What Is a Layer 2?

A layer 2 (L2) is a network that processes transactions outside a main blockchain (the "layer 1", almost always meaning Ethereum in practice), then posts compressed results back to it, inheriting the L1's security while escaping its congestion. The problem being solved is brutally simple: Ethereum settles for the world but processes transactions in the low double digits per second, and during the 2021 mania a single swap could cost more in gas than most people were swapping. The scaling debate consumed years of the industry's attention; rollups are the answer Ethereum's roadmap settled on, and by now chains like Arbitrum, Optimism, and Base carry a large share of all Ethereum-adjacent activity at a fraction of mainnet cost.

We were early to take L2s seriously. Our 2022 protocol coverage centred on GMX largely because it chose Arbitrum when serious DeFi on an L2 was still a novelty, and the reasoning we wrote then (L2 performance makes trading-grade applications possible at all) aged well. The costume has changed, with dozens of L2s now competing, but the logic hasn't.

How a rollup works, minus the whiteboard

A rollup executes transactions on its own fast, cheap network, then "rolls up" batches of them into compressed data posted to Ethereum. Ethereum stores that data and acts as the supreme court. The two rollup families differ in how the court verifies.

Optimistic rollups (Arbitrum, Optimism, Base) post results assumed-valid, with a challenge window, typically about a week, during which anyone can submit a fraud proof showing a batch was invalid. Security assumption: at least one honest party is watching. Practical consequence: withdrawing back to Ethereum through the native bridge waits out the window, which is why third-party fast bridges exist for the impatient.

ZK rollups (zkSync, Starknet, Scroll, and a fast-growing field) post a cryptographic validity proof with every batch. Ethereum checks the math, so there's nothing to challenge and no week-long wait. Harder to build, mathematically heavier, and widely viewed as the endgame design as the proving technology matures.

Either way, the load-bearing property is the same: the rollup's state can be reconstructed and enforced from data on Ethereum. That's what separates a true rollup from a sidechain like Polygon's original PoS chain, which runs its own validators and its own security budget. A perfectly usable network, but if its validator set fails you, Ethereum owes you nothing. "Layer 2" gets used loosely as marketing; whether the L1 actually backstops the network is the question that sorts the label from the reality.

The caveats that belong in the brochure

L2s inherit Ethereum's security for settlement, not for everything. Most today run a single centralized sequencer that orders transactions. It can't steal funds, but it can go down or censor until decentralisation roadmaps deliver. Many retain upgrade keys or security councils that could, in theory, change the rules under you. The community grades these trust assumptions publicly (L2Beat's framework is the standard), and the grades deserve a look before you park serious value. And bridges between chains remain the industry's historical disaster zone. The largest hacks on record include several bridge drains, which is why our bridging guide leans hard toward native and battle-tested routes over whatever's fastest today.

Using them

Day to day, an L2 feels like Ethereum with the meter turned down: same wallets, same address format, same DeFi primitives, fees typically cents rather than dollars. Ethereum amplified the cost gap in March 2024 by adding blob space, a data lane priced specifically for rollups. The mechanics that matter: your wallet must be on the right network before anyone sends you anything (same address, different chain is the classic mistake), each L2's tokens are distinct assets from their mainnet cousins until bridged, and gas on Ethereum-aligned L2s is still paid in ETH. Several major L2s have their own tokens too, largely governance instruments born of some of the biggest airdrops ever run. You'll find the ones we track on our coin pages, where the usual tokenomics questions apply with extra force: an L2 token's relationship to the network's actual usage is often loose, and pricing that gap honestly is the whole game.

Frequently asked questions

Is a layer 2 as safe as Ethereum itself?

Not quite, and the gap varies by network. A true rollup inherits Ethereum's settlement security, but most L2s today add trust in a centralized sequencer and sometimes upgrade keys. For working funds the difference rarely matters; for long-term cold storage, mainnet remains the conservative choice.

Why are layer 2 fees so much cheaper?

Because hundreds of L2 transactions share the cost of one compressed posting to Ethereum, and since March 2024 that posting uses blob space priced far below regular calldata. You're splitting one mainnet bill with everyone else in the batch instead of paying it alone.

Which layer 2 should I use?

For a first run, one of the large incumbents (Arbitrum, Optimism, or Base) is the boring, correct answer: deepest liquidity, broadest app support, most scrutiny. Differences that matter to a normal user are small; pick where the app you want lives, and check L2Beat if you're curious about trust rankings.