A Layer 2 blockchain is a network built on top of a base-layer chain like Ethereum that handles transactions offchain and settles them back to the base layer for finality. The point is to scale — to process thousands of transactions per second at fees measured in cents — while inheriting the security of the underlying L1. In 2026, Layer 2s are where most Ethereum transaction activity actually happens, and where most stablecoin volume actually settles.
​

This primer covers what makes something an L2 versus a sidechain, the two dominant families (optimistic rollups and ZK rollups), where Arbitrum, Optimism, Base, zkSync, and Starknet sit, and how to pick one for a real workload. If you want the L1 side first, see what is a Layer 1 blockchain.
​

A Layer 2 inherits its security from a Layer 1. That inheritance is the defining line. On a rollup, every transaction's data and state root are posted back to Ethereum, and Ethereum is the ultimate arbiter of what is final. That inheritance is what separates an L2 from a sidechain — a sidechain has its own validator set and its own security budget, even if it bridges assets to and from Ethereum. Ethereum's Layer 2 documentation uses this inheritance test as the working definition, and L2Beat tracks every major L2 against it.
​

Almost every production L2 in 2026 is a rollup, and almost every rollup is either optimistic or ZK. The division is purely cryptoeconomic.
​

Optimistic rollups assume every batch is valid and rely on fraud proofs plus a challenge window (typically seven days) to catch cheating. Arbitrum, Optimism, and Base are the market leaders. Their strength is near-perfect EVM equivalence — existing Ethereum contracts run with minimal changes — and deep liquidity. Their weakness is the seven-day withdrawal delay, which third-party intent-based bridges have largely solved at the UX layer.
​

ZK rollups prove every batch is valid with a cryptographic validity proof that Ethereum verifies in milliseconds. zkSync Era, Polygon zkEVM, Scroll, Linea, and Starknet are the major players. Their strength is fast finality, no withdrawal delays, and stronger long-term security assumptions. Their historical weakness — harder EVM compatibility and expensive proving — is largely closed in 2026 as zkEVMs mature and prover hardware improves.
​

Not every "Layer 2" in marketing copy is actually a Layer 2 by the security-inheritance test. Polygon PoS is historically a sidechain (its new zkEVM is a real rollup). Gnosis Chain is a sidechain. Validiums — rollups that post proofs to Ethereum but keep transaction data on a separate data availability layer — are a middle category; they trade some security inheritance for cheaper fees. The L2Beat taxonomy categorizes every chain explicitly, and the distinctions matter for anyone evaluating where to hold meaningful value.
​

By TVL, Arbitrum, Optimism, and Base lead the optimistic rollup side. zkSync Era, Starknet, Linea, Scroll, and Polygon zkEVM lead the ZK side. Base has been the fastest-growing chain on either side, driven by Coinbase integration and a deep stablecoin footprint. growthepie's fundamentals dashboard tracks transaction counts, fees paid, and daily active addresses across every major L2, and is the cleanest source for 2026 market share.
​

Pre-2024, L2s paid Ethereum calldata prices for posting transaction data, and that cost dominated their fees. EIP-4844 introduced blobs — a dedicated, cheaper data lane sized for rollup payloads — and cut L2 fees across the board by roughly an order of magnitude. In 2026 every major rollup posts to blobs, and the data-availability side of the cost equation keeps compressing as Ethereum's danksharding roadmap lands PeerDAS and, eventually, full danksharding.
​

An L2's security model has three pieces: the correctness guarantee (fraud proofs for optimistic, validity proofs for ZK), data availability (can anyone reconstruct state), and the exit game (can users escape if the sequencer goes rogue). L2Beat's risk framework scores each L2 on these three, and is the single best reference for "what am I actually trusting here." In 2026 both Arbitrum and Optimism run permissionless fraud proofs; all the major ZK rollups verify validity proofs on Ethereum; and every rollup posts data to Ethereum (blobs or calldata, not offchain).
​

Most L2 economic activity in 2026 is stablecoin activity. USDC is native on Arbitrum, Optimism, Base, Polygon zkEVM, and zkSync Era. USDT has deep liquidity on Arbitrum. Base is the largest single source of new stablecoin onramp flow. For teams that need to move stablecoins between L2s — from Base to Arbitrum, from zkSync to Optimism, from L2 back to L1 — intent-based cross-chain protocols replace slow native bridges, and Eco's execution network routes stablecoin movement across the 15 chains where production liquidity lives, settling in seconds rather than the seven-day wait optimistic rollups impose natively.
​

Three questions usually decide it. First, what is your contract footprint? If you are porting existing Solidity with exotic opcodes, an optimistic rollup (Arbitrum, Optimism, Base) or a Type 2 zkEVM (Scroll, Linea) minimizes surprises. Second, what is your finality requirement? If you need fast withdrawals or strong cryptographic guarantees, a ZK rollup. Third, where is your liquidity? DeFi TVL is heaviest on Arbitrum and Base; stablecoin onramp flow is heaviest on Base; retail activity is spread across all of them. Most teams in 2026 end up deploying on two or three L2s and routing between them.
​

A Layer 2 inherits its security from a Layer 1 — transaction data and state roots post back to the base chain, which is the ultimate arbiter of finality. A sidechain has its own validator set and its own security budget, even if it bridges assets to and from the base chain. Polygon PoS is a sidechain; Arbitrum, Optimism, and zkSync are L2s.
​

L2s batch thousands of transactions into a single Ethereum settlement. You pay for the batch once, not per transaction. After EIP-4844 introduced cheaper blob data, L2 fees dropped roughly 10x overnight. In 2026, simple transfers on major L2s typically cost a few cents.
​

A properly constructed rollup — optimistic with working fraud proofs, or ZK with a verified validity proof — inherits Ethereum's security for transaction correctness and data availability. Remaining trust lives in sequencer centralization, upgrade keys, and the maturity of the fraud-proof or prover system. L2Beat scores each L2 on these axes.
​

Base, Arbitrum, and Optimism carry the deepest stablecoin liquidity among optimistic rollups; zkSync Era and Polygon zkEVM lead on the ZK side. For cross-L2 stablecoin movement, intent-based routing networks settle across chains in seconds rather than the seven-day native withdrawal delay.
​

Native withdrawals go L2 → L1 → L2, which is slow. In practice, intent-based cross-chain protocols let a solver front destination-chain liquidity against your source balance, clearing L2 → L2 transfers in seconds. This is the default UX for stablecoin movement between L2s in 2026.
​

A rollup posts transaction data to Ethereum, so anyone can reconstruct state. A validium posts the validity proof to Ethereum but keeps transaction data on a separate data-availability layer, which is cheaper but relies on that DA layer staying honest. Rollups are the stricter L2 category; validiums sit in a hybrid gray zone.
​