Layer 2s (L2) are blockchains built on top of Ethereum that process transactions quickly and cheaply while inheriting its security. They're Ethereum's answer to the scaling problem: instead of making L1 faster (which would compromise decentralization), it delegates activity to specialized L2s.
In 2026 L2s process over 90% of Ethereum volume. If you use DeFi, NFTs, or any modern dApp, you're almost certainly operating on an L2 without realizing it.
Why do Layer 2s exist?
Ethereum L1 is limited by design: ~15-25 transactions per second and variable fees from $0.30 to $50+ during congestion. Increasing those numbers directly would require:
- Bigger blocks (compromises decentralization: only large nodes could validate)
- Shorter block time (compromises security)
- Reducing validator requirements (compromises trust)
Instead, Ethereum adopted the rollup-centric strategy: L1 maintains its role as a secure decentralized settlement layer, and execution is delegated to L2s that process transactions off-chain and publish the result to L1.
The result: users see fees of $0.01-$0.20 with fast transactions, while maintaining Ethereum's economic security.
How rollups work
Two main types of rollups:
Optimistic Rollups
- How it works: execute transactions off-chain and publish results to L1 assuming correctness ("optimistic"). Anyone has 7 days to submit a "fraud proof" if they detect an error.
- Pros: full EVM compatibility, more mature.
- Cons: native L1 withdrawals take 7 days (challenge window).
- Examples: Arbitrum, Optimism, Base.
ZK Rollups
- How it works: execute transactions off-chain and publish results to L1 along with a cryptographic proof that the execution is correct. L1 verifies the proof.
- Pros: fast L1 withdrawals (no challenge window), stronger cryptographic security.
- Cons: more limited EVM compatibility (improving fast), more complex.
- Examples: zkSync, Starknet, Linea, Scroll, Polygon zkEVM.
The main Layer 2s in 2026
Arbitrum
Leading DeFi TVL among L2s. Operated by Offchain Labs. Features:
- Nitro client with WASM as intermediate environment.
- Stylus: enables smart contracts in Rust, C, C++.
- Arbitrum Orbit: framework for other projects to launch L3/L2.
- Token: ARB (DAO governance).
- Key ecosystem: GMX, Camelot, Pendle, Radiant.
More detail in our Arbitrum vs Optimism comparison.
Optimism
Operated by OP Labs. Main thesis: OP Stack as an open-source modular framework that other projects use to launch their interoperable L2s (the "Superchain").
- Bedrock client + OP Stack.
- Superchain: Base, World Chain, Zora, Mode, Cyber use OP Stack.
- Token: OP (governance).
- RetroPGF model: retroactive distribution to contributors.
- Ecosystem: Velodrome, Synthetix, Sonne Finance.
Base
Operated by Coinbase, built on OP Stack. Features:
- Fees typically lower than Arbitrum/Optimism.
- Excellent UX for new users (Coinbase wallet integration).
- No native token (unlike ARB and OP).
- Explosive growth since 2024 — one of the most active L2s.
Base is probably the best option for users coming from Coinbase, given the direct funnel of fiat → Coinbase → Base.
zkSync Era
ZK rollup operated by Matter Labs. Features:
- EVM-compatible (not identical, minor differences).
- Fast L1 withdrawals.
- ZK token launched in 2024.
- Good DeFi ecosystem (Velocore, Maverick, SyncSwap).
Starknet
ZK rollup with its own VM (not EVM). Cairo language.
- Pros: most cryptographically optimized.
- Cons: have to learn Cairo, smaller ecosystem.
- Token: STRK (launched 2024).
- Ecosystem: Ekubo, JediSwap, Nostra.
Linea, Scroll, Polygon zkEVM
ZK rollups maintaining full EVM compatibility. Newer but growing.
Quick comparison
| L2 | Type | Token | DeFi TVL | Typical fee | Ecosystem |
|---|---|---|---|---|---|
| Arbitrum | Optimistic | ARB | $3.5B | $0.02-$0.20 | Deep DeFi |
| Optimism | Optimistic | OP | $1B | $0.03-$0.30 | Superchain |
| Base | Optimistic (OP Stack) | — | $4B+ | $0.01-$0.15 | Consumer + Coinbase |
| zkSync | ZK | ZK | $0.5B | $0.05-$0.25 | Emerging DeFi |
| Starknet | ZK | STRK | $0.3B | $0.05-$0.20 | Native Cairo |
How to use a Layer 2
- Get ETH on CEX (or stablecoins): Binance, Coinbase, Bybit, OKX.
- Withdraw directly to an L2 if your exchange supports it. Many CEX already support native withdrawals to Arbitrum, Optimism, Base, zkSync without going through L1 (saves bridge fees).
- If not, bridge from L1:
- Native bridge: the official one for the L2 (e.g., bridge.arbitrum.io).
- Third-party bridges: Across, Hop, Stargate. Faster, charge a fee.
- Configure the network in your wallet: add the L2 in MetaMask, Rabby, Frame.
- Operate normally: once on the L2, you use dApps as if it were Ethereum, but with near-zero fees.
Risks to know
1. Centralized sequencer. Today all major L2s have a sequencer operated by the company (Offchain Labs on Arbitrum, OP Labs on Optimism, Coinbase on Base). If the sequencer goes down or censors, there are "force inclusion" mechanisms but they're technical.
2. Smart contract risk. The rollup contracts and the DeFi protocols you use carry exploit risk. New L2s have less audit history than L1 protocols.
3. Stage 1 vs Stage 2. The L2Beat classification indicates maturity. Stage 0 = fully centralized. Stage 2 = fully decentralized. Almost all current L2s are at Stage 1 — operators can theoretically revert transactions under certain conditions.
4. Withdrawal periods. Optimistic rollups take 7 days to withdraw to L1 (unless you use third-party bridges that advance liquidity for a fee).
Which L2 to choose?
For deep DeFi: Arbitrum (highest TVL, GMX, Pendle). Coming from Coinbase or want minimum fees: Base. Value Superchain and RetroPGF: Optimism. Need fast L1 withdrawals: zkSync, Starknet. Need full EVM compatibility: Arbitrum, Optimism, Base, zkSync.
You don't have to choose only one. Many users keep funds on several L2s depending on the protocols they use.
The future of L2s
Clear roadmap for 2026-2028:
- Greater sequencer decentralization (Stage 2 target).
- Interoperability between L2s (Superchain, Polygon Aggregation Layer).
- Dedicated data availability (Celestia, EigenDA) to lower fees further. More on this in our Modular Blockchains guide.
- Native Account Abstraction for better UX.
- ZK rollups dominant: as EVM compatibility improves, ZK rollups will gain share.
FAQ
Are L2s as secure as Ethereum? They inherit Ethereum's economic security (data is on L1) but have additional smart contract and sequencer centralization risk. For small-to-medium holdings: yes, secure enough. For corporate treasury millions: probably prefer L1.
Can I move funds directly between L2s? Yes, with bridges like Across, Hop, Stargate. There's no native direct bridge between L2s — each has only one bridge with L1.
Why so many L2s if they're all Ethereum? Each optimizes for different cases (consumer, deep DeFi, ZK, modularity). The market will differentiate them over time: probably 3-5 large L2s and many small specialized ones will coexist.
Do I need ETH on each L2 for fees? Yes, but minimal amounts ($1-5 usually enough). Some L2s allow paying fees with stablecoins (account abstraction).
Do L2s issue their own token? Some yes (ARB, OP, ZK, STRK), others no (Base, Linea). The token is usually governance, doesn't capture fees directly.
Conclusion
Layer 2s are already where most of Ethereum happens today. They're not an alternative, they're the new default layer. If you do more than passive holding, practically any activity —swaps, lending, perps, NFTs— happens on L2.
Knowing the differences between Arbitrum, Optimism, Base, and ZK rollups is part of the basic kit of a 2026 crypto user. And as with everything in this space: start with small amounts, get familiar with bridges and wallets, and scale up when you understand the flow.
