ZK-rollup projects: a complete guide

Most of the projects that topped ZK-rollup lists a few years ago are no longer running. dYdX left for its own chain. Sorare moved to Solana. zkLend closed down after an exploit drained its lending pools. Polygon switched off its zkEVM entirely in July 2026. The proving technology all of them bet on, though, is in better shape than it has ever been, and the chains running it settle real money every day.

ZK rollups outlasted the projects that pioneered them. A list written today shares almost no names with one written in 2022, at the chain level and at the application level.

What is a ZK rollup?

A ZK rollup executes transactions away from the base chain, then posts a cryptographic proof back to that chain showing the resulting state is correct. The base layer checks the proof instead of re-running every transaction. Because correctness is proven rather than assumed, withdrawals do not need the multi-day challenge window that optimistic rollups rely on.

"Zero-knowledge" is a misleading name for most of these chains. The proofs establish that a batch of transactions was executed correctly, not that the transactions are private. Only a handful of ZK rollups, Aztec being the clearest example, actually hide transaction contents.

The dimension that matters most when you pick one is how closely it matches the Ethereum Virtual Machine. Vitalik Buterin's zkEVM type framework describes that spectrum, from chains where identical contract bytecode runs unchanged to chains with their own virtual machine and their own language. Everything downstream, meaning your tooling, your audit scope, and how much of your existing Solidity you can reuse, follows from where a chain sits on it.

What are the most popular ZK-rollup blockchains?

All figures and stages read from L2BEAT on August 11, 2026. Stage is L2BEAT's decentralization rating, where Stage 0 means a privileged operator can still intervene and Stage 2 means the contracts govern themselves.

Two things in that table would have surprised anyone reading the 2022 version of this page.

The first is that the two largest ZK rollups by value are not zkEVMs competing for general Ethereum activity. Mantle got there by converting, having run as an optimistic rollup before moving to validity proofs on Succinct's SP1 in September 2025, which cut its withdrawal window from seven days to roughly six hours. Lighter got there by being one application, a perpetuals exchange whose order matching and liquidations are proven in custom circuits rather than executed in a general VM.

The second is that the most decentralized ZK rollup is a privacy chain with almost no value on it. Aztec reached Stage 2 in June 2026 after onchain governance revoked ownership of the rollup contract, making the code immutable. Every general-purpose zkEVM in the table is still Stage 0.

Starknet

Starknet uses STARK proofs and its own Cairo virtual machine, so applications are written in Cairo rather than Solidity. That buys prover throughput and makes compute-heavy applications practical, at the cost of a smaller developer pool and thinner cross-chain tooling than EVM chains get for free. It is the only general-purpose ZK rollup at Stage 1, meaning users have a working exit path that does not depend on the operator staying honest.

Its recent growth has been concentrated in Bitcoin-backed DeFi rather than general Ethereum activity, pushed along by a BTCFi incentive program that ran from September 2025 through March 2026.

Linea

Linea is a zkEVM built by Consensys that matches EVM behavior closely enough that most existing Solidity contracts deploy without modification. It uses a gnark-based SNARK proof system, which requires a trusted setup.

Proving and sequencing are both operated by Consensys. Per L2BEAT, only whitelisted proposers can publish state roots, and withdrawals freeze if those proposers stop. Permissionless proving is a stated goal rather than something shipped. If you are deploying existing Ethereum contracts and want the shortest path, Linea is the least friction in the table. Do not describe it to your own users as decentralized yet.

ZKsync Era

ZKsync Era runs EraVM, its own virtual machine, rather than matching EVM bytecode directly. Solidity compiles for it, so source-level porting usually works, but bytecode-level assumptions do not carry over. The tradeoff bought prover performance and a native account abstraction model that makes gas sponsorship and session keys easier than on chains where they are bolted on.

The ZK Stack that powers it is now used by other chains, Abstract among them, which matters more for its long-term position than its own value secured. If you maintain an older integration, note that ZKsync Lite, the original pre-Era deployment, was discontinued in May 2026. Our ZKsync Era overview covers the developer-facing differences in more depth.

Scroll

Scroll targets bytecode-level EVM equivalence and has moved to an OpenVM-based proof system. The equivalence commitment historically cost it prover throughput relative to chains willing to change the VM. Its value secured is well below Linea's and ZKsync Era's, so treat ecosystem depth, not the technology, as the open question when you evaluate it.

Mantle

Mantle launched as an optimistic rollup and moved to ZK validity proofs without asking applications to migrate. Nothing about that was possible in 2022, when a rollup's proof system was a founding decision. Now it is closer to a configuration a chain can change under you, which is a good reason to check what a chain runs today rather than trusting the label it launched with.

Taiko

Taiko Alethia is the closest thing in production to a fully Ethereum-equivalent zkEVM, where the same bytecode runs identically on Ethereum and on the rollup with no adapter layer. Its value secured is small next to the chains above, and we do not currently serve it, so check Taiko's own documentation for its live proving and decentralization status before you commit to it.

What happened to Polygon zkEVM?

Polygon Labs shut down Polygon zkEVM on July 1, 2026. Both mainnet and the Cardona testnet are off, the sequencer no longer orders transactions, and applications deployed there stopped functioning. The shutdown was announced in June 2025, giving roughly a year of notice.

Assets sitting directly in wallets were snapshotted at shutdown and made claimable on Ethereum. Funds locked inside applications on the chain could not be migrated automatically, because once the sequencer stopped, those contracts stopped executing.

The chain traces back to Polygon's acquisition of Hermez Network in 2021, later rebranded and launched as Polygon zkEVM Mainnet Beta in March 2023. Our Polygon ZK rollups overview covers that history, and our Polygon zkEVM deprecation notice documents the endpoints we retired alongside it.

If you are still holding a Polygon zkEVM integration in a codebase, the endpoints are gone and there is no drop-in successor. Pick a target from the table above based on how much of your Solidity you need to keep unchanged.

What are the top ZK-rollup projects?

The 2022 version of this page ranked applications by market cap and user counts. That list has not survived, so it is worth being direct about what happened to it before naming a new one.

The two survivors moved in the same direction. Argent now operates as Ready, a self-custody card and wallet, and Starknet's own ecosystem write-ups list the Ready card as the spending leg of its Bitcoin collateral flow. DeversiFi became rhino, a cross-chain stablecoin liquidity layer. Neither sells a rollup to its users anymore, which is a fair signal of where the durable demand was.

What draws real usage today, grouped by the chain it runs on:

On Starknet

Ekubo is the main DEX and holds most of the wrapped-BTC and stablecoin liquidity. Vesu handles lending and sits at the center of the chain's Bitcoin collateral flows. Extended runs perpetuals across a wide asset list. Endur provides Bitcoin liquid staking, minting a yield-bearing BTC token the other three then accept. That composability, rather than any single application, is what the Starknet BTCFi push actually produced.

On Abstract

Abstract was built by Igloo Inc., the company behind Pudgy Penguins, and launched in January 2025 on the ZK Stack with EigenDA for data availability. Its applications are consumer-facing rather than DeFi, and it has drawn far more transactions than locked value. Its team treats that ratio as the goal rather than a shortfall, so read its transaction counts, not its TVL, if you want to judge whether it is working.

On Lighter

Lighter is both the chain and the application, which is unusual enough to be worth stating plainly. It runs a perpetuals exchange in custom ZK circuits, so order matching and liquidations are cryptographically proven rather than taken on trust. It spent eight months in private beta before opening its mainnet.

Lighter does one thing and still secures more value than Linea, ZKsync Era, and Scroll put together, per the same L2BEAT snapshot above. If you are picking a chain to run a single product rather than a platform, compare it against chains like these instead of against the general-purpose zkEVMs.

On Aztec

Aztec is where you go if the point is confidentiality rather than throughput. Contracts are written in Noir, and privacy is a property of the base protocol instead of something each application implements.

How do you choose a ZK rollup to build on?

Two checks matter more than anything in that table. Read the chain's current L2BEAT stage, because Stage 0 means an operator can still intervene and your users inherit that assumption. Then read who runs proving today rather than what the roadmap promises, since permissionless proving is unshipped on most of these chains.

How does Alchemy support ZK rollups?

We serve RPC for Starknet, Linea, ZKsync Era, Scroll, Mantle, and Abstract. We do not currently serve Taiko, Lighter, or Aztec. Linea, ZKsync Era, Scroll, and Abstract also have Data APIs. Endpoints for the RPC chains are in our supported chains reference, and the ZKsync RPC quickstart is a reasonable place to start if you are porting an Ethereum app.

Serving the RPC calls is the easy half. The harder half is that EVM chains still return balances, transfers, and token data in slightly different shapes, so a naive integration turns into one indexer per chain. Our Data API normalizes that into one surface on Linea, ZKsync Era, Scroll, and Abstract. It does not cover Starknet or Mantle. If your app sponsors user transactions, gas sponsorship works across the EVM chains here too.

You can add any of these chains from the dashboard on the free tier. No contract change, no minimum commitment, and no waitlist for a new chain. If you are weighing us against other providers on ZK-rollup coverage specifically, our node provider comparison lays out the field.

Planning your own chain instead of deploying to an existing one is a different decision, and our ZK rollup and rollup-as-a-service guide covers what running one involves.

Frequently asked questions

What are ZK rollups and how do they work?

A ZK rollup runs transactions off the base chain and posts a validity proof back to it. The base layer verifies the proof rather than re-executing the transactions. Because correctness is proven, withdrawals settle without the multi-day challenge window optimistic rollups require.

Which are the top ZK-rollup blockchains?

Mantle and Lighter currently secure the most value, followed by Starknet, Linea, and ZKsync Era. Rankings shift often, so check L2BEAT rather than trusting a fixed order. On decentralization the order is different. Starknet is the only general-purpose ZK rollup to reach Stage 1, and Aztec is the only ZK rollup of any kind at Stage 2.

What happened to Polygon zkEVM?

Polygon Labs shut it down on July 1, 2026, after announcing the sunset in June 2025. The sequencer is off, applications on the chain stopped working, and wallet balances were snapshotted for claiming on Ethereum. Alchemy retired its Polygon zkEVM endpoints at the same time.

How does ZKsync Era differ from other ZK rollups?

ZKsync Era runs its own EraVM instead of matching EVM bytecode, so Solidity source usually ports but bytecode-level assumptions do not. In exchange it offers native account abstraction, which makes gas sponsorship and session keys simpler than on EVM-equivalent chains.

Are ZK rollup proving systems decentralized?

Mostly not. Every general-purpose zkEVM in this guide sits at L2BEAT Stage 0, meaning a privileged operator can still intervene. Linea restricts state-root publishing to whitelisted proposers. Aztec is the only ZK rollup at Stage 2. Confirm each project's current status before describing it as decentralized.

Which ZK rollup should I use for a consumer app?

Abstract is purpose-built for consumer applications and optimizes for user activity over locked value, using the ZK Stack with EigenDA. If your users need to hold Bitcoin-denominated assets, Starknet's BTCFi ecosystem is the deeper option.