The problem they solve
A base blockchain has every participant verify every transaction, which is what makes it trustworthy and also what limits how many transactions it can handle. Raising that limit means requiring more of every participant, which concentrates the network among fewer of them.
Layer 2 sidesteps the trade rather than resolving it. Transactions happen elsewhere, in bulk, and only a compressed proof or summary goes to the base chain. Security is inherited; throughput is not constrained by it.
The types that matter
Rollups
batch many transactions, execute them off-chain and post the compressed data back. Optimistic rollups assume the batch is valid and allow a challenge window in which anyone can dispute it. Zero-knowledge rollups post a cryptographic proof of validity instead, removing the need to wait. Arbitrum, Optimism and Base are the widely used examples on Ethereum.
Payment channels
open a direct channel between two parties, allow unlimited transfers inside it, and settle only the final balance on the base chain. The Lightning Network on Bitcoin works this way.
Sidechains
run as independent chains with their own validators and a bridge to the base chain. Strictly they are not layer 2, since they do not inherit the base chain’s security, and they are frequently grouped with them anyway.
What changes for payments
Cost, mostly, and dramatically.
A Bitcoin transaction on the base chain costs dollars at busy moments. The same payment over Lightning costs a fraction of a cent and settles instantly, which is what makes small-ticket Bitcoin payments viable at all.
On Ethereum, base-chain fees put anything under fifty dollars in an awkward position. On Base or Arbitrum the same transfer costs cents.
Two constraints come with this.
Each layer 2 is a separate network for address purposes
USDC on Arbitrum and USDC on Ethereum are the same token on different rails, and sending between them without a bridge loses the funds in the usual way.
Optimistic rollups have withdrawal delays
Moving funds from the layer 2 back to the base chain involves the challenge window, historically about seven days. Bridges offer faster routes for a fee, and the underlying delay is a property of the design.
Where this is going
The direction is worth noting, because it changes what a merchant should plan for.
Ethereum’s own roadmap treats layer 2 as the place ordinary transactions belong, with the base chain settling batches rather than individual payments. Shopify built its USDC acceptance on Base rather than on Ethereum directly, which is the same logic applied commercially.
For a business the practical implication is that the network a customer pays on will increasingly be a layer 2 rather than a base chain, and support for those networks is worth asking a provider about specifically.