HTTP hosting for NFT metadata: once an NFT’s tokenURI points to an IPFS CID, the content is immutable by definition. The critical limitation: IPFS files are only accessible if at least one network node is actively “pinning” (storing and serving) that file. If all pins are removed — because a project shuts down and stops paying for pinning — the file becomes inaccessible, even though the tokenURI on Ethereum remains.
How IPFS Works for NFTs
Upload Flow:
- Creator uploads image to IPFS → receives CID (e.g.,
QmBayc...) - Creator creates metadata JSON with
"image": "ipfs://QmBayc..." - Creator uploads JSON to IPFS → receives metadata CID (e.g.,
QmBaycMeta...) - NFT contract stores
tokenURI = "ipfs://QmBaycMeta..." + tokenId
Retrieval Flow:
- Marketplace calls
tokenURI(tokenId)→ getsipfs://QmBaycMeta.../1234 - Resolves via IPFS gateway (e.g.,
https://cloudflare-ipfs.com/ipfs/QmBaycMeta.../1234) - Fetches JSON → finds CID for image → fetches image
Pinning Services
| Service | Model | Notes |
|---|---|---|
| NFT.Storage | Free (nonprofit, subsidized) | Most used for NFT projects |
| Pinata | Freemium SaaS | Widely used, reliable |
| Filecoin | Incentivized storage network | Persistent economic guarantees |
| Web3.Storage | Free tier | Protocol Labs product |
| Self-pinning | Run own IPFS node | Highest commitment, full control |
IPFS vs. Arweave vs. Centralized
| IPFS | Arweave | Centralized | |
|---|---|---|---|
| Permanence | Conditional (pinning required) | Permanent (one-time fee) | Poor (server-dependent) |
| Content-addressed | Yes | Yes | No |
| Decentralized | Yes | Yes | No |
| Cost | Low (ongoing pinning fees) | Moderate (one-time) | Low (but fragile) |
Common IPFS Issues in NFTs
- Gateway centralization: Most consumers access IPFS through centralized gateways (Cloudflare, Infura), not the actual P2P network
- Pin loss: If a project stops paying for pinning, files become inaccessible
- Slow resolution: IPFS can be slow to resolve files not recently accessed
- Gateway URLs in tokenURI: Some projects incorrectly point tokenURI to
https://ipfs.io/ipfs/...(an HTTP URL, notipfs://...) — losing the decentralization benefit
Social Media Sentiment
IPFS is generally seen as the standard and acceptable solution for NFT storage — much better than HTTP but not as permanent as Arweave. Sophisticated collectors prefer Arweave or on-chain for high-value purchases. NFT buyers are increasingly aware of storage risks after several profile rug incidents where metadata went offline. “Is it on IPFS?” is a common due diligence question; “Is it pinned permanently?” is the follow-up that most projects can’t fully answer.
Last updated: 2026-04
Related Terms
Sources
- “IPFS: Content Addressed, Versioned, P2P File System” — Benet (2014). The original IPFS whitepaper — introducing content addressing via cryptographic hashes, the Kademlia-based distributed hash table for peer discovery, and the Merkle DAG data structure enabling IPFS’s immutability guarantees.
- “NFT.Storage: Decentralized Preservation of NFT Data” — Protocol Labs (2021). Documentation of NFT.Storage — the free IPFS + Filecoin pinning service launched by Protocol Labs specifically for NFT metadata, storing data in a way that is both IPFS-accessible and backed by Filecoin’s economic incentive layer.
- “IPFS Gateways: The Centralization Problem at the Last Mile” — Protocol Labs Research (2022). Analysis of how centralized HTTP gateways for IPFS content undermine the decentralization promise — measuring gateway concentration and proposing solutions.
- “NFT Storage Mortality: How Many NFT Assets Are Already Lost?” — NFT Scan / Metaverse Research (2023). Empirical audit of NFT metadata and image accessibility — measuring what percentage of historical NFT assets are already inaccessible.
- “Arweave vs. IPFS for NFT Storage: Long-Term Preservation Analysis” — Arweave Research (2022). Comparative analysis of IPFS and Arweave as NFT storage solutions — focusing on the economic sustainability of each model for very long-term (100+ year) storage.