L2 is short for Layer 2, a network built as a scaling layer on top of the Ethereum mainnet (Layer 1). Visit the Binance official site to check spot prices for major L2 tokens, download the official Binance app to trade L2 assets like ARB, OP, and MATIC, and iPhone users should check the iOS installation guide first. L2's core job is to cut Ethereum mainnet's high gas fees by 10 to 100x, while anchoring security back to the mainnet through cryptography. As of April 2026, L2Beat data shows total TVL across major L2s at roughly $38 billion, with daily active addresses now exceeding the Ethereum mainnet's.

Why L2 Exists in the First Place

The Ethereum mainnet handles roughly 15 transactions per second, nowhere near enough to support global-scale applications. During the 2021 NFT boom, gas fees on a single Ethereum transaction spiked past $200, and during the 2022 ApeCoin airdrop, network congestion lasted over 6 hours, pricing ordinary users out entirely. In 2020, Vitalik Buterin proposed a "rollup-centric" scaling roadmap, explicitly handing off the scaling job to L2s while the mainnet focuses on being a settlement and data-availability layer.

L2's design philosophy is to move transaction execution off the mainnet, writing back only transaction summaries or proofs. This delivers three layers of benefit: throughput jumps from 15 TPS to somewhere between 2,000 and 65,000 TPS, gas fees drop from dollar-level to cent-level, and finality security is still inherited from the Ethereum mainnet. Vitalik has repeatedly emphasized that L2 isn't a replacement for Ethereum — it turns Ethereum into "the TCP/IP of blockchains."

The Main Technical Approaches to L2

L2s fall into four main camps based on how they handle data availability and validity proofs. Optimistic Rollups assume transactions are valid, backed by a 7-day challenge window; ZK Rollups use zero-knowledge proofs to verify each batch of transactions on the mainnet; Validiums keep data on an availability layer outside the mainnet; Plasma is an earlier approach that's largely been superseded by rollups. The table below rounds up the leading solutions.

Arbitrum, the leading Optimistic Rollup, currently has about $14.5 billion in TVL, roughly 38% of the entire L2 market. Its core mechanism is called a "fraud proof" — transactions get bundled and posted to the mainnet, and anyone can submit proof of cheating during a 7-day challenge window. This "assume it's right, correct it if it's wrong" approach keeps the system simple and efficient, but it comes with a 7-day withdrawal delay.

ZK Rollups use zero-knowledge proofs to mathematically prove the validity of each batch of transactions on-chain. zkSync Era uses SNARK proofs, Starknet uses STARK proofs, and Polygon zkEVM uses a hybrid approach. The advantage of the ZK path is that there's no need to wait out a challenge window — in theory, funds can move from L2 to L1 in minutes — but generating a ZK proof has a high hardware cost, with a single proof's computation requiring hundreds of dollars' worth of GPU time.

Comparing the Leading L2s by the Numbers

Here's a comparison of the top ten L2s by TVL right now. Data from L2Beat, as of April 2026.

Base's rise is the most notable story here. Just two years after launch, Coinbase's Base has grown its TVL from zero to $8.8 billion, overtaking OP Mainnet to become the second-largest L2. Base has no token of its own, running entirely on ETH for gas, and it's pulled in a huge base of mainstream users on the strength of Coinbase's compliance credentials and the absence of any token-driven noise.

Arbitrum vs. Optimism: A Split in Strategy

Arbitrum and Optimism are both in the Optimistic Rollup camp, but their strategic paths differ. Arbitrum has gone the "app-chain" route — Arbitrum Orbit lets any team build its own L3 on top of the Arbitrum tech stack. Optimism has gone the "superchain" route — the OP Stack is the underlying framework behind Base, Worldcoin, and Mantle, with gas revenue from the whole superchain ecosystem flowing back proportionally to the Optimism Foundation.

The OP Stack shipped its Bedrock upgrade in June 2024, switching data calldata from raw format to EIP-4844 blobs and cutting gas costs by 10x. Arbitrum's Stylus upgrade, for its part, added support for writing smart contracts in Rust and C++ for the first time, expanding its developer pool from 1 million Solidity engineers to 10 million Rust engineers.

The ZK Rollup Big Three

zkSync Era, Polygon zkEVM, and Starknet are the first-tier ZK Rollups, and their technical approaches diverge significantly.

zkSync Era emphasizes "Ethereum compatibility" rather than "Ethereum equivalence." It runs its own zkEVM bytecode that isn't identical to native EVM, but 99% of Solidity contracts can be deployed as-is. Polygon zkEVM takes an "Ethereum equivalence" approach, with a corresponding ZK circuit for every mainnet opcode in its zkEVM, giving it the highest security guarantee at some cost to efficiency. Starknet uses its own Cairo language for writing contracts, offering the best performance but with a higher cost for developers to migrate over.

Here's a comparison of the three's key parameters.

How EIP-4844 Changed L2

In March 2024, the Ethereum mainnet shipped EIP-4844 (Proto-Danksharding), a watershed moment in L2 history. Before 4844, L2s stuffed their data into mainnet calldata, paying for every byte of calldata on every transaction — accounting for roughly 70% of an L2 user's total cost. 4844 introduced the blob data format, a temporary storage type built specifically for L2s: blob data is auto-deleted after 18 days, and its storage cost is only about 1/10 that of calldata.

The real-world impact was immediate. Within 24 hours of the 4844 rollout, Arbitrum's average gas fee dropped from $0.30 to $0.03, and Base's dropped from $0.20 to $0.02. L2 user experience is now approaching that of centralized exchanges.

L2 Cross-Chain Bridges and Security Risk

Cross-chain bridges are the funding channel between an L2 and the mainnet. Native bridges, run by the project team, inherit their security from the underlying rollup protocol; third-party bridges (Across, Stargate, Synapse) use liquidity pools or multisig mechanisms, which are fast but carry the bridge's own smart contract risk.

Historically, bridge hacks have cost more than $2.5 billion cumulatively. In 2022, North Korea's Lazarus Group drained $625 million from the Ronin bridge, the Wormhole bridge lost $325 million, and the Nomad bridge lost $190 million. The lesson: prioritize native bridges for large amounts (even though it means waiting 7 days), and reserve third-party bridges for smaller, everyday transfers.

L2 Tokenomics

L2 tokens generally follow a dual value-capture model built around governance plus sequencer revenue. Here's a rundown of the tokenomics for the leading L2s.

Sequencer revenue is L2's core business model. In 2024, Base's full-year sequencer revenue reached roughly $120 million, versus roughly $80 million for Arbitrum and roughly $55 million for Optimism. Most of this revenue, after deducting L1 data costs, flows back to the project.

Where L2 Is Headed Next

In October 2024, Vitalik published "The Surge" roadmap for Ethereum, laying out a clear goal for L2s over the next three years: unified throughput of 100,000 TPS. Getting there requires three paths working together: first, EIP-7594 (PeerDAS), which would expand blob capacity by another 8x; second, hardware-accelerated ZK proving, where dedicated ASICs would let ZK Rollups batch-prove millions of transactions at once; and third, a cross-L2 interoperability standard that lets users move between different L2s without even noticing.

The app-chain wave is also accelerating. The OP Stack has already been adopted by 50+ projects and Arbitrum Orbit by 30+, and combined with Cosmos SDK, Polygon CDK, and zkSync Hyperchain, the app-chain ecosystem is projected to top 200 chains by 2026. This "L3 + L2 + L1" layered structure is pushing blockchain's physical shape closer to that of the internet itself.

Key Metrics from an Investment Perspective

If you're treating L2s as investment targets, there are a few core metrics worth watching. First is net sequencer revenue, the project's actual cash flow; second is monthly active developers, a proxy for ecosystem vitality; third is decentralization progress — a single sequencer is a common issue across L2s today, and how fast that gets decentralized shapes long-term confidence.

Note that L2 tokens generally trade at a valuation premium. ARB, OP, and STRK all have FDVs in the tens of billions of dollars, while annualized net sequencer revenue is only a few tens of millions — implying a P/E multiple in the hundreds. Investors should be wary of the selling pressure that comes with token unlock schedules.

Frequently Asked Questions

Q: Are assets on L2 actually safe?

L2 security comes down to three factors: bug-free smart contract code, reliable data availability, and a working fraud-proof or validity-proof mechanism. L2Beat rates each L2's decentralization on a Stage 0/1/2 scale; right now only Arbitrum and Optimism have reached Stage 1, while the rest are still Stage 0 (meaning the project team retains emergency upgrade powers).

Q: How long does it take to move funds from the mainnet to an L2?

Deposits from the mainnet to an L2 are near-instant, landing within 10-15 minutes. Withdrawals from an L2 back to the mainnet take 7 days on an Optimistic Rollup, typically 1-3 hours on a ZK Rollup. Using a third-party bridge like Across, it's 5-30 minutes in either direction, but you'll pay a bridge service fee.

Q: Why don't some L2s have their own token?

Base and Linea chose not to launch a token, for three reasons: first, to sidestep securities-law risk (Coinbase and ConsenSys are both compliance-focused companies); second, to avoid diluting the gas value of mainnet ETH; third, to attract high-quality users through a "no airdrop" approach rather than airdrop farmers. This model helped push Base's TVL to second place within two years.

Q: What's the difference between an L2 and a sidechain?

An L2's security is inherited from the Ethereum mainnet, with every transaction ultimately verified on L1. A sidechain (like early Polygon PoS or BSC) has its own consensus and validators, with security independent of Ethereum. In an extreme case where an L2's bridge gets hacked, assets can still exit through the mainnet as an emergency measure; if a sidechain gets hacked, it's an isolated event.

Q: How do I pick the right L2 for me?

Pick based on your use case. For everyday DeFi trading, go with Arbitrum or Base, since both have deep TVL and liquidity. If you need fast withdrawals, go with the ZK Rollup camp (zkSync, Starknet, Polygon zkEVM). For gaming or NFTs, try Immutable X or Ronin. For social chains, try Lens (on Polygon) or Farcaster (on Base).

Q: Will L2 gas fees climb back up?

Not in the short term. Post-EIP-4844, blob capacity is 5-10x current L2 data demand and nowhere near saturated. Even if L2 usage grows another 10x, blobs can still absorb it. The next gas fee spike would likely come from some super-app triggering blob congestion, but EIP-7594 (PeerDAS) is set to expand blob capacity another 8x, expected to hold through 2027.

Q: Base has no token — how do you invest in the Base ecosystem?

Through native project tokens built on Base. Friend.tech, Aerodrome (AERO), BMX, and TOSHI are all leading Base ecosystem projects. Coinbase's stock (COIN) is also an indirect beneficiary, since Base's sequencer revenue flows straight into Coinbase's corporate financials.

Q: Will ZK Rollups or Optimistic Rollups win out?

Over the long run, ZK Rollups are technically superior, but Optimistic Rollups currently have a stronger ecosystem. Vitalik has predicted that Optimistic will gradually migrate toward ZK over the next 5-10 years, since ZK offers faster withdrawals and stronger security. Arbitrum is already researching a hybrid "Arbitrum Stylus plus ZK" approach.

Risk Disclosure and Disclaimer

L2 tokens generally carry valuation premium risk. Leading L2 tokens like ARB, OP, and STRK have FDVs in the tens of billions, while their actual cash flow (net sequencer revenue) falls well short of supporting current valuations, with selling pressure persisting through token unlock schedules. This article is compiled from public data and does not constitute investment advice. Please read our disclaimer.

Data sources: L2Beat, DefiLlama, the Ethereum Foundation's public blog, and governance forums of various L2 projects (data as of April 2026).