ZKsync (ZK & ZK Nation)

Overview

ZKsync is an Ethereum layer-2 built by Matter Labs on validity (ZK) proofs. ZKsync Era went live on mainnet in March 2023, and the stack has since generalised into the Elastic Network – a set of ZK-powered chains built with the open-source ZK Stack that settle to the same shared bridge. Its governance sits with ZK Nation, the community of ZK token holders and delegates that controls the protocol contracts.

ZK is close to a pure governance asset: transaction fees on ZKsync are paid in ETH, so ZK's standing protocol role is voting weight over upgrades, the token itself, and governance operations. The one exception so far was the ZKnomics staking pilot, which paid newly minted ZK to holders who staked and delegated, and which has been paused since May 2026 (below). That separation – a governance token that is not the gas token – puts ZKsync alongside other L2 network DAOs the wiki tracks, such as Arbitrum, the Optimism Collective, Starknet, Scroll, and GnosisDAO.

The ZK token & distribution

ZK's tokenomics set a ceiling of 21 billion tokens – not a supply that exists, and until December 2025 not one the contract enforced (see the next section) – with roughly two-thirds (66.7%) earmarked for the community and the remainder split between investors and the Matter Labs team under a one-year lock followed by a linear unlock from June 2025 to June 2028. The token generation event and initial airdrop landed in June 2024.

The community airdrop distributed 3.675 billion ZK (17.5% of supply) across 695,232 wallets, based on a snapshot of ZKsync Era and ZKsync Lite activity taken on 24 March 2024 – about 89% weighted to users who transacted and met an activity threshold, and 11% to ecosystem contributors. Details are set out in Matter Labs' airdrop announcement (archived; the post has since been removed from zknation.io) and reported at the time.

Supply: a ceiling the contract enforces, and the part it has never minted

The 21 billion figure quoted in nearly every write-up of ZK is a ceiling, not a supply – and until December 2025 it was not a ceiling the token contract knew anything about. Two keyless reads, both re-runnable, separate the numbers.

• Minted supply. totalSupply() on the ZK token – 0x5A7d6b2F92C77FAD6CCaBd7EE0624E64907Eaf3E, native to ZKsync Era – returned 14,901,043,049.600285 ZK at Era block 71,859,407, read 6 September 2026. That is just under 71% of the ceiling, leaving 6,098,956,950 ZK that has never been minted.
• The ceiling. maxSupply() on the same contract returns exactly 21,000,000,000 ZK.

The gap is not tokens parked somewhere else. ZK's bridged counterpart on Ethereum, 0x66a5cfb2e9c529f14fe6364ad1075df3a649c0a5, held 17,996,312.2 ZK at Ethereum block 25,914,925 the same day – roughly 0.12% of the Era figure, nowhere near six billion. Re-read on 16 October 2026 at Era block 72,513,540, totalSupply() returned 15,436,021,378.7 ZK, about 535m more than on 6 September, which leaves 5,563,978,621 ZK unminted. The unminted remainder is schedule rather than custody: it comes into existence as the unlock curve and governance-approved programmes call for it, each through a minter the governance system has to authorise. A reader who takes 21 billion as the denominator for a holdings or quorum calculation is using a number about a third larger than the one that exists.

The ceiling arrived by vote, and it arrived late. ZIP-14, authored and sponsored by ScopeLift, upgraded the token to ZKTokenV3 and added three things: a public burn any holder may call against their own balance, a role-gated burnFrom, and the maxSupply() function read above – with BURNER_ROLE_ADMIN assigned to the ZKsync Protocol Governor timelock. L2BEAT's governance team published its FOR vote and its review, calling the change minimal, safe and forward-compatible. It executed on Ethereum at block 23,967,898, 8 December 2025 11:50:47 UTC. Before that transaction the ZK contract carried no maximum at all, and “fixed supply of 21 billion” was a claim in a tokenomics document rather than a property of the asset. That both facts are now true on-chain is checkable in one call each, which is the point of stating them this way.

Note also that ZIP-14 travelled as a ZIP through the Protocol Governor, not as a Token Program Proposal, even though its subject was minting and burning rights over ZK – the proposal classes below split on which contracts an upgrade touches, and the boundary is not always where the topic suggests.

The staking pilot, and what it actually measured

Between 9 February and 11 May 2026 ZK Nation ran Season 1 of the ZKnomics staking pilot (TPP-12): holders staked ZK, delegated the voting power that came with it to an active delegate, and were paid newly minted ZK in return. Tally supplied the infrastructure. The stated purpose was to rehearse staking ahead of a decentralised sequencer and to raise active delegated voting power. The programme's own Season 1 review reports the results first-party: a peak of 355m ZK staked against a 400m target (about $7m), 4,397 addresses, 5.3m of an available 10m ZK actually distributed as rewards, and zero incidents. Active delegated voting power went from 958m ZK at the announcement to 1.163bn at the close – net new delegation of 205m ZK. Unminted reward tokens were left unminted and the season's capped minter had its minter roles revoked.

The reply underneath it is the part worth reading. Curia checked the programme's headline metric against a second one, and the two disagree. Roughly 76% of the pilot's delegation reached six addresses; the top three alone received 209.7m ZK, more than the entire net gain of 205m, so removing them leaves the headline roughly flat. And measured over the same five-proposal rolling window the programme's own “active delegate” filter uses, average proposal participation moved from 957.76m to 968.77m – about one per cent. A programme can move delegated power a great deal and move votes actually cast barely at all. That gap is exactly what voting power activation and delegate accountability are about, and it is a useful caution for any DAO planning to buy participation with emissions: the thing rewarded was delegation, and delegation is what grew.

Season 2 is paused. The programme admin announced the pause in the same post, citing alignment with a decentralised sequencer that has slipped to late H2 2026, custodian limitations that kept institutional holders out, and an APR model it wants to reassess before committing to a permanent staking contract. The pilot's capped minters expire on 31 December 2026.

Three-body governance

ZK Nation's defining feature is a three-body structure – a deliberate departure from the single-house token vote used by most protocol DAOs. Each body checks the others (governance 101):

• Token Assembly – ZK holders who delegate voting power to ZKsync addresses. The Assembly submits and votes on proposals to upgrade the protocol, the token, and the governance system itself.
• Security Council – 12 technical experts (auditors, security engineers) who are signers of a security multisig. They review and actively approve protocol upgrades passed by the Token Assembly and, facing an imminent threat, can freeze the protocol and push time-sensitive fixes.
• Guardians – protectors of the ZK Credo, a minimum of 5 of 8 allocated seats, holding veto power to block decisions that clash with the ecosystem's stated values.

Splitting proposing power (Assembly), technical safety (Council), and a values veto (Guardians) is a different answer to the capture and rushed-upgrade risks that a plain token vote carries – closer in spirit to a constitutional separation of powers than to one-token-one-vote.

Onchain governors & proposal types

The three bodies act through three onchain governor contracts on ZKsync Era, each paired with a timelock and each owning one proposal class:

• Protocol Governor executes ZKsync Improvement Proposals (ZIPs) – upgrades to the protocol contracts.
• Token Governor executes Token Program Proposals (TPPs) – proposals that assign minting and burning rights over ZK.
• GovOps Governor facilitates Governance Advisory Proposals (GAPs) – governance and operations decisions.

Standard upgrades follow a set path: Token Assembly approval → Security Council risk review → timelocked execution. Emergency changes route instead through the Emergency Upgrade Board, which accepts approvals from all three governance bodies before an expedited upgrade can execute.

Proposals to simplify governance (October 2026)

Two draft proposals posted in October 2026 would change the structure described above. Neither had gone to a vote when read on 16 October 2026.

• Dissolving the Guardians. A draft posted by the Guardians on 7 October would move their veto and fallback-approval roles on all three governors to the ZKsync Foundation, and replace the Emergency Upgrade Board with one that needs approval from the Security Council and the Foundation only. It cites the needs of Prividium operators and other institutional counterparties. A delegate who replayed the pending ZIP-18 upgrade on a fork reported that the Token and GovOps governors hard-code today's Guardians as their L2 veto holder; in reply, rafa, who posted the draft below, said engineering had confirmed the dissolution would execute after ZIP-18, the resubmitted v31 upgrade.
• Optimistic governance for ZIPs. A draft from the ZKsync Association on 12 October would let Security Council members propose protocol upgrades that proceed unless the Token Assembly vetoes them, the optimistic-governance pattern. Today every ZIP needs at least 630m ZK voting for it and a simple majority of votes cast. Under the draft, 630m ZK voting against would stop a ZIP, and a veto could be overridden only with at least 1.26bn ZK for and more votes for than against.

BORGs: smart contracts wrapping legal entities

The Security Council and Guardians are structured as BORGs (blockchain organizations) – legal entities whose behaviour is constrained by onchain rules embedded in their charters, so a multisig cannot simply act outside its mandate. It is the same "cybernetic org" pattern that Safe multisigs and adjacent tooling have popularised, applied here to give a values veto and a security freeze real legal and onchain teeth rather than social convention alone.

How Caper relates

ZKsync's Guardians are a backstop against capture: a minimum of five of eight allocated seats can veto a decision that clashes with the ecosystem's stated values. Until 29 August 2026 Caper had no equivalent, and this section said so – that its answer to a bad decision was the exit right rather than a veto seat. That contrast no longer holds, because a caper has since acquired a veto of its own.

A caper proposal now clears two gates. The ballot is legislative: it decides which typed option carries a mandate. Triggering that option then locks the caper's trailing TWAP as a baseline and opens the market window, and the action executes if and only if the TWAP measured across that window is at or above the baseline. That is a veto in the standard optimistic-governance sense – but the seat is not appointed, because there is no seat. Anyone holding the token can exercise it, and exercising it means selling hard enough to hold the price under the baseline for a real share of the window. Holding is consent.

Set beside ZK Nation's three bodies, the difference is who pays for saying no. A Guardian veto is free to the Guardian, and its legitimacy rests on the charter it is judged against. A caper's veto costs the vetoer whatever they sell into, and its legitimacy rests on nothing but the price that results. Neither is obviously the better answer: a values veto can block a decision no market would have punished, and a price veto can be outbid by anyone willing to buy the token back up. The exit right is still there underneath both – but in a caper it is now the last line of defence rather than the first.