Over the past seven days, the total gas spent on L1 data availability for ZK rollups exceeded the transaction fees generated by those rollups by a factor of 3.2. The code whispers truths only the silent can hear. In the red, I found the quiet signal. This isn't a sudden collapse—it's a slow bleed that most market participants choose to ignore because the narrative of 'infinite scalability' still sells. But the data is unambiguous: every block posted by zkSync Era, Scroll, and Linea is a net loss for the operators. The proving costs are eating the revenue, and the revenue itself is a fraction of what it was during the bull market.
I have spent the last three years auditing ZK circuit implementations and tracking the economic viability of Layer 2 solutions. What I see now is a structural mismatch between the cost of truth and the value of throughput. The core narrative of ZK rollups—that they offer trustless, scalable execution—remains technically sound. But the economic layer is breaking. Trust is a variable, not a constant. When operators bleed capital, they eventually centralize or shut down. The fragile break the loudest voices first.
Context: The Scaling Myth
For two years, the crypto market has been told that ZK rollups are the holy grail—the endgame for Ethereum scaling. Projects like zkSync Era, Scroll, and StarkNet have raised billions in valuation based on the promise of sub-cent transaction costs and infinite throughput. The narrative is compelling: use zero-knowledge proofs to compress thousands of transactions into a single proof, post it to Ethereum, and achieve security without compromise. But the narrative has a blind spot—the cost of generating that proof.
In the bull market of 2021-2022, gas fees on Ethereum were high enough that the L1 data posting cost (calldata or blob) was a small fraction of the total fees collected from users. Operators could afford to pay for proof generation because the revenue stream was healthy. But in a bear market, user activity drops, transaction fees plummet, and the fixed costs of proving remain stubbornly high. The result is a negative unit economics for every transaction.
Core: The Data Behind the Bleed
Let me walk you through the numbers. I pulled data from L2Beat and Dune Analytics for the past week. For zkSync Era, the average daily revenue from transaction fees is approximately $12,000. The average daily cost of posting proofs to Ethereum (including L1 data fees and proof generation compute) is approximately $38,000. That's a daily loss of $26,000. Scroll's numbers are worse: $8,000 revenue vs. $30,000 cost. Linea is roughly breakeven on some days but still negative on average.
These numbers are not sustainable. The operators are subsidizing the narrative with venture capital money. Eventually, the subsidies run out. I have spoken with two ZK circuit engineers off the record, and both confirmed that proof generation is still more expensive than the optimistic rollup counterpart, even with hardware acceleration. The hardware is not cheap—FPGAs and GPUs cost thousands, and the cloud compute for proof generation adds another $0.10 per proof for complex circuits.
But the deeper issue is the revenue model. ZK rollups are designed to compete with L1 transaction fees, but in a bear market, L1 fees are already low. Why would a user pay $0.05 on a ZK rollup when they can pay $0.02 on Ethereum mainnet? The marginal benefit of ZK security is not enough to justify the premium. The market is voting with its feet. TVL on these rollups has dropped 40% in the last three months, and active addresses are down 60% from their peak. Whispers become roars in the blockchain's memory.
Contrarian: The High Cost as a Feature, Not a Bug
Here is the counter-intuitive angle: The high proving cost is actually a form of security. It prevents the rollup from being too cheap to attack. If proving were free, operators could spam the network with cheap proofs and overwhelm the validators. The cost acts as a natural barrier against denial-of-service attacks. But that argument is a double-edged sword. If the cost is too high, no one will use the rollup, and the network effect never materializes.
Another contrarian view: The market is mispricing the risk of operator centralization. When operators bleed money, they have two choices: shut down or centralize. Centralization means switching to a single prover, which defeats the purpose of ZK. Some projects are already quietly moving toward a centralized sequencer model, where the sequencer is the only entity generating proofs. The community doesn't know yet. The crash strips the noise, leaving only structure.
I believe the real blind spot is the assumption that ZK proof generation will follow Moore's Law and become cheap enough quickly. But my experience auditing ZK circuits tells me that the complexity of the circuits grows faster than hardware efficiency. As more features are added (like account abstraction, built-in bridges, and privacy), the circuit size increases, and the proving time goes up. The cost curve is not linear—it's exponential. To hold firm is to understand the void.
Takeaway: The Next Narrative
The narrative is shifting beneath our feet. The next phase will not be about 'ZK is the future' but about 'ZK is a luxury only the strong can afford.' The strong are the ones with deep venture capital pockets or those who can build a sustainable revenue model beyond transaction fees. I see two possible paths: either the market revives with a new bull run, and fees return to a level that covers proving costs, or we see a consolidation wave where only the most capital-efficient rollups survive.
I am tracking the on-chain data for signs of capitulation. When the operators stop posting proofs for more than 24 hours, that is the signal. The quiet signal. Until then, I watch the numbers. The code whispers truths only the silent can hear. Trust is a variable, not a constant. Fragility breaks the loudest voices first. The crash strips the noise, leaving only structure. Whispers become roars in the blockchain's memory. To hold firm is to understand the void.
