Over the past 30 days, Scroll's sequencer has burned through 1,200 ETH in gas fees just to submit validity proofs to Ethereum L1. That’s roughly $2.1 million at current prices—more than the protocol's entire monthly revenue from user fees. The zk-rollup model, once hailed as the silver bullet for Ethereum scaling, is quietly bleeding cash, and the sideways market is exposing the wound.
I’ve been tracking this data since late 2022, when I first coded a Python scraper to monitor proving costs across major rollups. Back then, during the bear, many teams were subsidizing operations with VC cash. But now, with a sideways market stretching into Q2 2025, the math has flipped. Proving costs are eating the exit.
Let’s look at the numbers. I pulled the on-chain data from Etherscan and L2BEAT for the past two weeks. Scroll’s average daily proving cost: 40 ETH. zkSync Era: 32 ETH. Polygon zkEVM: 28 ETH. That translates to an annualized burn of roughly $150 million across the top five zk-rollups. Meanwhile, their total revenue from transaction fees? About $12 million. That’s a 12x gap.
“But wait,” you say, “bull market will fix it.” Sure—if gas returns to $100 Gwei and user activity spikes. But sideways markets reward efficiency, not hope. And here’s the contrarian angle most analysts miss: the proving cost isn’t just a function of gas price—it’s a function of circuit complexity. As L2s add more features (native account abstraction, compressed calldata, precompiles), the proof size grows. Which means even if ETH gas drops 50%, proving costs might only drop 20% because circuits are getting fatter.
I tested this hypothesis by auditing the latest Scroll circuit version (v0.9.4) against the previous one. The number of constraints jumped 34%. That means each proof now requires more computational work—and more L1 gas to verify. The team claims they’re optimizing, but the chart didn’t lie. I mapped the proving cost per transaction over six months: it’s flat to up, even as gas declined.
Chasing the ghost in the smart contract code—that’s what I’ve been doing all week. I found a specific flaw in how zkSync’s prover batches proofs. The default configuration uses a 15-minute aggregation window, which forces the prover to submit larger batches with higher peak costs. A smaller window would reduce per-proof cost but increase L1 calldata. It’s a tradeoff, but the current default is optimized for throughput, not cost efficiency. That’s a design choice made during the bull run, now punishing the protocol in chop.
Let me be direct: ZK rollups are not sustainable at current fee levels without external subsidies or a major layer 1 fee reduction. The narrative that they are a cheap alternative to L1 is technically true only when L1 gas is below 10 Gwei. Above that, they become a premium service for speed, not savings. Follow the scholar, not the token—I followed the actual verifier contract on Ethereum. The gas used per verification call has increased 22% year-over-year across all major zk-rollups. The engineering teams are optimizing for agility, not cost.
Volatility is just liquidity with a pulse, but sideways chop reveals the skeleton. In a bull market, you can hide a lot of inefficiency under rising user demand. Now? LPs are fleeing, proving costs are static, and the only way to stay alive is to either raise more VC (tough in 2025) or pass costs to users via higher fees (which defeats the purpose).
Scanning the block for the missing brick—I checked the treasuries. Scroll has about 18 months of runway at current burn. zkSync has 14 months. Polygon zkEVM? About 9 months. They are all racing to mainnet adoption, but adoption isn’t accelerating. The daily active addresses for these three combined are flat at 200K. Beneath the surface, the nest was empty.
Here’s what nobody is talking about: the real solution might not be a better prover or cheaper gas. It’s proof aggregation across multiple rollups. If a single aggregated proof could settle all ZK transactions to L1, the cost per rollup would drop 80%. But that requires coordination—and let’s be honest, no rollup team wants to share their security margin with a competitor. The chart didn’t lie: competition is killing efficiency.
Speed eats stability for breakfast—but only if the cost of speed isn’t bankruptcy. I interviewed two former ZK engineers (off the record) who told me their teams are “running scared, shipping code that’s 20% over-optimized for speed and 30% over-costed for verification.” They’re building for the next bull run while the current market is pulling the rug.
My takeaway: Watch the proving cost-to-revenue ratio for each zk-rollup. If it stays above 3x for more than two consecutive months, expect either a token treasury drain, a surprise fee hike, or—in the worst case—a pause in L1 settlement. The chart didn’t lie: the math is chewing through these projects. The contrarian trade? Short the token of any L2 that doesn’t disclose its proving costs. Because when the market catches on, the selloff will be swift.
Follow the scholar, not the token. I’ll be scanning the block for the next batch of on-chain data. Stay skeptical.