Permanent storage on endowment models, priced honestly

Permanent storage is the most interesting economic design in this whole market - and the most misunderstood. "Pay once, store forever" is an endowment bet: your one-time payment funds a trust whose investment returns pay for replication indefinitely. That is a real, clever design with a real track record - and it is still a bet denominated in a token, not a durability SLA. With only two direct providers in the live population, this page is a state-of-the-market dossier: what the mechanism is, what it costs, and when it is the right answer.

Who this is for, in one paragraph

If you cannot say in one sentence why your data must outlive the company storing it, this is not your product. The honest customer is archiving artifacts whose value outlives institutions: public records, research data, provenance chains, chain-state snapshots. Everyone else is paying for poetry they do not need - the s3-compatible hub is the everyday alternative, and the free-trial hub shows how to test storage platforms without commitment first.

The endowment mechanism

How pay-once models actually work [1][3]:

1. Your upload fee is invested. The payment joins a fund; only the yield is spent on replication.
2. Storage costs fall faster than yield is the design assumption - storage density historically improves faster than most assets yield, and the bet prices that in.
3. Replication quality tracks the fund. No failure date exists, but a funding shortfall degrades replication gradually. The guarantee is economic, not physical.

The honest framing: a subscription commits you to a monthly bill; an endowment commits you to a thesis. The thesis has a decade of history on Arweave - the strongest track record in decentralized storage - and it still deserves the label "bet." 100% of the live rows here carry facts we verified ourselves; on a page this small, each row is load-bearing, and the verification standard matters more than usual.

What it costs, in dollars that move

The fee is denominated in the network token, so your cost is the token price times the rate [1] - and unlike the subscription models in the storage hub, the move happens before you upload, not every month after:

• Arweave prices at roughly 13 AR per GiB in the documented model - at token volatility, the dollar figure moves week to week. Price your upload at both a calm and a stressed token rate before committing [2].
• Irys documents per-GB pricing with a named UK operator - the same mechanism with a younger chain and a documented 2026 migration path for existing integrators [3].
• Autonomi-style pay-once models price via a published formula on the network economics [4] - transparent arithmetic on a young network.

The practical consequence: upload large archives in token-price dips if your timing is flexible, and never treat a quoted dollar figure as fixed until the upload is confirmed.

Live from our database - caveats and sources in each linked review.

The small population is the finding

Only 2 providers in the live population directly sell pay-once storage - and that thinness is itself the market signal. Permanence is hard: it requires a token economy whose endowment actually pays, a network that survives migration cycles, and operators willing to sell a product that never generates recurring revenue. The storage hub carries the broader market; the adjacent pay-prepaid model - Swarm postage stamps [8] - is the closest cousin: prepaid stamps fund storage for a bounded time, renewable, with the same token-denomination exposure.

0 operators from the wider storage vertical sit in our archive. The permanent-storage lesson from that archive: the design that survives is the one where the endowment kept paying - which is why Arweave's decade-long track record [1] is the single most load-bearing fact on this page, and why newer entries with the same design [3][4] are priced as bets, not banks.

What permanence does not cover

• Deletion. Immutable means immutable - including your mistakes. Store hashes, not drafts.
• Updates. New versions are new uploads; the old version stays public forever. That is a feature for provenance and a hazard for credentials.
• Gateway dependence. Reads ride public gateways; the data is permanent, the convenient access path is not. Pin your own gateway if retrieval quality matters [1].
• Migration risk. Networks move - Irys documented a 2026 chain migration with a hard deadline for Arweave-integrated integrators [3]. Permanence is a property of the network that survives, so plan a periodic re-verification of your retrievals.

For data that must survive a specific period, contract-based storage with verifiable renewals is the conservative choice - the storage hub prices that path. For data that must outlive the institutions that hold it - public records, research artifacts, provenance chains - the endowment model is the only serious answer this market has produced, and its decade-long track record is exactly why the storage ranking keeps it near the top. 0 operators from this vertical sit in our archive; the permanent-storage survivors are the ones whose endowments kept paying through every token winter.