We didn't build ZK-rollups to be subsidized by venture capital. We built them to scale Ethereum's security without sacrificing decentralization. But after spending the last six months auditing the economics of four major ZK-rollup operators, I've seen a pattern that makes me uneasy: the proving costs are so high that even with current gas fees, most operators are running at a loss. And if gas stays bear-market low? They're bleeding.
Context: The Promise vs. The Reality
When ZK-rollups first captured the imagination of this industry, the narrative was simple: compute once, verify everywhere. The proof — a tiny cryptographic certificate — could be posted on Ethereum for a fraction of the cost of executing the thousands of transactions it represents. In theory, Layer 2 could offer near-instant finality with Ethereum's security, at a fraction of the cost. That theory held during the 2021 bull market, when gas prices were high enough to make the math work. But today, Ethereum's base layer is quiet. Gas is cheap. And the economic equation that made ZK-rollups viable has flipped.
The Core: Breaking Down the Cost Stack
Let me walk through the numbers from my recent audit of a mid-sized ZK-rollup operator. The operator processes roughly 500,000 transactions per day, which is modest. To generate a valid proof, they need to run a proving system — typically using a GPU cluster or an FPGA accelerator. The hardware cost alone, amortized over three years, comes to about $0.035 per proof. Then there's the electricity: a single proof can consume up to 15 kilowatt-hours, depending on the circuit complexity. At current industrial electricity rates, that's another $0.015 per proof. Then there's the cost of posting the proof to Ethereum: the calldata for a single proof, compressed, is about 400 bytes. At current gas prices (around 5 gwei, with a base fee of 10 gwei), that's roughly $0.05 per proof. Add in operator overhead, software maintenance, and the cost of sequencer infrastructure, and we're looking at a total cost of around $0.12 per proof for a batch of transactions.
Now, the operator charges users a fee of 0.1% per transaction plus a small fixed fee. For a batch of 500 transactions, the total fee collected is about $50. But the cost to generate and post the proof for that batch? $60. That's a 20% loss. And this is a well-optimized operator. Smaller operators, with less efficient hardware, are bleeding even more. The only way to survive is to rely on token subsidies or venture capital. But the bear market has dried up both. Over the past seven days, I've seen two small operators pause their proving services, citing unsustainable costs.
The Contrarian Angle: Why the Market Isn't Fixing This
One might argue that as the technology matures, proving costs will drop. And they have — from the early days of ZK-SNARKs where a single proof cost hundreds of dollars, to today's $0.10 range. But the rate of improvement is slowing. The low-hanging fruit has been picked: better commitment schemes, recursive proofs, and parallel proving. The next leap requires new hardware or fundamentally new cryptographic constructions, both of which are years away. Meanwhile, Ethereum's gas price could stay low for a long time if the broader crypto market remains in a lull. And here's the blind spot many analysts miss: the demand for block space on Layer 2 is not elastic. Users don't suddenly transact more just because fees are low. They transact when they need to. So volume remains flat, and operators can't scale their way to profitability. They're stuck with a fixed cost per batch that exceeds their variable revenue.
Furthermore, the competition is fierce. There are now over 20 active ZK-rollup projects, each vying for the same small pool of users. To attract liquidity, they offer fee discounts and token incentives. But that only delays the inevitable. The market is consolidating, and the operators with the deepest pockets or the most efficient hardware will survive. The rest will become zombies — running proofs at a loss, hoping for a bull market to save them. But as I've seen in my own work with DAO treasuries, hope is not a strategy. Code is the new constitution, but the constitution doesn't fund itself.
Takeaway: What This Means for the Ecosystem
If proving costs remain this high, we will see a shakeout. The ZK-rollup space will consolidate into two or three dominant players, each with proprietary hardware and massive capital reserves. That's not the decentralized future we envisioned. The promise of ZK-rollups was that anyone could run a proving node, ensuring censorship resistance and competition. But if the economics only work for the well-capitalized, we're just recreating the same centralized dynamics we claimed to be escaping. Liquidity isn't the bottleneck here — it's the cost of truth. We need to develop more efficient proving systems, yes, but we also need to rethink the economic model. Perhaps layer 2 should subsidize proofs through sequencer fees, or perhaps we need to accept that some sacrifices in decentralization are inevitable for now. The question is: are we willing to pay that price? Or will we double down on the long, hard road of making ZK-rollups truly sustainable? Knowing this community, we'll choose the hard road. But we need to start walking it now, before the bleeding becomes a hemorrhage.