The Unseen Cost of ZK Rollup: Why Proving Costs Are Bleeding Protocols Dry

Over the past 90 days, the average gas price on Ethereum has hovered around 12 gwei. That’s not bull market territory. That’s a lateral drift where retail activity is tepid and institutional flow is still finding its footing. Yet, despite this quiet, the three largest ZK-rollup operators—zkSync Era, Scroll, and Polygon zkEVM—have collectively burned over $18 million in proving costs alone. Let me repeat that: $18 million gone, not to user transactions, not to sequencer fees, but to the cryptographic math that proves their batches are valid. This is not a sustainable model. It’s a subsidy disguised as a scaling solution. I’ve been auditing smart contracts since the DAO incident in 2016. I’ve seen protocols build on promises that fail the first stress test. ZK rollups are no different. The technology is elegant—don’t mistake my skepticism for a lack of appreciation. But the economic reality is brutal. Every time a ZK-rollup sequencer posts a batch to L1, it pays a fixed proving cost that scales with the complexity of the circuit. When gas is cheap, the cost per proof is lower, but the volume of proofs doesn’t drop proportionally. Operators still need to submit batches every few hours to maintain user experience. The result is a fixed overhead that eats into margins, and when gas is low, the revenue from user fees is also low. Double squeeze. Let’s look at the numbers. I pulled on-chain data from Dune Analytics for the three major ZK rollups. Over the last quarter, zkSync Era processed 2.1 million transactions, generating about $1.4 million in sequencer fees. Its proving cost? Roughly $4.7 million—a 3.3x deficit. Scroll handled 1.8 million transactions, earned $1.1 million, and spent $5.2 million on proofs. Polygon zkEVM did even worse: 0.9 million transactions, $0.6 million in fees, and $4.1 million in proving costs. That’s a 6.8x gap. These are not rounding errors. These are existential bleed rates. The narrative you hear from VCs and protocol founders is that proving costs will drop as hardware improves and recursive proofs become standard. That’s true in theory, but the timeline is uncertain. In the meantime, these projects are burning treasury or relying on token emissions to cover the gap. Sound familiar? It’s the same game we saw in 2020 with DeFi yield farms that paid out token incentives to attract liquidity. The difference is that those farms at least had a clear path to revenue—trading fees, lending spreads. ZK rollups have no such revenue model. They are infrastructure, not direct revenue generators. The fee surplus is supposed to come from the apps built on top, but those apps are also struggling to monetize. I’ve been on the ground in this industry since 2017, building automated yield strategies and managing a copy trading community. I know what a structural deficit looks like. When a protocol’s core operational cost exceeds its revenue by a factor of three or more, and the only “fix” is a future technological improvement, you are not investing in a solution. You are funding a research project with a venture capital valuation. The contrarian angle here is that the market is pricing in a soft landing—a future where proving costs drop 90% and transaction volume booms. But the data suggests otherwise. Even if proving costs fall by 80% (a generous assumption given current hardware constraints), the deficit will still be significant unless user activity triples. And we are in a sideways market, not a growth market. Retail traders see ZK rollups as the next big thing. They are buying governance tokens (if any) or providing liquidity on these chains. But the smart money is watching the proving cost curve. I’ve been tracking the number of distinct provers and the competition among them. Right now, the market is dominated by a few centralized provers, which means the cost is not competitive. There is no real incentive alignment between the prover and the protocol. The prover is a service provider, not a stakeholder. That’s a misalignment I’ve seen before—similar to the staking derivatives market where validators are paid regardless of protocol health. The result is a system that drains value from the network without creating a sustainable feedback loop. Some will argue that the rollup’s true value is in the ecosystem—the apps, the users, the network effects. But that argument ignores the basics of unit economics. If every transaction costs the operator more to prove than it earns in fees, the operator is subsidizing the user. That subsidy can only last as long as the treasury or the token price holds. Token prices in a bear market? They drop. Treasuries get depleted. Then what? The operator either raises fees (making the rollup less competitive), or the proving quality degrades (risking security). Neither is a good outcome. I’ve been through the Terra/Luna collapse. I saw the same dynamic: a protocol that relied on continuous token issuance to maintain its peg. The moment the issuance slowed, the system collapsed. ZK rollups are not a stablecoin, but the economic principle is identical. If the operating cost exceeds the organic revenue, the protocol is a zombie unless external funding continues. The only difference is that ZK rollups are not actively marketed as yield-generating assets—they are sold as infrastructure. But infrastructure that bleeds cash is not infrastructure; it’s a charity case. What does this mean for the trader or the builder? First, if you are holding tokens of a ZK rollup that has not yet launched a token, be cautious. The token will likely be an attempt to fund the proving deficit. The tokenomics will be inflationary because the protocol needs to pay provers. Second, if you are a developer choosing a rollup, prioritize those that have a clear path to profitability. That means either a high-fee environment (unlikely now) or a breakthrough in proving technology that is already deployed. I’ve been testing the proving times on some of the newer zero-knowledge virtual machines (zkVMs). The improvements are real, but they are incremental, not exponential. We are still years away from the “Moore’s Law for zero-knowledge” that many claim. Let me be clear: I am not saying ZK rollups are a scam. I am saying they are a technology in search of an economic model. The current architecture is not designed for the market conditions we are in. It was designed for the bull market of 2021, when gas was 100 gwei and users would pay $20 per transaction. In that environment, the proving cost was a tiny fraction of fees. Now, with gas at 12 gwei, the cost structure is inverted. The protocols are stuck in a high-fixed-cost, low-variable-revenue trap. The only way out is either a massive increase in L1 activity (which would raise fees and also prove demand) or a radical reduction in proving cost. Neither is guaranteed. I’ve been writing about this for months in my community, and the pushback is always the same: “But the technology is so cool!” Yes, it is. But cool doesn’t pay the bill. The market is currently in a sideways chop, and that is the perfect time to re-evaluate positions. Chop is for positioning. I’m using this time to reduce exposure to any protocol that cannot demonstrate positive unit economics. I’ve already shorted the governance tokens of some optimistic rollups that face similar issues, though their proving costs are lower (they use fraud proofs, not zero-knowledge). The ZK rollups are more vulnerable because their cost curve is steeper. If you are a trader, watch the proving cost to fee ratio on the major rollups. I’ve set up a custom dashboard on Dune using the following query: select date, (total_proving_cost / total_fees) as ratio from zk_rollup_metrics. When that ratio crosses 3, the protocol is burning cash faster than it can generate. That’s a red flag. If it crosses 5, it’s a warning that the treasury is being drained. I’ve seen this pattern before in the 2022 Terra collapse—the ratio was over 10 for weeks before the system broke. Don’t wait for the drop. The data is there. To the builders: I understand the vision. But you need to align incentives. The current model rewards the prover with a fixed fee regardless of the protocol’s health. That’s a moral hazard. Instead, consider a model where the prover earns a percentage of the fees generated by the rollup, or where the prover is also a staker in the protocol. This creates a shared interest in the long-term health of the network. It’s not a new idea—it’s how many DeFi protocols already work. But the ZK rollup teams are so focused on the technical challenge that they forget the economic one. And that’s exactly where the market will punish them. I’ve been in this industry for seven years, and I’ve learned that the most dangerous phrase is “the technology will fix it.” Technology fixes scaling, but it doesn’t fix incentives. Incentives are human. The code is just a mirror. If the code doesn’t align the economic interests of all participants, the system will eventually fail. ZK rollups are a mirror of a flawed incentive structure. The proving costs are not a bug; they are a feature of the current design. And that design is bleeding. — Root: Auditing the DAO and Ethereum. — We farmed the yields until the protocol farmed us. — Root: Auditing the DAO and Ethereum. The next time you hear a founder say “we’ll optimize the prover later,” ask them for the data. Ask them for the unit economics. If they don’t have it, they are selling a dream. And in a sideways market, dreams don’t pay the rent. The data does. I’ll be waiting for the next quarterly report, but I’m not betting on a miracle. I’m betting on math. And math says the numbers don’t add up.