Explainer · Scaling
Layer 2s, explained
October 1, 2026
"Layer 2" gets used as if everyone already knows what problem it solves. In plain terms: it is a separate network that processes transactions on behalf of a base blockchain like Ethereum, then settles a compressed summary back to it. Here is why the base chain can't simply scale up on its own, how the two main rollup designs differ, and what a user is actually trusting when they use one.
The constraint that makes a second layer necessary
A base blockchain like Ethereum — often called Layer 1 — gets its security from having many independent participants verify every single transaction. That is the whole point: no one can cheat, because everyone is checking. But it is also the bottleneck. The more transactions competing for that verification, the more expensive and congested the network gets. A base chain can't simply "get faster" without weakening the decentralization that makes it trustworthy in the first place — the tradeoff is structural, not an oversight to be engineered away.
A Layer 2 is the answer to that constraint, and the most common form it takes is called a rollup. A rollup processes a batch of transactions on its own, separate network, then submits a compressed summary back to the base chain, which still ultimately verifies that everything in it was done correctly. Because the base chain only has to check the summary rather than replay every transaction itself, far more activity can move through the system, while still inheriting most of Layer 1's security — disputes are still settled there.
Optimistic vs. zero-knowledge rollups
- Optimistic rollups. Assume a batch is valid by default and give anyone a window of time to challenge it with proof of fraud if something is wrong. Arbitrum is a well-known example of this approach. Optimistic rollups tend to be simpler to build, at the cost of that challenge window before a transaction is fully final.
- Zero-knowledge (ZK) rollups. Generate a mathematical proof that a batch is valid before it is even accepted, so there is nothing to challenge after the fact. ZK rollups tend to finalize faster, since there's no challenge window to wait out, generally at the cost of more complex infrastructure to build and operate.
The tradeoffs, not just the benefits
Faster and cheaper isn't free — it is a different, usually smaller, set of risks than the base chain's own, and they are worth naming specifically rather than waving at. Sequencer centralization is the first: most rollups today rely on a single operator, the sequencer, to order and post transaction batches to the base chain, rather than the decentralized validator set that secures Layer 1 itself. Decentralizing that role is an active, ongoing project across the industry, not yet the default anywhere.
Bridge risk is the second: moving assets onto or off a rollup means trusting that rollup's own bridge back to the base chain, which is exactly the kind of software that has historically been the most-attacked part of any Layer 2. And fragmented liquidity is the third, a consequence of scaling this way at all: every additional rollup splits activity and trading depth across another separate environment, so a market that is deep on one Layer 2 can be thin on another, with a bridge standing between them.
The read
Layer 2s aren't a workaround or a lesser version of the base chain — they are the scaling answer for a system that refuses to trade away decentralization to get faster. But what a user is trusting changes with the design: a challenge window in one case, a proof system and a bridge in another, a particular operator set in either. Knowing which rollup is being used, and what is being trusted to get that speed, is exactly the diligence that's easy to skip.
This is general-circulation educational content, not investment advice. It explains how Layer 2 scaling works in general; project names mentioned are examples of a category, not recommendations, and nothing here is a recommendation to use, hold or transact on any specific network.
For deeper coverage of a specific Layer 2 discussed here, see Blockchain IQ's research on Arbitrum.