In this guide
The scaling problem
Ethereum has a problem. It's popular. Too popular for its own infrastructure. When lots of people want to use it at the same time, gas fees spike. During the 2021 NFT boom, simple token transfers cost $50-100. A complex DeFi transaction could cost $200+.
This isn't a bug. It's by design. Ethereum prioritizes security and decentralization over speed. Every transaction has to be verified by thousands of nodes worldwide. That takes time and limits throughput.
The blockchain trilemma (coined by Vitalik Buterin) says you can only optimize two out of three: decentralization, security, or scalability. Ethereum chose the first two. Layer 2s are the answer to the third.
Instead of processing every transaction on Ethereum directly (Layer 1), you can batch them on a separate network (Layer 2) and submit a compressed summary back to Ethereum. You get Ethereum's security with much higher speed and much lower fees.
What Layer 2s actually are
A Layer 2 is a separate blockchain that runs on top of Ethereum (the Layer 1). It processes transactions off-chain, then posts the results back to Ethereum for final settlement. Think of Ethereum as a courthouse and L2s as arbitration centers. Most disputes get resolved without going to court, but the court is always there as the final authority.
The key innovation is that L2s inherit Ethereum's security. If the L2 shuts down or tries to process fraudulent transactions, Ethereum can step in. Your funds are ultimately secured by Ethereum's validator set, even though transactions happen elsewhere.
L2s achieve much higher throughput because they bundle hundreds or thousands of transactions into a single batch. Instead of paying Ethereum gas fees for each individual transaction, the cost of one Ethereum transaction gets split across all the transactions in the batch. A swap that costs $15 on Ethereum might cost $0.05 on an L2.
This approach became even cheaper after Ethereum's "Dencun" upgrade in March 2024, which introduced "blob space" specifically for L2 data. L2 fees dropped by 90%+ overnight. Transactions on Arbitrum and Base now cost less than a penny.
Optimistic vs ZK rollups
There are two main types of Layer 2 rollups, and the differences matter.
Optimistic rollups assume transactions are valid unless proven otherwise. They post transaction data to Ethereum and give a "challenge period" (usually 7 days) where anyone can prove a transaction was fraudulent. If nobody challenges it, the batch is finalized. If someone does challenge it, the protocol verifies the dispute on Ethereum.
Arbitrum and Optimism are the two biggest optimistic rollups. They work now, are battle-tested, and support basically every Ethereum app. The downside: that 7-day challenge period means withdrawing funds from L2 back to Ethereum takes a week. Third-party bridges can speed this up, but they add trust assumptions.
ZK (Zero Knowledge) rollups use mathematical proofs to verify transactions. Instead of assuming validity and waiting for challenges, they generate a cryptographic proof that all transactions in a batch are valid. This proof gets verified on Ethereum. Once verified, the batch is finalized immediately.
ZK rollups are theoretically superior. Faster finality, no 7-day wait for withdrawals, and stronger security guarantees. But the math is incredibly complex. zkSync, StarkNet, Scroll, and Polygon zkEVM are the major ZK rollups. They're newer and still maturing, but progressing fast.
My take: optimistic rollups won the first round because they shipped earlier. ZK rollups will likely win long-term because the technology is fundamentally better. Both will coexist for years.
Major L2 networks
Arbitrum: The largest L2 by TVL (total value locked), with over $15 billion in early 2026. It has the most DeFi protocols and the deepest liquidity. ARB is its governance token. If you're moving to L2 for the first time, Arbitrum is a safe default choice.
Optimism (OP Mainnet): Close behind Arbitrum in size. Known for the "Superchain" vision where multiple L2s share infrastructure. It's the foundation for Base (Coinbase's L2) and several other chains. OP token is used for governance.
Base: Built by Coinbase using Optimism's technology. No native token (yet). Huge growth in 2024-2025 because Coinbase funnels millions of users directly to it. Tons of activity around meme coins, social apps, and DeFi. Fees are absurdly low, usually under $0.01.
zkSync: The leading ZK rollup by activity. Has its own token (ZK). Growing ecosystem of DeFi and gaming apps. Technically impressive but still building out its app ecosystem.
StarkNet: Uses a different proof system (STARKs instead of SNARKs). Very technically advanced with its own programming language (Cairo). Smaller ecosystem but strong developer interest. STRK is the native token.
How to use Layer 2s
Using an L2 is almost identical to using Ethereum. The same wallet (MetaMask, etc.) works. The same DeFi protocols (Uniswap, Aave) are deployed. You just need to switch networks.
Option 1: Bridge from Ethereum. Use the official bridge for each L2 (bridge.arbitrum.io, app.optimism.io/bridge) or a third-party bridge like Across or Stargate. Send ETH from Ethereum to the L2. For optimistic rollups, bridging in is instant. Bridging back takes 7 days (or use a fast bridge).
Option 2: Withdraw directly from an exchange. Most major exchanges now support direct withdrawals to L2s. On Coinbase, you can withdraw ETH directly to Base, Arbitrum, or Optimism. On Binance, same thing. This is the cheapest option since you skip the Ethereum bridging fee entirely.
Option 3: Buy on the L2 directly. Some L2s have on-ramps where you can buy crypto with a credit card and receive it directly on the L2. No bridging needed.
Once your funds are on an L2, everything works the same. Swap tokens on Uniswap, lend on Aave, mint NFTs. The experience is identical but transactions confirm in seconds and cost fractions of a cent.
Risks and tradeoffs
Sequencer centralization: Most L2s currently rely on a single "sequencer" that orders transactions. If the sequencer goes down, the L2 stops processing. This has happened to Arbitrum and other L2s. Decentralizing sequencers is a work in progress.
Bridge risk: Moving funds between L1 and L2s involves bridges. Bridges have been the target of some of crypto's biggest hacks. Use official bridges when possible, and avoid bridging huge amounts through new or unaudited services.
Upgrade risk: L2s are still being upgraded frequently. The teams behind them have significant control over the code through admin keys and upgrade mechanisms. This is improving but most L2s aren't fully trustless yet. L2Beat.com tracks the decentralization status of every L2.
Liquidity fragmentation: With dozens of L2s, liquidity gets spread thin. The best trading prices and deepest pools might be on different chains. Moving between them takes time and costs fees. This is one of the biggest user experience challenges in crypto right now.
Despite these tradeoffs, L2s are clearly the future of Ethereum. The roadmap is "rollup-centric," meaning Ethereum itself is optimized for supporting L2s rather than trying to scale the base layer directly. If you're using Ethereum DeFi today and not on an L2, you're paying 10-100x more than you need to.
The bottom line
Layer 2s solve Ethereum's biggest weakness: cost and speed. They're not a compromise. They're the plan. If you're interacting with Ethereum at all, you should be doing it through an L2 for 95% of activities. The fees are pennies, the speed is instant, and the security inherits from Ethereum itself.
Related reading: cross-chain bridges, Ethereum, and DeFi.