The Problem Ethereum Had to Solve
Ethereum processes approximately 15-20 transactions per second on its mainnet. Visa processes approximately 24,000 TPS at peak. This gap has been Ethereum's central challenge since it became the hub for decentralized finance and Web3 applications. When demand exceeded capacity — as it did repeatedly in 2020, 2021, and 2022 — gas fees surged to levels that made the network unusable for anyone with modest holdings. Simple token transfers cost $50-200. Complex DeFi operations cost $500-2,000.
These fees priced out the majority of global users and created an opportunity for rival blockchains like Solana, BNB Chain, and Avalanche to capture users and developers who couldn't afford Ethereum's mainnet. The scalability problem was existential: without a solution, Ethereum risked losing its ecosystem advantage to cheaper, faster competitors.
The answer the Ethereum community converged on was not to make mainnet faster — that would require compromising decentralization, which the community was unwilling to do. Instead, the solution was to build a Layer 2 ecosystem: separate chains that process transactions off-chain and then batch-commit them to Ethereum mainnet in compressed form. This is the rollup-centric roadmap, and in 2026 it is delivering results that would have seemed impossible two years ago.
The Glamsterdam Hard Fork: The Upgrade That Multiplied L2 Capacity
The path to 4,800 TPS started with EIP-4844 (Proto-Danksharding) in March 2024. This upgrade introduced "blobs" — a dedicated data channel for L2s to post transaction batches to Ethereum mainnet at dramatically lower cost than using standard calldata. Before EIP-4844, L2 data posting cost $0.10-0.50 per transaction; after, it dropped to approximately $0.001-0.01. This single upgrade reduced average L2 fees by 90% overnight and accelerated L2 adoption substantially.
Glamsterdam, in early 2026, extended this breakthrough by increasing blob capacity approximately 10x. Where EIP-4844 created the blob channel, Glamsterdam widened it dramatically — reducing L2 data costs another 70-80% and removing the bottleneck that was still limiting how many transactions L2s could process per Ethereum block.
The practical impact was immediate. Base went from processing approximately 20 TPS to over 300 TPS in the weeks after Glamsterdam. Arbitrum One scaled from 40 TPS to 500+ TPS. Combined across the top 10 L2 networks, Ethereum's aggregate throughput crossed 4,800 TPS — a 24x improvement from the 200 TPS measured before EIP-4844. Average transaction fees on major L2s now average $0.001-0.01, making Ethereum accessible to users globally regardless of transaction size.
⚡ Ethereum Layer 2 Comparison — May 2026
| L2 Network | Type | TVL | Avg Fee | Finality | Best Use Case |
|---|---|---|---|---|---|
| Arbitrum One | Optimistic | ~$17B | $0.005 | ~1 min | DeFi, derivatives, DEX |
| Base | Optimistic | ~$9B | $0.003 | ~2 sec | Consumer apps, social |
| Optimism | Optimistic | ~$7B | $0.005 | ~2 sec | DeFi, OP Stack chains |
| zkSync Era | ZK Rollup | ~$4B | $0.01 | Instant | Payments, security-critical apps |
| Starknet | ZK Rollup | ~$2.5B | $0.01 | Instant | Gaming, complex computation |
TVL estimates Q2 2026. Fees are approximate averages for simple token transfers post-Glamsterdam.
Arbitrum: Ethereum's DeFi Capital
Arbitrum One maintains its position as the largest Ethereum L2 by TVL at approximately $17 billion. It hosts the majority of Ethereum DeFi activity that has migrated off mainnet — leading DEXs like Camelot and Trader Joe, lending platforms including Radiant Capital, and the perpetuals ecosystem anchored by GMX and its successors.
Arbitrum's dominance stems from early launch timing (giving protocols time to mature), full EVM compatibility (any Ethereum dApp deploys with minimal changes), and a strong developer ecosystem supported by the ARB governance DAO — one of the largest by treasury value. Arbitrum Nova, a separate chain optimized for high-volume, low-value transactions, processes millions of gaming and social application transactions daily at near-zero fees.
Base: Coinbase's Path to Mass Adoption
Base, launched by Coinbase in August 2023 and built on Optimism's OP Stack, has emerged as the L2 with the clearest path to mainstream consumer adoption. With Coinbase's 110 million+ verified users as a direct acquisition funnel, Base offers what no other L2 has: tight integration with the most widely used US crypto on-ramp.
Base has become home to "onchain consumer apps" — decentralized social platforms, creator monetization tools, and consumer-facing applications that benefit from blockchain's ownership properties without mainnet's prohibitive fees. Farcaster, a decentralized social protocol, primarily runs on Base. By May 2026, Base processes over 300 TPS and has crossed 9 million weekly active addresses — comparable scale to some national payment networks.
ZK Rollups: The Emerging Frontier
While Optimistic Rollups dominate current market share, ZK Rollups represent the architectural ideal: cryptographic proofs provide instant withdrawal finality to Ethereum mainnet, versus the 7-day waiting period of Optimistic systems. zkSync Era and Starknet have achieved full EVM compatibility in 2024-2026 — a previously major limitation — and proof generation times have fallen dramatically through hardware acceleration and algorithmic improvements.
The ZK vs. Optimistic debate will likely resolve through market selection over 2026-2028. As ZK proof generation costs continue falling, ZK Rollups' advantages (instant finality, stronger cryptographic security guarantees, no 7-day bridge delays) become increasingly relevant for high-stakes financial applications. Most industry observers expect ZK Rollups to eventually displace Optimistic Rollups as the dominant L2 architecture — the question is when, not if.
Frequently Asked Questions
Q: How do I bridge ETH to an Ethereum Layer 2?
A: Use the official bridge for each L2 (bridge.arbitrum.io, bridge.base.org, app.optimism.io/bridge). You send ETH to the bridge contract on Ethereum mainnet, and ETH appears on the L2. For cross-L2 bridging, aggregators like Bungee and Li.Fi find the optimal route. Always verify you're using official URLs before submitting.
Q: Is my ETH safe on a Layer 2?
A: Established Optimistic Rollups (Arbitrum, Base, Optimism) have strong security records, though most currently rely on centralized sequencers with fraud proof backstops. ZK Rollups provide stronger cryptographic security. For large, long-term holdings, Ethereum mainnet remains the gold standard. L2s are best for active use with amounts you can afford to have at somewhat elevated risk.
Q: What is the 7-day withdrawal period on Optimistic Rollups?
A: Withdrawing directly to Ethereum mainnet from an Optimistic Rollup requires a 7-day challenge period for fraud proofs. However, third-party bridges (Across, Hop Protocol, Stargate) use liquidity pools to give you ETH immediately for a small fee, bypassing the wait. Most users never experience the 7-day delay.
Q: What is the Optimism Superchain?
A: The Optimism Superchain is a network of L2s built on the OP Stack that share bridging infrastructure and security. Base, Mode, Zora, and many other chains are part of the Superchain. The vision is seamless asset and message flow across all OP Stack chains — collectively forming a decentralized alternative to monolithic blockchains.
Q: Can I use MetaMask on Layer 2s?
A: Yes. All major Ethereum L2s are EVM-compatible and work with MetaMask. Add the L2 network through MetaMask's network settings or via Chainlist.org, then bridge ETH to pay gas. Your same Ethereum address works identically across all EVM-compatible L2s — same private key, same wallet.
Related: Solana vs Ethereum 2026 • DeFi TVL Recovery Analysis