Ethereum Layer 2 Explained: Rollups, Examples and Why They Matter
A layer 2 Ethereum network processes transactions off the main chain and posts data or proofs back to Ethereum, cutting fees to cents while borrowing Ethereum's security.

In this guide
- What Is Ethereum Layer 2?
- Why Layer 2s Exist: The Scaling Problem
- Layer 2 vs Layer 1: Key Differences
- Optimistic vs ZK Rollups
- Data Availability and Blobs: What EIP-4844 Changed
- Bridging to Layer 2 and Withdrawal Times
- Popular Ethereum Layer 2 Examples
- Layer 2 Risks: Sequencers, Bridges and Upgrade Keys
- How to Start Using a Layer 2
- Bottom Line
- Sources
A layer 2 Ethereum network is a separate blockchain network that processes transactions off the main Ethereum chain, then posts compressed data or proofs back to Ethereum so it inherits much of Ethereum’s security. You get the same apps and the same ETH, but fees are often a few cents or less instead of dollars.
Ethereum’s main chain, called layer 1 (L1), deliberately limits how much work each block can hold so that ordinary computers can still verify the chain. That keeps it decentralized but makes block space scarce and, at busy times, expensive. Layer 2 Ethereum networks (L2s) are the answer the Ethereum community has settled on: move execution elsewhere, keep settlement and security on L1.
This guide explains why L2s exist, how the two main kinds of rollups work, what the 2024 “blob” upgrade changed, how bridging and withdrawals work, the best-known networks and the risks you take on when you use one.
What Is Ethereum Layer 2?
A layer 2 is any system built on top of a base blockchain that handles transactions on its own while relying on that base chain for final security. On Ethereum, almost all major L2s today are rollups: they “roll up” hundreds or thousands of transactions into a batch, execute them on their own network, and publish the batch to Ethereum.
Three ideas define a true rollup:
- Execution happens off-chain. The L2 runs the smart contracts and updates balances on its own servers and nodes.
- Data is posted to Ethereum. Enough information to rebuild the L2’s state is published on L1, so anyone can check it or reconstruct balances if the operator disappears.
- Ethereum enforces correctness. A contract on L1 accepts the L2’s new state only if it passes a fraud-proof challenge window or comes with a valid cryptographic proof.
Because most L2s run the Ethereum Virtual Machine (EVM), or something compatible with it, the same wallets, token standards and developer tools work across them. For you, using an L2 usually means switching the network in your crypto wallet and moving funds over a bridge.
Why Layer 2s Exist: The Scaling Problem
Every Ethereum block has a gas limit, a cap on the total computation it can contain. When more people want to transact than fits in a block, they bid against each other and the base fee rises. During busy periods in 2021, simple token swaps on L1 could cost $50 or more. Our guide to Ethereum gas fees explains how that fee market works in detail.
Raising the gas limit dramatically would make blocks bigger and nodes more expensive to run, pushing verification toward a few large operators. That trade-off is often called the blockchain trilemma: it is hard to maximize decentralization, security and scalability at the same time on one chain.
Ethereum’s chosen path is a “rollup-centric roadmap.” L1 focuses on being a secure, neutral settlement layer and a place to store data cheaply. L2s compete on speed, fees and features. One L1 transaction that posts a batch can represent thousands of L2 transactions, so the fixed costs of using Ethereum are shared across many users.
Layer 2 vs Layer 1: Key Differences
Layer 1 is the base blockchain with its own validators and consensus. Layer 2 borrows security from that base chain instead of providing its own. The table sums up how they compare in practice.
| Feature | Ethereum layer 1 | Typical Ethereum layer 2 (rollup) |
|---|---|---|
| Who orders transactions | A large, open set of validators via proof of stake | Usually a single sequencer run by the L2 team |
| Security source | Its own consensus and staked ETH | Ethereum, through posted data and proofs |
| Typical fee for a swap | Cents to dollars, depending on demand | Often fractions of a cent to a few cents |
| Block or confirmation time | 12-second slots | Often under 1 to 2 seconds (soft confirmation) |
| Finality on Ethereum | About 13 minutes for full finality | After the batch is posted and finalized on L1 |
| Main extra risks | Smart contract bugs in apps | Sequencer downtime, upgrade keys, bridge bugs |
A sidechain, by contrast, is a separate blockchain with its own validators that connects to Ethereum through a bridge but does not post enough data or proofs for Ethereum to enforce its correctness. Sidechains can be fast and cheap, but if their validators misbehave, Ethereum cannot help you. That is the key reason rollups are considered the stronger design.
Optimistic vs ZK Rollups
Rollups differ mainly in how they convince Ethereum that their batches are valid.
Optimistic rollups
An optimistic rollup assumes each batch is correct unless someone proves otherwise. After a new state is posted, there is a challenge period, typically about seven days, during which anyone watching the chain can submit a fraud proof (also called a fault proof) showing a specific step was computed wrongly. If the challenge succeeds, the bad state is rejected. Arbitrum, OP Mainnet and Base use this approach.
ZK rollups
A ZK rollup (zero-knowledge rollup) attaches a validity proof to each batch: a compact piece of cryptography that proves every transaction was executed correctly. The Ethereum contract verifies the proof, and once it checks out, the new state is accepted with no waiting period for challenges. zkSync, Starknet, Scroll and Linea are examples. Generating these proofs takes heavy computation, but verifying them on L1 is cheap.
| Aspect | Optimistic rollups | ZK rollups |
|---|---|---|
| How validity is ensured | Fraud proofs during a challenge window | Validity proofs checked on L1 |
| Native withdrawal to L1 | About 7 days | Usually hours, once a proof is verified |
| EVM compatibility | Very high; most code runs unchanged | High and improving; some use their own languages (Starknet uses Cairo) |
| Operating cost | Lower; no proofs needed in the normal case | Higher; proof generation needs significant computing power |
| Security assumption | At least one honest party watches and challenges | The math and proof system are sound and correctly implemented |
| Examples | Arbitrum One, OP Mainnet, Base | zkSync Era, Starknet, Scroll, Linea |
Neither type is strictly better. Optimistic rollups launched earlier and hold more activity; ZK rollups offer faster native withdrawals and a security model that does not depend on someone watching. Several optimistic teams have said they plan to add ZK proofs over time, so the line between the two may blur.
Data Availability and Blobs: What EIP-4844 Changed
Data availability means the information needed to verify or rebuild an L2’s state is published where anyone can download it. Without it, a fraud proof cannot be built and users cannot prove what they own. For rollups, posting this data to Ethereum is the single biggest cost.
Until 2024, rollups posted data as calldata, the input field of a normal transaction, which is stored forever and costs 16 gas per non-zero byte. That kept L2 fees well above where they could be.
The Dencun upgrade in March 2024 activated EIP-4844, also called proto-danksharding. It added a new kind of data container called a blob: about 128 KB of data attached to a block, priced in its own separate fee market and automatically deleted by nodes after roughly 18 days (4,096 epochs). Eighteen days is plenty of time for anyone to download the data and for fraud proofs to run, but nodes do not have to store it forever, so it can be priced far lower than calldata.
The effect on fees was immediate: many L2 transactions dropped from tens of cents to a cent or less. Ethereum has since raised blob capacity. The Pectra upgrade in May 2025 doubled the target number of blobs per block, and the Fusaka upgrade in December 2025 introduced PeerDAS, which lets nodes verify blob data by sampling small pieces instead of downloading all of it, followed by “blob parameter only” forks that lifted capacity further. Blob prices still rise when demand exceeds the target, so check a live tracker if fees matter to you.
Some networks post their data somewhere other than Ethereum, such as a separate data availability committee or chain. These are often called validiums or “optimiums.” They are cheaper but rely on that outside party keeping data available, so they carry more risk than a full rollup.
Example: Suppose a token swap uses 150,000 gas and, hypothetically, ETH is $3,000.
- On L1 at a base fee of 10 gwei (1 gwei = 0.000000001 ETH): 150,000 × 10 gwei = 1,500,000 gwei = 0.0015 ETH = $4.50.
- On a rollup, execution happens on L2 at, say, 0.01 gwei: 150,000 × 0.01 = 1,500 gwei = 0.0000015 ETH, or about $0.0045.
- The rollup must also post your transaction’s data to L1. Compressed, it might be 150 bytes. As calldata at 16 gas per byte and 10 gwei, that is 150 × 16 × 10 = 24,000 gwei, or about $0.07.
- In a blob, if blob gas costs a hypothetical 1 gwei per byte, the same 150 bytes cost 150 gwei, or about $0.00045.
Total on the rollup after EIP-4844: roughly $0.0045 + $0.00045 ≈ $0.005, compared with $4.50 on L1. Real numbers move with demand, but the ratio shows why L2 fees fell so sharply in 2024.
Bridging to Layer 2 and Withdrawal Times
To use an L2, you move assets there through a bridge, a set of smart contracts that locks tokens on one chain and makes an equivalent amount available on another.
Canonical (native) bridges
Each rollup has an official bridge whose contracts on Ethereum hold the deposited funds. Deposits from L1 to L2 usually arrive in minutes. Withdrawals back to L1 are slower because Ethereum must be sure the L2 state is correct:
- Optimistic rollups: you must wait out the challenge period, about seven days, then submit a final transaction on L1 to claim your funds.
- ZK rollups: you wait until the batch containing your withdrawal is proven and verified on L1. Depending on the network’s proving schedule and any built-in safety delay, that ranges from under an hour to about a day.
Third-party and liquidity bridges
Third-party bridges let you skip the wait. A liquidity provider gives you funds on the destination chain almost immediately for a fee, then collects your tokens through the slow route itself. These are convenient, especially for moving directly between two L2s, but you are trusting another set of smart contracts and sometimes a set of operators.
Many centralized exchanges also let you deposit and withdraw directly on popular L2s, which avoids bridging entirely if you are moving funds to or from an exchange account.
Tip: Before you bridge, confirm the exact network name in your wallet, check that the token you are sending is supported on the destination, and keep a little ETH on the L2 to pay gas. Sending tokens to the right address on the wrong network is one of the most common ways people lose funds.
Popular Ethereum Layer 2 Examples
The networks below are among the most widely used. This is a description, not an endorsement; activity, fees and security setups change, and the independent site L2BEAT tracks them in detail.
- Arbitrum One: an optimistic rollup developed by Offchain Labs, live since 2021. It hosts a large share of L2 DeFi activity and is governed by the ARB token holders through the Arbitrum DAO.
- OP Mainnet (Optimism): an optimistic rollup built on the open-source OP Stack. The OP Stack is also used by many other chains, which together form what Optimism calls the Superchain.
- Base: an OP Stack rollup incubated by Coinbase and launched in August 2023. It has no network token of its own and is closely linked to Coinbase’s apps, which has made it a common on-ramp for new users.
- zkSync Era: a ZK rollup from Matter Labs, launched on mainnet in 2023, with EVM compatibility and native account abstraction (smart-contract wallets by default).
- Starknet: a ZK rollup from StarkWare that uses STARK proofs and its own programming language, Cairo, rather than the EVM directly.
Other notable networks include Scroll and Linea (ZK rollups) and Polygon’s various chains. If you are comparing Ethereum’s approach with a single high-throughput chain, see our explainer on Solana and how it compares to Ethereum.
Layer 2 Risks: Sequencers, Bridges and Upgrade Keys
Rollups inherit a lot from Ethereum, but not everything. These are the main risks to understand.
Centralized sequencers
The sequencer is the node that receives L2 transactions, orders them and posts batches to Ethereum. On most major L2s it is run by a single company. A sequencer cannot steal funds in a proper rollup, but it can go offline, halting the network for a time, or delay specific transactions. Good designs include a forced inclusion path that lets you submit a transaction directly through Ethereum if the sequencer ignores you, though it is slower and more technical.
Upgradeable contracts
The L1 contracts that hold bridged funds can usually be upgraded. If a small group controls the upgrade keys with no delay, they could, in principle, change the rules. Many L2s now use a security council (a multi-signature group) and exit windows that give users time to withdraw before an upgrade takes effect. L2BEAT rates rollups in “stages” (Stage 0, 1 and 2) based on how much they still rely on such training wheels.
Bridge hacks and bugs
Bridges hold large pools of locked assets, which makes them attractive targets. Some of the biggest crypto thefts, including the roughly $600 million Ronin bridge hack in 2022, hit bridges. Canonical rollup bridges have so far been far safer than many third-party bridges, but any smart contract can contain bugs.
Proof system and liquidity risk
A bug in a fraud-proof or validity-proof system could let an invalid state through. Separately, a token on an L2 is only as good as the bridge backing it, and thinly traded L2 tokens can be hard to sell at a fair price. Our guide to DeFi and its risks covers smart contract risk more broadly.
How to Start Using a Layer 2
- Pick a network that supports the app you want. Most app websites list the chains they run on.
- Add the network to your wallet. Popular wallets include major L2s by default or let you add them with one click from a trusted list.
- Fund it by withdrawing from an exchange directly to the L2, or by using the official bridge from Ethereum.
- Keep some ETH for gas. Most major L2s use ETH as their fee token.
- Plan your exit. Know whether you will withdraw via the native bridge (and its waiting time), an exchange or a third-party bridge.
Bottom Line
Layer 2 Ethereum networks exist because L1 block space is deliberately scarce, and rollups let thousands of users share the cost of using it. Optimistic rollups rely on a challenge window and fraud proofs; ZK rollups rely on validity proofs and offer faster native withdrawals. Since EIP-4844 introduced blobs in March 2024, fees on most L2s have fallen to cents or less. Before you move meaningful funds, check how decentralized the sequencer and upgrade controls are and how you will withdraw. For the bigger picture, start with our complete Ethereum guide or browse the crypto hub.
Sources
Frequently asked questions
What is Ethereum layer 2 in simple terms?
It is a faster, cheaper network that runs on top of Ethereum. It bundles many transactions together, processes them on its own and then records a compressed summary on Ethereum, which acts as the final judge of who owns what.
What is the difference between layer 2 and layer 1?
Layer 1 is the base blockchain with its own validators and consensus, like Ethereum mainnet. Layer 2 relies on that base chain for security and data, which lets it offer much lower fees and faster confirmations at the cost of some extra trust in its operators.
Optimistic vs ZK rollups: which is safer?
Both can be secure, but they rely on different assumptions. Optimistic rollups need at least one honest party to submit a fraud proof within the challenge window, while ZK rollups depend on the correctness of their cryptographic proof system. In practice, how decentralized the sequencer and upgrade controls are often matters more than the proof type.
Why does withdrawing from an optimistic rollup take 7 days?
The waiting period gives anyone time to challenge an incorrect batch with a fraud proof before Ethereum treats it as final. You can avoid the wait with a third-party bridge or an exchange, but you pay a fee and take on that service's risk.
Do I need ETH to use a layer 2?
On most major Ethereum L2s, including Arbitrum, OP Mainnet, Base and zkSync Era, gas fees are paid in ETH, so you need a small amount of ETH on that network. Some networks and wallets let you pay fees in other tokens, but ETH is the default.
Are layer 2 tokens like ARB or OP needed to use the network?
Usually not. Tokens such as ARB and OP are mainly used for governance, while fees are paid in ETH. Holding them is a separate investment decision with its own risks.