Layer 2 (Blockchain Scaling)
Layer 2 (L2) refers to secondary networks built on top of a base blockchain (Layer 1) to increase throughput and reduce transaction fees, while inheriting the security guarantees of the underlying L1.
Ethereum’s L1 processes ~238 TPS theoretical (in practice much less), far below Visa’s ~45,000 TPS. L2 chains address this scalability gap without changing Ethereum’s L1 security model.
How L2s Work
L2 chains:
- Process transactions off the main Ethereum chain
- Periodically post compressed transaction data (or proofs) back to Ethereum L1
- Inherit Ethereum’s security — to revert an L2 transaction, an attacker would need to compromise Ethereum itself
Users bridge assets from L1 to L2, transact cheaply and quickly on L2, and can withdraw back to L1 at any time (subject to withdrawal periods for optimistic rollups).
Main L2 Types
Optimistic Rollups
Assume transactions are valid by default (“optimistic”). Fraud proofs allow anyone to challenge an invalid transaction within a challenge period (typically 7 days). After the period, the state is final.
- Arbitrum — largest L2 by TVL; Arbitrum One + Nova chains
- Optimism — second largest; powers Base (Coinbase’s L2)
- Base — Coinbase’s L2; grew rapidly in 2024
Tradeoff: 7-day withdrawal delay to L1 (mitigated by liquidity bridges).
ZK Rollups (Zero-Knowledge Rollups)
Use zero-knowledge proofs to cryptographically prove the validity of a batch of transactions without revealing transaction details. Near-instant finality — no challenge period needed.
- ZKSync — one of the first ZK rollups; ZKSync Era launched 2023
- StarkNet — uses STARKs (different ZK proof system); enables Cairo smart contract language
- Scroll — EVM-equivalent ZK rollup
- Polygon zkEVM — Polygon’s ZK rollup
Tradeoff: More computationally expensive to generate proofs; historically harder to support full EVM compatibility (improving rapidly).
Key Metrics (2024–2025)
| L2 | Type | Notable Feature |
|---|---|---|
| Arbitrum | Optimistic | Largest TVL; most DeFi activity |
| Base | Optimistic | Coinbase-backed; fastest user growth |
| Optimism | Optimistic | Superchain architecture (OP Stack) |
| ZKSync Era | ZK | First major ZK rollup with EVM support |
| StarkNet | ZK | STARK proofs; native Cairo language |
Major L2s collectively process millions of transactions worth billions of dollars annually.
Dencun Upgrade (March 2024)
EIP-4844 (Proto-Danksharding) introduced “blobs” — a new data storage format for L2 transaction data posted to Ethereum L1:
- Blob data is stored only temporarily (~18 days), not permanently in call data
- Dramatically cheaper than posting to regular call data
- L2 transaction fees dropped 50–90% overnight after Dencun
This was the single most impactful upgrade for L2 cost reduction to date.
L2 vs Ethereum L1
| Factor | Ethereum L1 | L2 (typical) |
|---|---|---|
| Security | Highest | Inherits L1 security |
| Cost | High (gas fees) | Very low (fractions of a cent) |
| Speed | ~12s finality | Near-instant on-chain; seconds |
| Decentralisation | ~8,600+ nodes | Fewer sequencer nodes |
| Ecosystem maturity | Deepest | Growing rapidly |
L2 and Ethereum’s Value
L2 activity increases demand for ETH in two ways:
- L2s post data to Ethereum L1 — this costs gas (ETH)
- L2s use Ethereum’s security — staked ETH is the economic foundation
Bylund (2025) argues L2s are complementary, not competitive: their combined market cap (~$9B) is ~3% of Ethereum’s, and they rely entirely on Ethereum remaining secure.
Sources: wikipedia-2026-ethereum | ethereumorg-2026-what-is-ethereum | bylund-2025-ethereum-investment-thesis