Overview

Starknet is an Ethereum layer-2 built by StarkWare on validity proofs: transactions are executed off-chain and a STARK proof is verified on Ethereum, inheriting L1 security without re-executing every transaction. Contracts are written in Cairo, a language purpose-built for provable computation. Protocol stewardship sits with the non-profit Starknet Foundation and, increasingly, with STRK holders voting through the governance portal.

Starknet belongs to the family of L2 network DAOs this wiki tracks — communities that govern a rollup's contracts and treasury through a token — alongside Arbitrum, the Optimism Collective, ZKsync and Polygon. What sets it apart is that STRK is not a pure governance chip: it also pays gas and secures consensus, so the same asset carries three distinct jobs.

STRK: one token, three jobs

Ten billion STRK were created in May 2022 and minted on-chain on November 30, 2022; supply inflates over time as the protocol mints staking and block rewards on a community-set schedule. The initial allocation splits across early contributors (20.04%), investors (18.17%), StarkWare (10.76%), grants and development partners (12.93%), community provisions and rebates (9.00% each), and the Foundation's strategic reserves and treasury (10.00% and 8.10%). Investor and early-contributor tranches unlock monthly through March 2027.

The token does three things. It is the gas asset — since v0.14.0 (Sept 1, 2025) transaction fees on Starknet are paid only in STRK. It is the staking asset that secures sequencing, consensus, proving and data availability. And it is the governance asset that weights votes over protocol upgrades. Bundling gas, security and voting into one token is a deliberate contrast with ZKsync, whose ZK token votes but never pays gas.

Governance: vSTRK, delegation and Snapshot X

To vote with STRK held on Starknet, a holder wraps it 1:1 into vSTRK, a token dedicated solely to governance. Voting power is the sum of STRK held on Ethereum L1 and vSTRK held on L2, and — as in most large token-DAOs — that power can be delegated to a representative rather than voted directly. Votes run on Snapshot X, the on-chain successor to off-chain Snapshot, so tallies are settled by contract rather than by a signature server.

Starknet's governance opened in phases. In the first phase a professional Builders' Council, seated by the Foundation, held 23.3% of voting power as a competence backstop while the broader delegate base matured. The Foundation has since run an application-based, tiered delegate program through the governance portal: it lends STRK voting power to vetted delegates, and delegates who miss participation standards risk having that power reallocated — a design meant to keep delegated weight active rather than dormant, one of the failure modes catalogued in how DAOs fail.

Progressive decentralization: staking and consensus

Starknet is decentralizing its stack in numbered stages rather than all at once. Staking v1/v2 are live, letting STRK holders stake — directly or by delegating to a validator — with a seven-day unstaking delay; validators currently attest to blocks. Staking v3 (Q4 2025 / Q1 2026) has validators attest and vote on sequenced blocks, and v4 (2026) is slated to fully decentralize both the sequencer and the prover, per the 2025 roadmap recap.

On the sequencing side, the Grinta release (v0.14.0, Sept 1, 2025) moved Starknet to multiple sequencers running Tendermint consensus with a 1/2 threshold — three sequencers operated by StarkWare at launch, with community-run sequencers to follow. Governance and staking are the two levers that hand each of these roles from StarkWare to the token community over time.

Bitcoin as a second security asset

In 2025 Starknet became the first rollup to run a dual-token consensus: alongside STRK, staked Bitcoin can help secure the network, with STRK carrying 75% and BTC 25% of consensus weight (a BTC staking-power weight of 0.25). Within weeks of the launch more than 650 BTC (~$72M) were staked and over 1,000 bridged. Admitting Bitcoin into consensus and validating the bridge design were themselves decided by on-chain governance votes — a concrete case of the token community, not a core team, setting the rollup's security model.

How Caper relates

Starknet's answer to the risk that delegated power goes stale is procedural: seat a competent council early, then lend voting weight to delegates on the condition that they keep showing up, and claw it back when they don't. It is a good design for keeping a large token-holder base engaged. Caper works on a smaller, earlier-stage problem and answers it differently. Instead of curating who holds power, it gives every holder a credible exit right — the ability to leave on defined terms and reclaim a canonical share of the treasury — so the discipline on a bad decision is that participants can walk with their fair slice, not that an underperforming delegate loses a seat. Both are ways to keep governance honest; one manages representatives, the other keeps the door open.