Blockchain technology example: why Layer 2 rollups matter

It is the clearest blockchain technology example of how a base chain can retain settlement authority while moving routine execution elsewhere.

The economic shift is straightforward. Ethereum Layer 1 remains expensive blockspace with high security guarantees. Rollups turn that blockspace into a settlement and data-availability layer. They process user activity away from the main chain, compress the relevant output, then anchor it back to Ethereum.

The result is visible in transaction economics. Ethereum L1 still operates around 15 transactions per second. Major rollups routinely process more than 250 TPS, while aggregate rollup throughput has moved above 325 TPS. Median fees on major networks such as Base, Arbitrum, and Optimism have fallen into a $0.01 to $0.10 range.

That does not mean scaling has eliminated risk. It has redistributed it: from direct L1 gas exposure toward sequencers, bridge contracts, proof systems, governance controls, and fragmented liquidity.

The mechanics: execution moves off-chain, settlement stays on Ethereum

A rollup is a blockchain infrastructure example built around a simple division of labor.

Ethereum is costly because every node must independently validate and retain a large amount of transaction data. That redundancy is the security model. It is also a throughput constraint. A rollup reduces the amount of work Ethereum must perform for each individual user action.

The rollup executes transactions in its own environment. It then batches thousands of those actions and submits a compressed representation to Ethereum. Depending on the design, Ethereum receives transaction data, a state commitment, and either a mechanism for challenging invalid execution or a cryptographic proof that execution was valid.

The core sequence looks like this:

1. Users submit transactions to the rollup. A swap, transfer, NFT mint, perpetuals trade, or smart-contract interaction happens on the Layer 2 network rather than directly on Ethereum L1.

2. A sequencer orders and executes those transactions. In most current architectures, this function is concentrated in one operator or a small set of operators. The sequencer produces the immediate user experience: fast confirmations and low apparent fees.

3. The rollup batches the activity. Instead of posting every transaction individually to Ethereum, it aggregates them into compressed batches.

4. Ethereum receives the settlement data. L1 becomes the final reference point for balances, state transitions, and withdrawals, subject to the rollup’s specific proof and bridge design.

5. Users rely on the rollup’s bridge for movement between layers. This bridge is not an accessory. It is one of the largest risk surfaces in the system.

This is how blockchain technology works in practice when the base layer reaches economic capacity. The chain does not suddenly become infinitely fast. It becomes a coordination layer for multiple execution environments.

Rollups do not make Ethereum cheaper by changing Ethereum’s security model. They make each unit of Ethereum blockspace cover more economic activity.

The distinction matters because low Layer 2 fees are not a free lunch. They are produced by shared batching. One Ethereum settlement transaction can represent a large number of individual L2 transactions. The marginal cost per user falls only because users share the cost of publication and settlement.

That model works well for high-frequency retail activity. It is less clean for capital that must move across chains, protocols, and bridges. Every additional domain adds settlement assumptions and liquidity fragmentation.

Dencun changed the cost structure, not Ethereum L1 gas

The decisive catalyst arrived on March 13, 2024, when Ethereum activated the Dencun upgrade and introduced EIP-4844, commonly called Proto-Danksharding.

Before Dencun, rollups largely published transaction data through Ethereum calldata. Calldata is durable, expensive, and designed for a broader set of use cases than recurring rollup data publication. For rollups, it became a direct operating expense. When L1 gas rose, rollup margins narrowed or end-user fees increased.

EIP-4844 introduced blob transactions. Blobs provide a dedicated, temporary channel for rollup data availability. The data remains available long enough for verification and reconstruction, but it does not need to occupy Ethereum’s permanent execution history in the same way calldata does.

The effect was mechanical rather than speculative:

The 80% to 99% drop in Layer 2 transaction costs did not mean Ethereum mainnet gas became cheaper. That claim is wrong. Dencun targeted the data-availability cost paid by rollups. Ethereum L1 users still compete for ordinary execution gas as before.

The difference is operationally significant. A decentralized exchange on an L2 can now support small swaps that were uneconomic on Ethereum mainnet. Games can settle more frequent actions. Social applications can subsidize transactions without carrying a large gas budget. Stablecoin transfers can compete with conventional payment rails on direct cost.

But cheap transactions also create a filtering problem. Fees are no longer a reliable barrier against spam, low-quality deployments, or circular incentive activity. A chain with near-zero transaction costs can post high activity while generating little durable economic value.

The relevant metric is not raw transactions. It is whether activity produces recurring fees, retained users, deep liquidity, and credible settlement demand.

Optimistic and ZK rollups solve the same problem differently

A layer 2 rollup example is incomplete without separating optimistic and zero-knowledge architectures. Both move execution off Ethereum. Both settle against Ethereum. Their key difference is how they establish that a transaction batch is valid.

Optimistic rollups, including Arbitrum, Optimism, and Base, assume submitted batches are valid unless someone proves otherwise. This is an economically rational default: most activity is expected to be valid, and computing a proof for every batch is not required upfront.

The trade-off is the fraud-proof challenge window. In the standard model, a withdrawal to Ethereum can involve a period of roughly seven days, allowing participants to dispute a fraudulent state claim.

ZK rollups, such as zkSync and Starknet, operate differently. They generate cryptographic validity proofs. Ethereum verifies the proof, rather than waiting for a potential challenger to demonstrate fraud. This provides a more direct path to finality once the proof is accepted.

The usual shorthand says ZK rollups are technically superior because they avoid long challenge windows. That is too clean.

ZK proof generation carries real computational overhead. Validity proofs improve verification properties, but they do not automatically make a network cheaper to operate. The cost of generating proofs, maintaining prover infrastructure, and supporting complex execution environments remains substantial.

Optimistic rollups, meanwhile, have benefited from mature EVM compatibility, established liquidity, and simpler deployment paths. Their economic position is not based only on legacy design. It is based on ecosystem depth.

For users, the distinction often appears at withdrawal time. For developers, it appears in tooling, execution compatibility, and deployment constraints. For capital allocators, it appears in liquidity depth and bridge exposure.

None of those concerns disappear because a protocol has a mathematically elegant proof system.

The sequencer is a throughput engine and a control point

Rollups are often described as decentralized networks. That description needs qualification.

A rollup may inherit Ethereum’s settlement security under defined conditions, but many production rollups still depend on centralized or tightly controlled sequencers for transaction ordering and immediate confirmations. The sequencer is efficient. It is also a point of operational concentration.

A sequencer can affect:

• Transaction inclusion. Users may face delays or temporary censorship if the primary sequencer fails or excludes activity.
• Ordering. Transaction ordering creates exposure to MEV extraction, liquidation priority, and arbitrage control.
• Fee policy. The sequencer typically mediates the fee market presented to users.
• Liveness. If the sequencer goes offline, the network may remain secure in a narrow settlement sense while becoming difficult or slow to use.
• Revenue capture. Cheap user fees do not mean the network lacks margin. The gap between fees collected and data-publication costs can accrue to the operator or governance structure.

This is the less promotional version of a blockchain infrastructure example. Scalability is not solely a throughput number. It is an allocation of power between validators, sequencers, provers, bridge contracts, and governance multisigs.

A rollup can be technically sound while still carrying governance concentration. It can have low fees while subsidizing those fees through token emissions or centralized operator revenue. It can offer fast soft confirmations while users wait materially longer for final settlement on Ethereum.

These distinctions become relevant during stress. Normal conditions hide infrastructure dependencies. Congestion, sequencer downtime, bridge incidents, and volatile liquidations expose them.

Low fees measure the cost of normal operation. They do not measure the cost of exiting during a system-wide liquidity event.

Why Base, Arbitrum, and Optimism hold roughly 83% of L2 DeFi TVL

By mid-2026, Base, Arbitrum, and Optimism collectively account for approximately 83% of total Layer 2 DeFi TVL. The concentration is not accidental. It reflects the same network effects seen in other financial infrastructure markets.

Liquidity clusters where execution is cheap, applications are available, wallets have support, and users already hold assets. Once that loop begins, smaller networks face a difficult acquisition problem. They can lower fees further, offer incentive programs, or launch a token. None of those automatically create deep, durable liquidity.

Arbitrum built early depth across DeFi, derivatives, lending, and decentralized exchange activity. Its advantage is not merely TVL. It is the density of applications and routes available for capital already on the network.

Optimism developed an ecosystem strategy around the OP Stack, extending its relevance beyond one chain. The thesis is that a shared technical framework can create interoperable network effects across many chains using common infrastructure.

Base entered with a different distribution advantage. Its connection to a large exchange ecosystem reduces user onboarding friction. That matters. The best bridge interface is often not a bridge interface at all; it is a familiar account, wallet, or fiat on-ramp that places users directly into the execution environment they need.

This creates a concentration loop:

1. More users create more liquidity.

2. More liquidity reduces slippage and improves execution.

3. Better execution attracts applications and market makers.

4. More applications create more transaction demand.

5. Higher activity improves the economics of batching and infrastructure investment.

Smaller rollups are therefore competing against more than technical specifications. They are competing against distribution, liquidity, developer mindshare, stablecoin inventory, and established bridge routes.

That competition can produce a liquidity sink. Capital enters a new chain for incentives, farms a yield differential, then leaves when emissions decline. The chain may retain transaction count but lose productive TVL. This is particularly visible where incentive-funded yields exceed organic protocol revenue by a wide margin.

A large advertised APY on a new L2 is not evidence of healthy infrastructure demand. It may be a transfer from token holders to mercenary liquidity. The drawdown starts when incentives normalize.

Throughput above 325 TPS is progress, not a final state

Ethereum’s approximate 15 TPS base-layer throughput is insufficient for global consumer-scale activity if every interaction must settle individually. Rollups materially change that constraint. Major networks can exceed 250 TPS, and system-wide Layer 2 throughput has passed 325 TPS.

Those figures are useful, but they need context.

TPS is a gross measure. It does not distinguish between a high-value stablecoin payment, a DEX arbitrage trade, a batch of account-abstraction operations, or low-cost spam. It also does not capture latency, sequencer uptime, withdrawal mechanics, state growth, or the cost of publishing data back to Ethereum.

Still, rollup throughput is economically meaningful because it changes what can be built without asking users to absorb L1 gas costs.

The next scalability frontier has several moving parts:

• More efficient blob usage. Rollups can continue optimizing how they compress and publish transaction data.
• Better proving infrastructure. ZK systems need cheaper, faster, and more distributed proof generation if they are to reduce operational overhead.
• Sequencer decentralization. Shared or distributed sequencing may reduce single-operator risk, though it can introduce latency and coordination costs.
• Cross-rollup interoperability. Liquidity spread across many networks is inefficient. Bridges and messaging layers attempt to solve this, but each additional connection expands the attack surface.
• Execution specialization. Some chains will optimize for general-purpose DeFi, others for payments, gaming, trading, or enterprise workflows. The market may not need dozens of undifferentiated general-purpose rollups.
• Future Ethereum data-availability upgrades. Proto-Danksharding is an intermediate step. The exact timing and final implementation path for full Danksharding remain unsettled.

This is where the optimistic narrative meets infrastructure reality. More chains do not necessarily create more usable capacity. If they split users and liquidity into isolated pools, aggregate throughput can rise while execution quality declines.

A trader does not care that ten chains each have low fees if the required stablecoin liquidity sits on only one of them. A developer does not benefit from an inexpensive chain if users cannot enter it without bridge friction. A protocol does not gain from higher TPS if MEV, fragmented collateral, and weak liquidity make market execution worse.

The practical test: settlement quality over headline fees

For anyone evaluating decentralized network examples, the right question is not, “Which rollup is cheapest today?” That is the lowest-value question in the stack.

The more useful tests are structural:

• Is transaction data posted to Ethereum in a way that permits reliable state reconstruction?
• Does the rollup use fraud proofs, validity proofs, or a more transitional architecture?
• Who controls the sequencer, upgrade keys, and bridge contracts?
• What happens if the sequencer stops producing blocks?
• How long does it take to exit to Ethereum through the canonical path?
• Is TVL supported by lending, trading, stablecoin usage, and organic fees, or by temporary emissions?
• Can liquidity move across the relevant application set without repeated bridge risk?
• Does the chain have a durable reason to exist beyond lower fees than its neighbors?

These are not academic questions. They determine whether a rollup is a scalable execution layer or an incentive-driven side venue.

The distinction becomes sharper during market stress. In a calm market, fragmented liquidity looks like choice. In a drawdown, it can become trapped collateral, delayed exits, thin order books, and wider spreads. Bridges become more heavily used precisely when their operational and smart-contract risks matter most.

Layer 2 scaling has solved a major part of Ethereum’s retail cost problem. It has not solved capital fragmentation, governance concentration, or the economic durability of every network that labels itself a rollup.

The yield sustainability verdict

Layer 2 rollups are the strongest current blockchain technology example because they demonstrate a workable scaling trade: Ethereum keeps final settlement and data availability, while specialized networks absorb high-volume execution.

Dencun and EIP-4844 made that trade economically viable at consumer scale. Fees falling by 80% to 99% changed the addressable market for on-chain activity. Major rollups charging roughly one to ten cents per transaction are no longer theoretical infrastructure.

But the market has already started selecting winners. Base, Arbitrum, and Optimism controlling about 83% of L2 DeFi TVL is a concentration signal, not a temporary footnote. Liquidity, distribution, and application density compound.

The sustainable yield is not the promotional APY attached to a new chain. It is the fee revenue and settlement demand generated after incentives fade. On that measure, the Layer 2 sector is real infrastructure—but much of its long tail remains an unproven liquidity sink.