Ethereum's Rollup-Centric Roadmap: Where Layer 2 Scaling Actually Stands

When Vitalik Buterin published "A rollup-centric Ethereum roadmap" in 2020, it formalized something many in the community already suspected: Ethereum's base layer would not scale on its own, and the path forward ran through Layer 2. Four years later, rollups are no longer a future promise — they are live infrastructure processing billions of dollars in transactions. But the picture is messier, more promising, and more technically interesting than either the boosters or the skeptics tend to admit.

The Case Rollups Were Built On

Ethereum's base layer processes roughly 15–30 transactions per second, constrained by block size limits designed to keep node operation decentralized. Increasing that limit directly — as some chains have done — risks centralizing validation to operators who can afford the hardware, weakening the censorship resistance that gives Ethereum its security guarantees.

Rollups sidestep this tradeoff. The key insight is that execution and settlement can be separated: rollups execute transactions off-chain at high throughput, then post compressed transaction data and cryptographic proofs back to Ethereum's base layer for final settlement. The base layer does not need to re-execute every transaction — it only needs to verify that the rollup's claimed state transition is valid.

Two verification approaches dominate: optimistic rollups assume transactions are valid unless challenged, relying on a fraud-proof window (typically 7 days) during which anyone can submit a proof of invalid state. ZK-rollups generate cryptographic validity proofs for every batch, allowing near-instant finality without a challenge period. Both inherit Ethereum's security model: fraud proofs and validity proofs both ultimately anchor to Ethereum's consensus layer.

Where Rollup Adoption Actually Is

Arbitrum One and Optimism are the two largest optimistic rollups by total value locked and transaction volume. Between them, they regularly process more transactions per day than Ethereum's mainnet. Arbitrum's ecosystem in particular has attracted significant DeFi activity — GMX, Camelot, and dozens of smaller protocols operate primarily on Arbitrum.

On the ZK side, zkSync Era and Polygon's zkEVM have both launched on mainnet, bringing EVM-compatible ZK execution to production. StarkNet uses Cairo rather than the EVM, which gives it more performance headroom but requires developers to rewrite contracts. Base, Coinbase's Optimism-stack rollup, has attracted substantial retail activity and demonstrated that brand-backed rollup launches can bootstrap liquidity quickly.

The total value locked across major L2s has grown from near zero in early 2021 to tens of billions of dollars. Transaction fees on rollups run between 1–10% of equivalent Ethereum mainnet fees for most operations. For small transactions that were effectively impossible on mainnet due to gas costs, rollups have genuinely opened up new use cases.

What EIP-4844 Changed

The most important infrastructure upgrade for rollups in recent years was EIP-4844, also known as proto-danksharding, which shipped with the Dencun hard fork in March 2024. Before 4844, rollups posted their compressed transaction data as calldata to Ethereum — expensive, permanently stored, and subject to the same gas market as regular transactions.

EIP-4844 introduced "blobs" — a new data type specifically designed for rollup data posting. Blobs are cheaper than calldata by design (a separate fee market), pruned after approximately 18 days (reducing long-term storage requirements on full nodes), and sized to give rollups roughly 3–10x more data capacity per block.

The effect on rollup fees was immediate and substantial. Average L2 transaction costs dropped by 80–90% the week after Dencun, and several rollups temporarily achieved sub-cent transaction fees. Blob space is not unlimited — sustained demand pushes blob fees up — but the base cost floor dropped significantly, and the architecture creates a clear path to further capacity increases via full danksharding.

The Fragmentation Problem

Rollup proliferation has created a genuine problem: liquidity and user attention are fragmented across dozens of chains, each with its own bridge, its own ecosystem of applications, and its own security model. Bridging assets between L2s requires either going through Ethereum mainnet (slow and expensive) or using third-party bridges (faster but introducing additional trust assumptions).

Cross-rollup composability is largely broken. A DeFi position on Arbitrum cannot interact atomically with one on Optimism. Smart contracts cannot easily call contracts on other L2s in a single transaction. The user experience of navigating a multi-rollup world — managing multiple networks in a wallet, understanding different bridging risks, tracking assets across chains — is genuinely bad.

The Superchain concept — Optimism's vision of interoperable rollups sharing a common sequencer and messaging protocol — is one attempt to solve this within a single rollup stack. The Base-Optimism relationship demonstrates this model: both chains share infrastructure and can pass messages to each other more efficiently than cross-stack communication allows. Polygon AggLayer and zkSync's Elastic Chain are alternative approaches to the same problem.

Sequencer Centralization and Its Implications

Most rollups today operate with a single, centralized sequencer — a server run by the rollup team that orders and batches transactions before posting them to Ethereum. This is a known temporary compromise: centralized sequencers enable faster iteration, lower latency, and simpler operations during the bootstrapping phase.

The problem is that a centralized sequencer can censor transactions, extract MEV with no accountability, and represents a single point of failure. If Arbitrum's sequencer goes offline, the chain stops — users can still force-include transactions via Ethereum mainnet, but at mainnet fees and with significant delay.

Decentralized sequencing is on every major rollup's roadmap but has proven technically complex and politically contentious — sequencer revenue is substantial, and decentralizing it means distributing that revenue to token holders or validators rather than the rollup team. Espresso Systems and Astria are building shared sequencer infrastructure that multiple rollups could use, reducing the cost of decentralization by spreading it across chains.

What the Next Phase Looks Like

Full danksharding — the long-term vision for Ethereum's data availability layer — would increase blob capacity by roughly 64x, from 3 blobs per block to hundreds. Combined with improved ZK proof systems and decentralized sequencing, this would allow rollups to handle millions of transactions per second in aggregate, anchored to a base layer that most of the world can run on commodity hardware.

The timeline for full danksharding is measured in years, not months. Ethereum's development process is methodical and prioritizes correctness over speed. The intermediate step — PeerDAS, which distributes blob data across more nodes without requiring every node to download everything — is in active development and may ship in 2025.

For users and developers, the practical takeaway is that rollups are production infrastructure today, not a future promise. They are cheaper, faster, and more capable than Ethereum mainnet for most use cases. But the cross-chain fragmentation problem is real, the centralized sequencer issue has not been resolved, and the security assumptions of different rollup designs vary enough that understanding what you are using still matters.