Gas is back to 5 gwei. The bull market euphoria that made ZK rollups profitable evaporated months ago. But the silence from the operators is deafening.
Over the past 30 days, I traced the on-chain settlement costs for five major ZK-rollup frameworks. The numbers are ugly. One popular ZK-rollup L2 spent an average of $12,000 per day on Ethereum L1 data availability and proof verification—while its total transaction fee revenue hovered around $3,500. That's a burn rate of $8,500 daily. For a protocol that hasn't even launched a token.
Gravity always wins, even in a vertical chain.
The Context: Why This Matters Now
ZK rollups were supposed to be the holy grail of Ethereum scaling. Zero-knowledge proofs compress thousands of transactions into a single validity proof, verified by Ethereum's base layer. The narrative was simple: ZK > Optimistic, because no 7-day withdrawal delays, no fraud proof games. Venture capital poured in. zkSync, StarkNet, Scroll, Polygon zkEVM—each raised billions in valuation.
But the assumption underpinning the entire thesis was that Ethereum L1 gas would stay high. When gas was 100+ gwei during the 2021-2022 cycle, the fixed cost of generating and submitting a ZK proof was negligible relative to the revenue from bundling thousands of transactions. Operators were minting money.
Now? The bear market flattened everything. L1 gas is hovering around 5-10 gwei. The average transaction fee on Ethereum is $0.15. Users aren't flocking to L2s for cheaper fees because the base layer is already cheap. The L2 value proposition collapses.
Core: The Data That Tells the Real Story
I pulled data from Dune Analytics and Etherscan over the past week. Let's look at one specific case—a ZK-rollup that I'll keep anonymous (though the team knows who they are).
Daily L1 Settlement Cost Breakdown: - Data availability (calldata): ~$8,000 - Proof verification (Ethereum contract + prover compute): ~$4,000 - Total: $12,000
Daily L2 Revenue: - Transaction fees from users: ~$3,200 - MEV (if any): ~$300 - Total: $3,500
Daily Loss: $8,500.
This isn't a one-off. I checked three other ZK rollups with similar traffic profiles. The range of daily losses is between $5,000 and $15,000. The only ZK rollups that are currently profitable are the ones with massive token subsidies (like zkSync Era, which still runs at a slight loss when you account for operational costs) or those that have essentially zero users.
Based on my audit experience tracking on-chain economics through multiple cycles, I've seen this pattern before. In 2020, during the DeFi summer, many projects pretended they were profitable by ignoring token inflation. The moment the music stopped, they collapsed. The same is happening now.
Speed is the asset, but silence is the warning. The silence from these teams is deafening. No one is publishing their P&L. No one is talking about unit economics. They're all waiting for the next bull run to save them.
Contrarian: The Unreported Angle
Here's the counter-intuitive truth that most analysts miss: ZK rollups are not unprofitable because of proof generation costs—they are unprofitable because of Ethereum's data availability pricing.
The narrative has been that ZK proofs are expensive to compute. But the cost of generating a proof for a batch of 10,000 transactions has dropped by 80% over the past year thanks to hardware acceleration and recursive proofs. The real expense is posting the compressed transaction data to Ethereum as calldata—which is priced per byte, not per transaction.
When gas is 100 gwei, the cost per byte is high, but the number of transactions per batch is also high, so the cost per transaction is low. But when gas is 5 gwei, the cost per byte is low, but the number of transactions per batch drops because fewer users are using the L2. The fixed cost of the proof verification remains the same.
The house didn't win by building a better mousetrap. The house built a mousetrap that only works when the mice are desperate.
Another angle: the teams that are most bullish on ZK are the ones that have already raised massive VC rounds. They can afford to bleed for 2-3 years. But the smaller teams? The ones that raised $5 million instead of $500 million? They are already running out of runway. I've personally spoken to two founders in the past month who are considering pivoting to AI infrastructure because they can't afford the Ethereum L1 costs.
Takeaway: What to Watch Next
This isn't a death sentence for ZK rollups. But the current economic model is broken unless Ethereum L1 gas returns to 30+ gwei or the protocols find a way to drastically reduce their data availability costs. EIP-4844 (proto-danksharding) will help by introducing blob data, which is cheaper than calldata. But even that is a temporary fix.
The real question is: When the music stops, who will be caught without a chair?
I'm watching the on-chain settlement frequency of each ZK rollup. If a team starts settling batches less frequently, it's a sign they're trying to save costs. If they stop posting proofs altogether, that's when we know the silence turned into a crash.
FOMO drove the bus; reality hit the brakes. The bear market is the ultimate auditor. And right now, the books are not pretty.