Layer 1 vs. Layer 2: Understanding Blockchain Scaling Solutions

Update, 6 August 2026. This page is no longer our standing reference for Ethereum and Layer 2s. It has been superseded by Ethereum and Layer 2s: How the Rollups Work, What Their Numbers Measure, and Why Two Trackers Disagree by Four Times, which carries the current measurements. This page is kept as the shorter introduction to the Layer 1 and Layer 2 distinction.

Every blockchain runs into the same wall, often called the scalability trilemma: pick two of three — secure, decentralized, fast — and the third gets compromised. Bitcoin and Ethereum picked security and decentralization first, which is exactly why their base layers cap out at a modest number of transactions per second, and why fees spike the moment demand does.

Layer 1 is the base chain itself — Bitcoin, Ethereum, Solana, the network validating transactions under its own consensus rules. Scaling it directly means touching something fundamental: block size, block time, the consensus mechanism itself. Each of those changes forces a tradeoff most communities are reluctant to make, and getting the whole network to agree on the change is its own battle.

Layer 2 sidesteps the fight entirely by building on top rather than rebuilding underneath. Ethereum’s rollups — Optimism, Arbitrum — batch thousands of transactions off-chain, then post a compressed proof back to the base layer. Users get a fraction of the fee and a fraction of the wait, while still inheriting most of Ethereum’s security, since any dispute can ultimately be settled on Layer 1.

Rollups aren’t the only design. Sidechains run as fully independent chains with their own validator sets, bridged to the main chain — faster and cheaper, at the cost of some security. State channels, the model behind Bitcoin’s Lightning Network, let two parties transact directly off-chain and settle only the opening and closing balance on-chain, which makes them well suited to frequent, small payments between the same two counterparties.

For everyday users the net effect is optionality. Moving a large position, the higher fee for Layer 1 settlement is worth paying for the security; running frequent small transactions, a Layer 2 wins on both cost and speed. As these networks interoperate more seamlessly, that distinction will likely matter less — but understanding it now is what makes sense of the fee, speed, and security tradeoffs showing up across the market today.

What Ethereum is for

Bitcoin is a ledger that records who owns what. Ethereum is a ledger that also records what code should run and lets anyone deploy it. That is the entire difference, and it produces two very different assets. Bitcoin’s pitch is that it will not change. Ethereum’s pitch is that it is a settlement layer other things get built on, which means it has to change, and has: it moved from proof of work to proof of stake, restructured its fee market, and rebuilt its scaling roadmap around rollups. Judging one by the other’s criteria produces most of the bad analysis in this sector. The comparison we do make, on value locked rather than rhetoric, is in our Bitcoin reference.

The practical consequence is that Ethereum is where the activity is. On figures pulled on 3 August 2026, DefiLlama counted $74.6 billion of value locked across 461 chains; Ethereum held $40.76 billion of that, or 54.7%. Almost every application described in the DeFi reference either runs on Ethereum or on a chain that settles to it.

Proof of stake, in plain terms

Validators lock up a deposit and are selected to propose and attest to blocks. Behave correctly and the deposit earns a modest return funded by new issuance and by fees. Behave incorrectly — sign two conflicting blocks, go offline at scale — and part of the deposit is destroyed. That is the whole security model: capital at risk rather than electricity spent. It removes the energy argument entirely and introduces a different one, which is that stake concentrates where capital concentrates, and a large share of it now sits behind a small number of staking services and exchanges.

Our view is that validator and client concentration is the risk worth tracking on Ethereum, in the same way that mining pool concentration is the risk worth tracking on Bitcoin, and for the same reason: both are places where a decentralized system quietly acquires a small number of load-bearing operators. It is measurable from public data, and it is the next item in our concentration series.

The fee market, and why gas behaves the way it does

A transaction pays for computation, measured in gas. Each block has a target size and a base fee that rises when blocks run above target and falls when they run below, which makes fees respond to congestion within a few blocks rather than through an open auction for every slot. The base fee is destroyed rather than paid to a validator, which means heavy usage removes ether from supply; a priority tip on top buys faster inclusion. None of this makes Ethereum cheap at peak demand. It makes the price of block space predictable enough to quote, which is a smaller and more useful claim than the one usually made for it.

Rollups: what they actually inherit

A rollup executes transactions on its own network, compresses them, and posts the data plus a validity argument back to Ethereum. The important word is data. A rollup that posts its transaction data to Ethereum can be reconstructed and exited by anyone even if its own operators disappear; a chain that keeps that data elsewhere cannot, and calling both of them Layer 2s obscures the only difference that matters in a crisis.

Two proof designs exist. Optimistic rollups assume the posted result is correct and allow a challenge window during which anyone can prove fraud, which is why native withdrawals from them take days. Zero-knowledge rollups post a cryptographic proof of correctness with each batch, so finality does not wait on a challenge period, at the cost of far more complex engineering. Both are legitimate; neither is finished. Most also still run a single sequencer, which means one operator decides transaction order and can, in principle, censor or extract value, with an escape hatch to the base layer as the backstop.

The honest summary is that rollups have delivered the cost reduction they promised and have not yet delivered the decentralization. When you read that a Layer 2 is secured by Ethereum, the questions to ask are whether transaction data is posted to Ethereum, whether the proof system is live rather than announced, who can upgrade the contracts and how quickly, and whether a user can force a withdrawal without the sequencer’s cooperation. L2Beat publishes exactly these answers per network, and reading it takes ten minutes.

Sidechains, validiums and the labels that hide risk

A sidechain is an independent blockchain with its own validators, connected by a bridge. It is not secured by Ethereum in any sense; it is secured by its own validator set, and the bridge is an additional point of failure on top. A validium executes like a rollup but keeps data off-chain with a committee attesting to availability, which is cheaper and reintroduces exactly the dependency rollups were designed to remove. Neither is illegitimate. Both are routinely marketed with the same vocabulary as rollups, and the marketing is where the risk hides.

State channels are the fourth model and the oldest: two parties open a channel on-chain, transact privately as often as they like, and settle the net result later. This is the design behind Bitcoin’s Lightning Network, and its constraints — liquidity has to be committed up front, and a participant has to be online to defend against an outdated settlement — are the reason it suits recurring payments between known counterparties rather than general-purpose use.

Staking, liquid staking and where the concentration sits

Running a validator requires a fixed deposit and continuous uptime, which prices most people out and produces an obvious market: pooled and liquid staking services that take deposits of any size, run validators on your behalf and issue a token representing the staked position. That token can then be used as collateral elsewhere, which is efficient and quietly stacks risks — the staking service’s operational risk, the token’s peg to the underlying, and the lending protocol’s liquidation parameters — into a single position that reads on screen as one number.

It also concentrates the validator set. A pool with a large share of stake does not gain the ability to steal, but it does gain influence over transaction ordering and, at extremes, over which chain history the network converges on. This is the Ethereum analogue of the mining pool problem we measured in Three Pools Build 61% of Bitcoin’s Blocks, and we would treat any staking arrangement above roughly a fifth of total stake as a structural question rather than a competitive one.

Moving between layers without losing money

Bridging is the most dangerous routine action in this part of the market, because it concentrates value in contracts that have repeatedly been the largest single losses on record. Prefer a network’s canonical bridge over a third-party one where you can wait; understand that a fast bridge is usually a liquidity provider fronting you the asset for a fee rather than a faster settlement; and check on both sides that you are receiving the native asset rather than a wrapped claim issued by whoever built the bridge. Send a minimum test amount first. Keep a small balance of the destination network’s gas token before you need it, because arriving on a new chain with a valuable token and no way to pay for a transaction is a common and entirely avoidable trap.

What to check before trusting a network

Where is transaction data published. Is the proof system live in production or on a roadmap. Who holds the upgrade keys, is there a timelock, and how long is it. Can a user exit without the sequencer. What is the value locked, and how much of it is one protocol. L2Beat covers the first four, DefiLlama the last, and Etherscan and the equivalent explorers let you confirm that the contracts you are trusting are verified and match the documentation. Everything on this page can be checked from those four sources, and we would rather you checked than took our word.

The asset, and what proof of stake did to its supply

Ether is not a claim on anything and pays no dividend, but unlike Bitcoin it is consumed by usage. The base fee on every transaction is destroyed, while new issuance goes to validators. The result is a supply schedule that is not fixed but conditional: heavy network usage can burn more than is issued, and quiet periods do the reverse. Two honest points follow. This is a genuinely different monetary design from Bitcoin’s and deserves to be argued on its own terms rather than scored against a fixed cap. And it makes the supply path dependent on demand, which means anyone quoting a long-run supply figure for ether is quoting an assumption about usage, not a schedule.

Staking adds a second layer to that. A validator’s return comes from issuance, priority fees and whatever it can capture from transaction ordering, and the headline yield quoted by staking services blends all three, net of their cut. It is a real return in ether terms and it is not a risk-free rate: it carries slashing risk, operator risk, and, for liquid staking tokens, the risk that the token trades below the asset it represents at exactly the moment you want to sell.

Blobs, and why Layer 2 fees fell

Rollups used to post their compressed data into ordinary transaction calldata, which competed for the same block space as everything else and made Layer 2 costs track Layer 1 congestion. Ethereum now provides a separate, cheaper class of data that is retained by nodes for a limited window and then dropped, which is long enough for anyone to reconstruct or challenge the rollup state and short enough not to burden the chain forever. That change is most of the reason Layer 2 transaction costs fell by an order of magnitude rather than by a few per cent.

It also introduces a second fee market with its own supply and demand, so Layer 2 costs can now spike independently of Layer 1 when many rollups compete for data space at once. If you are budgeting for on-chain activity, the relevant number is no longer just the gas price on the base layer.

Smart accounts, and the end of the seed phrase as the only option

The default account on Ethereum is a key pair and nothing else: no spending limits, no recovery, no way to pay a fee in anything but the native asset. Account abstraction lets a smart contract act as the account instead, which makes it possible to add multiple signers, daily limits, session permissions, social recovery and fee payment in a stablecoin. This is the most consequential usability change in the ecosystem and it is arriving unevenly, wallet by wallet.

It is not free of trade-offs. A smart account is code, and code has bugs and an upgrade path, so it replaces one clean failure mode — lose the phrase, lose the money — with a set of subtler ones. Recovery mechanisms that let a group of guardians restore access also let a group of guardians be socially engineered. The comparison with plain key custody is in the self-custody reference, and the right answer depends on who is holding what and for how long.

The case against Ethereum, taken seriously

It is slower and dearer than the alternatives. True at the base layer, and deliberately so; the counter is that a settlement layer optimises for verifiability, not throughput, and that the throughput lives one layer up. Whether users care about that distinction is an open question that marketing cannot settle.

The roadmap keeps moving. Also true. An asset whose properties are subject to change is a harder thing to hold for a decade than one whose whole proposition is that it will not change, and dismissing that concern as ignorance is a failure of argument rather than a rebuttal.

The Layer 2 ecosystem fragments liquidity. Users now choose between a dozen networks with separate balances, separate gas tokens and bridges in between. Cross-chain standards are improving, but for now this is a genuine cost paid by users to solve a problem they did not create.

Concentration. Staking, sequencers, client software and even the infrastructure providers most applications use to read the chain all cluster around a small number of operators. We think this, rather than throughput, is the substantive risk, and it is the one least represented in the public argument.

Key takeaways

• The scalability trilemma forces a tradeoff: Bitcoin and Ethereum chose security and decentralization over base-layer speed.
• Layer 2s — rollups, sidechains, state channels — add speed and lower fees on top of a base chain rather than rebuilding it.
• Large transfers still favor Layer 1 settlement; frequent, smaller transactions are where Layer 2 wins on cost and speed.

Where this sits in the rest of our coverage

Ethereum and Layer 2 scaling is one of seven subjects we maintain as standing reference pages. The adjacent ones are the full Ethereum and Layer 2s reference, Bitcoin, stablecoins, decentralized finance, self-custody and wallets, exchanges and tax and regulation. Nothing here is investment advice.

Revision history

• 6 August 2026. Expanded from a short explainer into the standing Ethereum and Layer 2 reference. Added proof of stake, the fee market, rollup data availability and proof systems, sidechains and validiums, staking concentration, bridging practice and a verification checklist. Value-locked figures are the 3 August 2026 pull.
• Original publication. Layer 1 versus Layer 2 explainer.