Ethereum
Arbitrum
Optimism
Ethereum's Scalability Problem in Practice
At the peak of DeFi Summer in 2020, Ethereum gas fees — the cost to execute a transaction — regularly exceeded $50 for a simple token swap. During the NFT boom of early 2021, complex smart contract interactions could cost over $200. At these prices, Ethereum was economically accessible only to users moving substantial sums. A person trying to send $100 of crypto could pay fees that represented half the transaction's value.
This was not a temporary spike to be endured. It was a structural consequence of Ethereum's architecture: a single global blockchain where every node must process every transaction, with block space priced by auction. When demand for block space exceeds supply, fees rise until demand is priced out. Ethereum's bottleneck was real and its effects were exclusionary.
The long-term solution — sharding Ethereum's execution layer — remained years away on the research roadmap. In the interim, the ecosystem converged on a different approach: Layer 2 networks.
Layer 2 solutions execute transactions outside Ethereum's main chain (the Layer 1) and then publish compressed summaries of those transactions back to Ethereum for security. Users get fast, cheap transactions; Ethereum provides the underlying security and settlement finality. The combination resolved the fee crisis in a way that base-layer scaling could not deliver quickly.
Optimistic Rollups: Assume Validity, Challenge If Wrong
Arbitrum and Optimism are the leading optimistic rollup networks. The mechanism is elegant: transactions are executed by a sequencer off-chain, and the resulting state transition is posted to Ethereum with an assertion that it is valid. The "optimistic" part means Ethereum accepts this assertion by default.
During a challenge window (typically seven days), any watcher can submit a fraud proof demonstrating that the posted state transition is invalid. If the fraud proof succeeds, the invalid state is rolled back and the dishonest sequencer loses their staked collateral. If no challenge succeeds during the window, the state is finalized.
This design achieves high throughput because only the state commitments and transaction data travel through Ethereum — not the computational work of executing each transaction. Arbitrum processes thousands of transactions per second and settles them on Ethereum for fees measured in cents rather than dollars.
Coinbase launched Base, an optimistic rollup built on the OP Stack (Optimism's open-source rollup framework), in 2023. Within months, Base had grown to become one of the highest-transaction-volume chains in the ecosystem, demonstrating that a well-distributed company building on Layer 2 could rapidly acquire users who had never previously used crypto. Base's cheap fees — often below $0.01 per transaction — enabled micropayment use cases that were economically impossible on Ethereum L1.
ZK Rollups: Prove Validity Mathematically
The alternative approach uses zero-knowledge proofs. Rather than assuming transactions are valid and relying on fraud challenges, ZK rollups generate a cryptographic proof — a validity proof — that mathematically demonstrates the correctness of a batch of transactions. Ethereum verifies this proof; if it verifies, the state transition is accepted immediately, with no challenge window.
This is the approach taken by zkSync Era and Polygon's zkEVM, among others. The advantage of immediate finality (no seven-day challenge window) is significant for use cases requiring fast settlement. The disadvantage has historically been the computational cost of generating ZK proofs — but advances in proof systems (PLONK, STARKs, Groth16) and specialized hardware have dramatically reduced this cost.
StarkNet uses STARKs (Scalable Transparent Arguments of Knowledge), which have larger proof sizes but no trusted setup requirement. zkSync uses SNARKs, which have smaller proofs but require a trusted setup ceremony. The cryptographic trade-offs are subtle but the user-facing experience is similar: fast, cheap transactions with Ethereum-grade security.
Fee Reduction: Orders of Magnitude
The impact on transaction costs has been transformative. A token swap on Ethereum L1 during peak congestion might cost $10-50. The same swap on Arbitrum or Optimism costs $0.05-0.50. On zkSync or Base, costs drop further, often below $0.05.
EIP-4844, implemented in March 2024, introduced "blob" transactions — a new data format optimized for rollup data publication — and reduced L2 data costs by an additional 10-100x. After EIP-4844, transactions on leading L2 networks regularly cost fractions of a cent, making crypto economically accessible to anyone regardless of transaction size.
This fee reduction unlocked use cases that were previously economically absurd. Micro-tipping content creators, paying per-message for AI services, streaming payments where fractions of a cent per second flow to service providers, gaming economies where in-game items have sub-dollar values — all become viable when transaction costs approach zero.
User Growth: L2 Summer
The combined effect of low fees and Ethereum-equivalent security drove a user migration to Layer 2 networks throughout 2023-2024. By late 2023, the combined transaction volume on major L2 networks exceeded Ethereum L1. By early 2024, it was substantially higher.
Arbitrum's total value locked grew to over $10 billion. Base attracted hundreds of thousands of daily active users, becoming the home of a new social media protocol (Farcaster) and consumer crypto applications that had never found traction on expensive L1 networks. Optimism's Superchain vision — a network of rollups sharing security and a common token — attracted major partners including Coinbase (Base), OP Labs, and Worldcoin.
The growth was not just in financial applications. Gaming projects launched on L2 networks, issuing NFTs and in-game currencies at costs that made the economics of gaming viable. Social applications issued on-chain attestations for identity and reputation at costs below a cent. The developer ecosystem concentrated on L2s, with Ethereum L1 increasingly serving as a settlement and security layer rather than an application execution environment.
Sequencer Revenue and the Economics of L2s
Running a Layer 2 network is a profitable business. The sequencer — the operator that orders and publishes transactions — collects user fees and pays Ethereum for data publication costs. The difference is sequencer revenue.
During 2023-2024, leading L2 sequencers generated hundreds of millions of dollars in revenue annually. Arbitrum's sequencer generated estimated revenues exceeding $100 million in 2023. Base's sequencer (operated by Coinbase) generated substantial revenue that Coinbase disclosed as a new business line on its earnings calls.
This revenue is significant because it funds protocol development and creates sustainable business models for L2 teams without ongoing token inflation. It also raises governance questions: sequencers are currently centralized single operators. If a sequencer censors your transaction or goes offline, you can force-include your transaction through L1 — but this friction is meaningful. Decentralizing sequencers — distributing the ordering function across multiple parties — is an active area of research and development.
The Multi-Layer Ethereum Economy
The L2 ecosystem has produced a layered economic structure:
Ethereum L1 functions as a settlement and data availability layer. Its fee market is driven by rollup data publication and high-value transactions where L1 finality is required. L1 validators earn from this demand.
L2 networks function as the primary user-facing execution environment. Their fee markets are competitive and cheap. Sequencers earn the spread between user fees and L1 publication costs.
Application protocols — DeFi, NFT marketplaces, social apps — operate on L2 networks and compete for users through product quality and token incentives.
This multi-layer structure echoes the layering of traditional internet infrastructure: physical networks (analogous to L1), protocols (TCP/IP, analogous to the Ethereum protocol), and applications (HTTP, web apps, analogous to DeFi and consumer crypto apps). The analogy suggests that L2 networks may be the layer where most user-facing value is created and captured, with L1 serving as foundational infrastructure — valuable but less directly visible to end users.
What Comes Next
The L2 landscape is evolving rapidly. ZK proofs are becoming cheaper and more capable, enabling ZK-EVM equivalence — the ability to prove arbitrary EVM execution, not just simplified subsets. Full danksharding will eventually expand Ethereum's blob capacity, further reducing L2 costs. Interoperability between L2 networks — currently fragmented with assets and liquidity siloed — is an active development focus.
The core achievement is already substantial. Layer 2 solutions transformed Ethereum from a network accessible primarily to wealthy users into a global financial infrastructure where transaction costs no longer determine participation. The programmable financial system described in Ethereum's whitepaper has become practically accessible — and the economy it hosts is still growing.