CoinYQ Dossier

A Two-Century Prepayment Asked to Carry the Word “Forever”

Arweave turned a recurring cloud bill into an upfront economic wager. Its miners do not merely hash the newest block; they repeatedly prove access to old chunks packed across disks. The design is concrete, but permanence is a system outcome assembled from prices, replicas, node software and gateways—not a magic property stamped onto an AR payment.

Sam Williams tries to make forgetting more expensive than remembering

In 2017 Sam Williams and early contributors started from a familiar weakness of the web: a file usually survives only while someone keeps paying a host to serve it. Their answer was to make old data part of the mining competition itself, so the network would reward machines for proving that they still held its history.

That decision changed the unit being sold. Instead of renting disk space month by month, an uploader would pay for an estimated long future when data entered the weave. Arweave carried the idea into a live mainnet on 8 June 2018, turning each upload into part of a public ledger that later miners would have to remember.

The recall chunk makes yesterday part of today’s mining race

The early paper called the structure a blockweave: each new block belonged to a history that could challenge access to a past recall block. Current SPoRA documentation is more exact. A candidate block carries one or two pseudorandom 256 KiB recall chunks and cryptographic proofs tied to the accumulated block index.

Modern miners divide the weave into 3.6 TB partitions and pack each replica with entropy tied to their address. A VDF limits the rate at which candidates can be tried. More broadly and uniquely replicated data creates more useful opportunities than endlessly copying one easy slice.

That machinery has evolved. Replica 2.9, VDF changes and mining rules arrived through software and hard-fork releases. The reference implementation is open source, but AR holders do not cast a binding protocol-upgrade vote; maintainers propose and ship code, and node operators decide what chain-compatible version to run. The Digital History Association now describes itself as stewarding core research and the primary implementation; that influence still does not let it rewrite every independently operated node by an administrator key.

The endowment is a formula, not a trust fund manager

An upload fee has two time horizons. One part can reward the miner that includes the transaction; the rest enters protocol endowment accounting. Mining rewards combine released transaction fees, declining issuance and an endowment draw only when the model says the other components do not cover the storage burden.

Current documentation says the price prepays 20 replicas for 200 years at present costs. If storage’s all-in cost falls at least 0.5% per year and AR purchasing power does not undermine the calculation, the endowment’s modeled storage power can persist indefinitely. “Forever” is the conclusion of those assumptions, not a guaranteed fiat return earned in an external account.

A permanent ledger still needs someone to answer the door

Arweave’s core node stores, validates and proves the weave. It does not provide CDN-scale delivery, search, initial data seeding or large-upload bundling. Gateways, bundlers and other services perform those jobs. The protocol can preserve a transaction while a particular gateway fails to index or serve it.

Content policy is similarly layered. Node software supports local blacklists, and miners choose what data to store and serve. That protects operator autonomy and legal compliance, but it means “uncensorable” cannot honestly mean every participant must deliver every byte. Permanence concerns the replicated record; reachability is an ongoing network property.

On 5 September 2026 the arweave.net node reported arweave.N.1 at height 1,994,116 and 208 peers known to that gateway. It is direct evidence of a live network view, not a census of independent miners or proof that every historical item had equal replication.

AO borrows the archive without merging the two assets

AO extends the ecosystem from storage to parallel processes whose messages can be recorded on Arweave. Its protocol and 21 million AO token are separate. AO emissions rewarded AR holders and bridged assets, creating an economic link without replacing AR; it does not erase the separate scheduled-supply and debt-supply rules.

The expansion did not replace the original bargain. Foundations and developer groups could build new software, while gateways delivered data and miners kept proving access to it. AO made the archive useful to a new class of programs, yet the AR network still rose or fell on whether independent operators found it worthwhile to preserve and serve the weave.

The precise legacy is therefore narrower and stronger than the slogan. Arweave built a live market that repeatedly pays machines to remember. Whether a particular file remains easily reachable for generations still depends on the very things the slogan compresses: correct pricing, enough replicas, compatible upgrades and willing delivery infrastructure.

How the project changed

  1. 2017
    The storage protocol is designed

    Sam Williams and early contributors develop a ledger whose mining rewards depend on access to history.

  2. 2018-06-08
    Arweave mainnet launches

    The genesis block creates 55 million AR and begins the declining 11 million mining schedule.

  3. 2021-02
    SPoRA replaces the earlier access race

    Mining shifts toward succinct random-access proofs intended to reward broad replication.

  4. 2024-02-27
    AO emissions begin

    A separate compute token starts rewarding AR holders while using Arweave as its data foundation.

  5. 2025-02-08
    AO becomes transferable at mainnet

    Compute expands the ecosystem without becoming a new AR issuance mechanism.

Evidence and primary sources

Last evidence review: 2026-09-05

What is Arweave?

Arweave is a proof-of-work storage blockchain whose native AR pays for data inclusion, value transfers and miner rewards. Its ledger combines blocks with commitments to the accumulated dataset, while SPoRA mining asks for randomly selected historical chunks from miner-specific packed replicas. The permaweb is the application and retrieval ecosystem built above that base layer.

The familiar phrase “pay once, store forever” describes an incentive model, not a legal warranty. Upload pricing prepays 20 replicas for an estimated 200 years and places part of the fee in a protocol endowment. Long-term sufficiency depends on storage-cost decline, AR purchasing power, replication, miner participation and availability through gateways.

What problem does Arweave solve?

Ordinary web data survives only while a host keeps paying and serving it. Arweave changes the payment shape: the uploader contributes upfront, and miners compete for rewards by proving access to old data as well as accepting new blocks. The network tries to make preservation the profitable default.

Storage and access remain different jobs. A core node validates and stores the weave; gateways index and deliver it, bundlers aggregate uploads, and clients must seed new data. Individual operators may filter material. A confirmed transaction is therefore stronger than a promise by one cloud vendor, but it does not mean every URL will be served by every participant forever.

How does Arweave work?

A transaction may transfer AR, carry data, or do both. One AR is 10^12 winstons. Miners pack 3.6 TB partitions into address-specific replicas, use SPoRA to prove access to pseudorandom 256 KiB recall chunks, and combine that work with a verifiable delay function. The heaviest valid fork becomes the shared history.

Rewards come from immediately released fees, declining inflation and conditional endowment releases. The yellow paper created 55 million AR at genesis and schedules 11 million more through mining, toward roughly 66 million; current code uses 66,000,015.859279336957 and permits debt supply if a depleted endowment requires it. AO adds computation above Arweave and separately rewards AR holders with AO; it neither expands AR supply nor turns AR into AO governance.

Key facts

  • Mainnet launch: 8 June 2018; genesis supply: 55,000,000 AR; scheduled mining issuance: 11,000,000 AR; current scheduled-supply constant: 66,000,015.859279336957 AR; future debt supply is excluded.
  • One AR equals 1,000,000,000,000 winstons.
  • Arweave blocks commit to the accumulated block index and dataset rather than storing files in a single conventional chain field.
  • Current SPoRA proofs sample one or two 256 KiB recall chunks.
  • Replica 2.9 packs 3.6 TB partitions uniquely to a mining address.
  • A VDF limits how quickly recall candidates can be tested, keeping storage capacity relevant to mining.
  • Upload fees split between immediate miner reward and protocol endowment accounting.
  • The 20-replica, 200-year prepayment and 0.5% Kryder+ threshold are model inputs, not a fiat-value guarantee.
  • Nodes can use local blacklists; gateways and miners do not have to serve every item.
  • At review, arweave.net reported height 1,994,116 and 208 peers known to that node.
  • Protocol upgrades require software releases and adoption by operators; AR has no documented binding upgrade vote.
  • AO is a separate compute protocol and token whose message history uses Arweave.
  • AR ownership grants transfer and fee-payment capability, not shares, dividends, content ownership or redemption.
  • At review the protocol reward pool held 363,177.388909610291 AR and debt_supply was zero; code-defined available supply was 65,636,838.470369726666 AR, a different measure from exchange circulation.
  • The Digital History Association stewards protocol research and the primary implementation, but no administrator key can replace the software chosen by every node.

Official links

Categories

Related coins

Frequently asked questions

Does Arweave legally guarantee that a file survives forever?

No legal warranty was found. The protocol prices 200 years of replicated storage and models indefinite purchasing power under cost-decline and token assumptions. Actual retrieval also depends on replicas, seeding and gateways.

What is the blockweave?

It is the historical description of Arweave’s block structure and merkelized index, which lets new blocks commit to current history while mining challenges access to old data. Current docs often call the accumulated dataset simply the weave.

What does SPoRA prove?

A miner shows access to randomly selected historical chunks from a replica packed to that miner’s address. A VDF limits attempts, encouraging broad storage rather than only fast computation.

Is the endowment an investment account earning interest?

No. It is protocol accounting that withholds part of upload fees and releases AR when fees and inflation are insufficient under the reward formula. Its adequacy depends on modeled storage economics.

Can miners delete or censor data?

They cannot rewrite a confirmed transaction without defeating consensus, but each miner may choose what it stores or serves and nodes support local blacklists. Availability and universal delivery are separate from ledger inclusion.

Who approves Arweave upgrades?

Maintainers publish code and releases; miners and other node operators choose compatible software and the network follows valid consensus. No binding token-holder ballot was found.

Is AO the next version of AR?

No. AO is a separate compute protocol and has its own token. It records messages on Arweave and its distribution rewarded AR holders, but the two assets and governance surfaces remain distinct.

What rights does AR provide?

AR can be held, transferred and spent for protocol fees. Reviewed sources do not grant company equity, dividends, guaranteed storage value, redemption or ownership of data uploaded by others.

External trackers

Choose a tracking site for Arweave: