Are ZK Rollups Finally Catching Up to Optimistic Rollups?
ZK rollups have closed the gap on cost and finality, but optimistic rollups still lead in TVL and ecosystem depth. Here's where the race actually stands today.
Key takeaways
- ZK rollups and optimistic rollups solve the same problem with fundamentally different trust models. One relies on cryptographic proof, the other on economic incentives and a challenge window.
- ZK rollups have already won on finality and proving cost, while optimistic rollups still lead on throughput, ecosystem depth, and total TVL.
- The ecosystem gap is harder to close than the technical gap. Liquidity gravity and developer tooling compound over time in ways that math alone can't shortcut.
- Neither architecture is universally superior. The right choice depends on use case, not on which camp is "ahead."
ZK rollups have largely caught up to optimistic rollups on cost and finality speed. Proving costs have dropped over 90%, EVM compatibility is near-parity, and withdrawals that once took seven days now clear in under an hour. The gap on throughput and ecosystem depth, however, remains real, and that's what's keeping optimistic rollups dominant today.
The technical story and the market story are moving at different speeds. Seizing the difference is what actually helps you decide where to build or deploy capital.
The Race So Far – What "Catching Up" Actually Means
| At a glance: Optimistic rollups got there first. Networks like Arbitrum, Base, and OP Mainnet launched with near-perfect EVM compatibility, deep developer tooling, and a straightforward security model that was easy to reason about. That head start compounded quickly – protocols deployed, users followed, and liquidity concentrated. |
Optimistic rollups currently hold approximately 80% of total Layer 2 TVL, driven by years of ecosystem compounding, deep protocol integrations, and superior EVM compatibility that attracted developer gravity first.
ZK rollups, meanwhile, had a harder early story.
- Proving was expensive.
- EVM compatibility required significant engineering workarounds.
- Withdrawal times were fast in theory, but proof generation latency made real-world UX inconsistent.
"Catching up" meant closing those specific gaps – the concrete technical disadvantages that made ZK rollups a hard sell for builders who had a simpler option.
That gap is now closing fast. The L2 landscape splits cleanly into two camps: optimistic rollups dominate TVL and ecosystem maturity, while ZK rollups lead on finality time and proof economics.
For a deeper look at how each architecture works under the hood, see Rollups Explained: Optimistic vs. ZK Rollups in Ethereum Scaling.
How ZK Rollups Have Closed the Gap
| In short: On the three dimensions that once held ZK rollups back – proving cost, EVM compatibility, and withdrawal finality, the gap has either closed entirely or shrunk to the point where it's no longer a deciding factor for most builders. |
Proving Costs: From Barrier to Non-Issue
The cost of generating zero-knowledge proofs was once ZK rollups' most visible weakness. Proof generation required significant compute – expensive hardware, long latency, and high operating costs that made ZK rollups economically uncompetitive for most use cases.
That has changed dramatically.
Specialized hardware, such as ZK-ASICs and optimized GPU clusters, has dropped proving costs by over 90% compared to a few years ago. ZK rollup transaction costs are now consistently under $0.01 per transaction for high-volume chains, making them competitive even for micro-transactions.
Post-EIP-4844, the data availability side of the equation also compressed significantly. After EIP-4844 landed blob data, ZK rollup fees for simple transfers dropped to the single-digit cents range on most major rollups, roughly a 10x reduction from pre-4844 levels, and typically 50–100x cheaper than Ethereum L1.
Cost is no longer a meaningful differentiator between the two architectures.
EVM Compatibility: No Longer a Dealbreaker
Early ZK rollups required developers to either rewrite contracts in custom languages (Cairo for Starknet) or accept limited compatibility with the Ethereum toolchain. This was a real friction point. Optimistic rollups, which simply re-executed EVM transactions off-chain, had an obvious advantage.
Today, EVM equivalence is no longer a selling point. Type-1 zkEVMs like Taiko achieve complete bytecode-level equivalence with Ethereum L1, meaning Solidity contracts deploy without modification.
The cost gap has narrowed further: EIP-4844 and modular DA layers reduced calldata costs for optimistic rollups, while ZK proving costs fell due to hardware and algorithmic improvements. In the current landscape, fee differences are often driven more by sequencer design and market demand than by the fundamental rollup type.
Developers choosing between architectures today are largely making a tradeoff between ecosystem maturity and architectural properties, not between "works" and "doesn't work."
Finality Now Actually Usable
This is where ZK rollups have a native, structural advantage, and it's finally being expressed in practice.
Optimistic rollups require a seven-day fraud-proof challenge window for native withdrawals to Ethereum mainnet. The window exists because the network needs time for challengers to detect and dispute invalid state transitions. Without it, the security model breaks.
A trader who needs to return capital from an optimistic rollup to Ethereum L1 faces two options:
- wait seven days via native withdrawal,
- or pay a bridge fee of 0.05–0.20% for a sub-5-minute exit via protocols like Across or Stargate.
ZK rollups sidestep this entirely. Systems like zkSync Era and Polygon zkEVM typically process a full withdrawal in 15 to 45 minutes, depending on proof batching frequency. The validity proof is the correctness guarantee. Once Ethereum verifies it, the batch is final. No waiting for fraud proofs to time out.
For institutional participants running settlement workflows where same-day capital clearance matters, this difference is operationally significant. For retail traders with multi-day positions, it's meaningful but not decisive.
ByteByByte's Take
Optimistic rollups say: assume everyone is honest, punish bad actors after the fact. ZK rollups say: prove correctness upfront, no assumptions required. The seven-day window is the price of the optimistic model. What's changed is that ZK rollups have made their alternative practical enough to matter. Proving latency has dropped from tens of minutes to seconds on modern zkEVM systems. The architectural advantage was always there; the engineering has finally caught up to it. That shift matters most for use cases where capital efficiency and settlement certainty are the core product, including institutional settlement, fast perpetuals, and compliance-sensitive DeFi, not for the majority of retail DeFi activity where a five-minute bridge workaround is perfectly acceptable.
Where ZK Rollups Still Lag Behind
| In short: Despite closing the technical gaps on cost and compatibility, ZK rollups still trail on throughput, ecosystem depth, and prover decentralization. And these are harder to fix than the proving cost problem was. |
Throughput: The Proving Overhead Problem
Despite all the progress on cost and compatibility, ZK rollups still face a hard ceiling on raw transaction throughput, and it comes directly from the proving mechanism itself.
Generating a validity proof for every batch is computationally intensive. That overhead limits how frequently ZK rollups can submit batches and how many transactions fit into each one.
Proof generation overhead currently caps ZK rollup TPS well below optimistic rollup benchmarks. zkSync Era delivers approximately 28 TPS; Starknet approximately 19 TPS. For comparison, Arbitrum One achieves 40–62 TPS, and OP Mainnet reaches up to 89 TPS under load.
This gap matters most for high-frequency applications: trading platforms processing thousands of orders per second, on-chain gaming, or consumer apps expecting mass concurrent users. For standard DeFi interactions, the current ZK throughput ceiling is rarely the binding constraint.
Hardware acceleration and recursive proof techniques are expected to raise ZK throughput ceilings materially, but the gap is real today.
Ecosystem Depth & Liquidity Gravity
This is the hardest gap to close because it's a compounding one.
When GMX, Aave, Uniswap, and Curve all deployed their primary liquidity on Arbitrum One, they brought their user bases, governance participants, and integrating protocols with them. Each subsequent protocol that deployed on Arbitrum One found pre-existing counterparty depth, stable liquidity pools, and composable infrastructure already in place.
Arbitrum and Base together hold approximately 77% of all L2 DeFi TVL. No other single chain exceeds $6 billion: OP Mainnet sits at $5.6B, zkSync Era at $4.1B, Linea at $3.4B, Scroll at $2.1B, and Starknet at $1.5B.
Better proving math doesn't move liquidity. What moves liquidity is a compelling reason for the protocols already on Arbitrum to redeploy capital elsewhere. That reason hasn't fully materialized yet for the majority of DeFi activity.
Decentralization of Provers
This is the most underappreciated gap in the ZK rollup story right now.
ZK rollups are often framed as more trust-minimized than optimistic rollups because their security relies on cryptographic proofs rather than honest watchers. That's true at the proof level, but the infrastructure running those proofs tells a different story.
Most production ZK rollups run a centralized sequencer and a permissioned prover network. Decentralizing both is the active roadmap for every major rollup. L2Beat's risk framework scores each rollup on these dimensions.
Most ZK rollups use a single prover operated by the foundation. While the math ensures they cannot steal funds, they can still experience downtime or censor transactions. Optimistic rollups face a similar sequencer centralization risk, but their Stage 2 decentralization efforts are currently further along in the governance cycle.
The cryptographic guarantee is real. The centralization risk sits one layer up in the infrastructure generating the proofs.
The Honest Scorecard
Neither architecture has won. The right framework is use-case matching, not picking a winner.
Dimension | Optimistic Rollups | ZK Rollups | Verdict |
| Native withdrawal time | ~7 days | 15 min – 1 hour | ZK wins |
| Transaction fees | $0.02–$0.06 | $0.01–$0.06 | Roughly equal |
| Throughput (TPS) | 40–89 TPS | 19–28 TPS | Optimistic wins |
| EVM compatibility | Mature | Near-parity | Near-tie |
| Ecosystem/TVL | ~80% L2 share | ~20% L2 share | Optimistic wins |
| Security model | Economic incentives + fraud proofs | Cryptographic validity proofs | ZK stronger on paper |
| Prover decentralization | N/A | Mostly centralized | Gap remains |
Quick decision guide:
- Building or trading where the liquidity already is → Optimistic rollups (Arbitrum, Base)
- Need fast, native settlement finality → ZK rollups
- Institutional settlement, compliance-sensitive DeFi → ZK rollups
- High-frequency apps needing maximum TPS → Optimistic rollups for now
- Privacy-preserving transactions → ZK rollups (architectural alignment)
Will ZK Rollups Eventually Overtake Optimistic Rollups?
| In short: On technical metrics, very likely yes. The throughput gap is a solvable engineering problem. On ecosystem dominance, that's less certain, and depends more on where institutional adoption goes than on any proof system improvement. |
Hardware acceleration, recursive proofs, and optimized proving circuits are on a clear downward trajectory. As ZK proof generation costs continue to fall and batch frequencies increase, the throughput gap between ZK and optimistic rollups will narrow, and the selection criteria will shift further toward ecosystem composition and governance maturity rather than underlying architecture type.
The ecosystem gap is structurally different. As compliance frameworks for on-chain tokenized assets mature across major jurisdictions, ZK rollup TVL growth may accelerate beyond current projections, potentially narrowing the structural gap with optimistic rollup dominance established during the EVM-compatibility era.
The most credible path to ZK dominance runs through institutional adoption. The ZK cohort has delivered on cryptographic finality promises and is now pivoting toward institutional use cases that optimistic chains cannot serve without fundamental architectural changes.
That's a real growth vector. It's just a different race than the one most people are watching.
Sources and Further Reading
- Ethereum Foundation – "Rollups Documentation" https://ethereum.org/en/developers/docs/scaling/optimistic-rollups/
- Ethereum Improvement Proposals – "EIP-4844: Shard Blob Transactions" https://eips.ethereum.org/EIPS/eip-4844
- Vitalik Buterin – "An Incomplete Guide to Rollups" https://vitalik.eth.limo/general/2021/01/05/rollup.html
- zkSync Era – "Official Documentation" https://docs.zksync.io
- Starknet – "Official Documentation" https://docs.starknet.io
- Arbitrum – "Official Documentation" https://docs.arbitrum.io
FAQs About ZK Rollups Catching Up to Optimistic Rollups
They're already coexisting and differentiating by use case. There's no strong reason to expect one to fully displace the other. The architectures have different strengths, and different applications are optimizing for different tradeoffs.
