Layer 1 and layer 2: how blockchains scale

When many people try to use the same blockchain at the same time, transactions slow down and fees go up. This is not an occasional glitch: it is a consequence of how open networks such as Bitcoin and Ethereum were designed. The most widely adopted answer today is to split the work into layers.

This article explains what layer 1 and layer 2 are, why this division emerged, how the main solutions work and what risks they add. The information was checked on October 8, 2026.

The problem: why blockchains process so little

In a public blockchain, each node, that is, each computer participating in the network, receives and checks every transaction. This is what allows anyone to verify the record without trusting anyone. To review how this works, see How the Bitcoin blockchain works.

The cost of this verification is capacity. The paper that proposed the Lightning Network, published in 2016 by Joseph Poon and Thaddeus Dryja, noted that Bitcoin supported fewer than 7 transactions per second with 1-megabyte blocks. To reach the peak of 47,000 transactions per second that Visa had achieved in 2013, each block would need to be nearly 8 gigabytes, which no home computer could keep up with (Poon and Dryja, 2016). The figures are from that time and serve to show the order of magnitude of the problem.

The paper’s central point is that giant blocks would reduce the number of people able to validate the network and lead to centralization. ethereum.org makes the same argument: increasing the main network’s capacity would require bigger nodes and specialized hardware, which would reduce who can run a node (ethereum.org, layer 2).

This dilemma is often summarized as the trade-off between decentralization, security and scalability. A Bank for International Settlements (BIS) bulletin from July 2026 describes exactly this balance among the three factors and states that it helps explain both the existence of multiple layer 1 networks and the expansion of layer 2 solutions (BIS, Bulletin 126).

What layer 1 is

Layer 1 is the main blockchain: the network that has its own consensus mechanism and where transactions are settled with finality. Bitcoin and Ethereum are examples. How networks reach agreement on the record is explained in How consensus mechanisms work.

There are two ways to gain capacity. The first is to change layer 1 itself, which ethereum.org calls on-chain scaling: bigger blocks, new data rules, protocol changes. It requires consensus across the whole network and runs into the limit you saw above. The second is to process transactions outside it, off-chain scaling (ethereum.org, scaling).

What layer 2 is

Layer 2 is a network built on top of a layer 1. It processes transactions outside the main network and uses layer 1 to record results and settle disputes.

The criterion that matters is where the security comes from. For ethereum.org, a true layer 2 solution inherits the main network’s security. Other solutions interact with Ethereum but have their own security (ethereum.org, scaling):

• Sidechains: independent blockchains, linked to the main chain by bridges, with their own consensus rules. If their validators fail, the main network does not protect you.
• Validiums and plasma: use proofs linked to Ethereum but keep data outside it, which changes the guarantees.

In practice, many services present themselves as “L2s” without meeting this criterion. Always ask: if the operator disappears, can I recover my funds through the main network?

Rollups: the dominant solution on Ethereum

Rollups execute transactions outside layer 1, bundle many of them into batches and post the data to the main network. Because the data stays on layer 1, anyone can reconstruct the network’s state and challenge errors. According to ethereum.org, rollups have become Ethereum’s main scaling path (ethereum.org, scaling). Their role in Ethereum’s strategy is also covered in What Ethereum is and how it works.

There are two main types.

Optimistic rollups. They assume batches are valid. After a batch is posted, a window opens in which anyone can challenge it with a fraud proof, which demonstrates the error on the main network. Whoever posted a faulty batch loses the bond they deposited. Security depends on at least one honest participant keeping watch. The cost to you shows up at withdrawal: to move funds back to Ethereum, you have to wait for the challenge period to end, about seven days. Some services will pay out the withdrawal earlier, for a fee (ethereum.org, optimistic rollups).

Zero-knowledge rollups (ZK-rollups). Along with each batch, they submit a validity proof, a cryptographic certificate that the transactions were processed correctly. A contract on layer 1 checks the proof before accepting the batch. That is why there is no challenge period, and withdrawals are released as soon as the proof is verified. The drawbacks: generating proofs requires a lot of computation, sometimes specialized hardware, which can concentrate operation among a few participants, and adapting complex Ethereum programs to this format is technically difficult (ethereum.org, ZK-rollups).

Layer 1 also changed to help

Rollups pay to post data to Ethereum, and this was one of their largest costs. The EIP-4844 proposal created a transaction type with blobs, cheaper chunks of data that remain available for about 18 days and can then be deleted by nodes. The idea is that the data only needs to exist long enough for honest participants to reconstruct the rollup’s state (EIP-4844). The Fusaka upgrade, in December 2025, introduced PeerDAS, which according to ethereum.org makes publishing and retrieving this data more efficient (ethereum.org, layer 2).

Payment channels: Bitcoin’s Lightning Network

On Bitcoin, the main layer 2 solution follows a different design. The Lightning Network uses payment channels: two people record on the blockchain only the opening and closing of a channel and, between those two moments, exchange payments off-chain as many times as they like. Channels connect into a network, and a payment can pass through several of them before reaching its destination (Poon and Dryja, 2016).

The paper itself points out risks. Each participant needs to monitor the blockchain, or delegate monitoring to third parties, to prevent the other party from trying to close the channel with an old balance. Failing to react in time can mean losing funds. The authors also cite as the biggest systemic risk the forced closing of many channels at the same time, which could congest the main network.

The risks the extra layers bring

A layer 2 can be faster and cheaper, but it is not a “risk-free” version of the main network. ethereum.org itself states that an L2’s security depends on the technology, the security of its contracts and the project’s maturity, and recommends starting with small amounts (ethereum.org, layer 2).

The main points of attention:

• Centralized operator. Many rollups depend on a single sequencer, the operator that orders and posts transactions. It can influence the order of operations, go offline or fail to include transactions. Some rollups allow sending the transaction directly through layer 1 to get around the block, but this feature varies from network to network (ethereum.org, optimistic rollups).
• Bridges. Moving assets between networks requires bridges, which are additional contracts and operators. The BIS notes that bridges and multi-chain issuance reduce friction but create new dependencies on trust, governance and operational resilience (BIS, Bulletin 126).
• Fragmentation. With many layer 1s and layer 2s, infrastructure, liquidity and assets end up split across networks, according to the same bulletin. A token on one L2 may not be the same as the “same” token on another.
• Upgradeable contracts. In the editorial team’s assessment, this is a little-noticed point: many L2s are controlled by contracts that their administrators can change. Whoever holds that key has power over the funds.
• Withdrawal delays. On optimistic rollups, exiting through the official route takes about a week.
• Network mistakes. Sending assets to the right address on the wrong network can make recovery difficult or impossible. Before transferring, check which network the platform or wallet operates on. Choosing between wallets is covered in Crypto wallets: custodial and non-custodial.

As with any crypto asset, the tokens circulating on these networks can lose all their value, and transactions generally cannot be reversed.

Key takeaways

• Layer 1 is the blockchain that defines consensus and final settlement; layer 2 processes transactions outside it and uses layer 1 as an anchor.
• Optimistic rollups rely on challenges via fraud proofs and have slow withdrawals; ZK-rollups use validity proofs and require heavy computation. On Bitcoin, the Lightning Network uses payment channels.
• More capacity comes with new points of trust: sequencers, bridges, upgradeable contracts and liquidity fragmented across networks.