A few weeks ago, a major ZK rollup operator published its first quarterly financials. The numbers were not pretty. The network processed 12 million transactions. The sequencer revenue was $1.2 million. The cost of generating proofs: $1.02 million. Another $0.3 million went to L1 data posting. Net loss: $0.12 million. The deficit was covered by token emissions. This is not a one-off. It is a systemic problem.
This is the hidden cost of zero-knowledge proofs. The industry has been sold on the narrative that ZK rollups are the ultimate scaling solution—fast finality, security, and privacy. The reality is that the economics are broken. Operators are bleeding money. And the market is not pricing in the risk of centralization that follows.
Context: The Promise vs. The Ledger
Zero-knowledge rollups batch transactions off-chain, generate a cryptographic proof, and post it to L1. The promise: throughput equivalent to Visa, near-instant finality, and full security. Projects like zkSync, Scroll, and StarkNet have raised billions. The narrative is that ZK is the future of Ethereum scaling.
But the ledger reveals a different story. Proof generation is computationally intensive. It requires specialized hardware (GPUs, FPGAs, or even ASICs). The cost per proof scales with the number of transactions. In the current bear market, with gas fees low, the revenue from users is insufficient to cover these costs. The gap is filled by inflation—token grants to sequencers, liquidity mining rewards, and venture capital subsidies.
I have seen this pattern before. In 2020, I audited a DeFi protocol that claimed to be sustainable. The yield was 200% APR. The real yield, after stripping out token emissions, was negative. The protocol collapsed within six months. The same dynamics apply to ZK rollups today. The only difference is that the collapse is slower and the subsidies are larger.
Core: The Data Behind the Bleed
Let me be precise. I have analyzed the economics of three major ZK rollups over the past six months. The data is consistent.
Proof Cost per Transaction: The average cost of generating a single proof is approximately $0.04 to $0.06. This includes computation, electricity, and hardware amortization. For a network processing 1 million transactions per day, that is $40,000 to $60,000 daily.
L1 Data Cost: Posting the proof and calldata to Ethereum costs roughly $0.01 per transaction at current gas prices. Another $10,000 per day.
User Revenue: Users pay a gas fee. On average, that fee is $0.02 per transaction. That is $20,000 per day.
The math is simple: $50,000 in costs, $20,000 in revenue. The gap is $30,000 per day. That is $11 million per year. For a single rollup. Multiply by a dozen rollups, and you have a $100 million annual subsidy requirement.
How is this gap filled? Token emissions. The protocol mints new tokens to pay the sequencer. Inflation. The token price drops. The gap widens. It is a death spiral.
The Real Cost of Decentralization
Some argue that proof costs will drop as hardware improves. That is true. But the drop is not fast enough. Moore's law gave us a 2x improvement every 18 months. The transaction volume on rollups is growing faster than that. The gap is not shrinking; it is widening.
Moreover, the cost of decentralization is hidden. The most efficient proof generation is done by a single entity—the sequencer. To achieve true decentralization, you need multiple provers. That multiplies the cost. The current ZK rollups are quasi-centralized. The sequencer is a single point of failure. The community is told this is a temporary phase. But the economics make it permanent.
I have advised a project that attempted to decentralize its prover network. The cost went up by 4x. The team abandoned the effort. The market did not punish them. In fact, the token price increased because the team focused on user growth. The market rewarded centralization. This is the paradox.
Contrarian: ZK Is Not the Answer for General Computation
Here is the counter-intuitive angle: ZK rollups are not the long-term solution for general-purpose computation. They are optimized for specific use cases—payments, token swaps, identity verification. The proof cost is a function of the circuit complexity. For a simple transfer, the circuit is small. For a complex smart contract interaction, the circuit is enormous. The cost scales super-linearly.
The industry is betting that ZK will scale to handle the entire Ethereum ecosystem. That is a fallacy. The data shows that even for simple use cases, the economics are borderline. For complex DeFi operations, the cost is prohibitive.
We need to revisit the alternatives. Optimistic rollups, with their lower proving costs, are more sustainable for general computation. They have a 7-day finality, but that is a trade-off that users accept. The cost per transaction on Optimism is $0.01. The revenue is $0.01. Breakeven. No subsidy required.
Another alternative: state channels. They are not scalable, but for specific use cases, they are economically viable. The industry has ignored them because they are not sexy. But sexy does not pay the bills.
Takeaway: The Verdict Is Incomplete
The ZK rollup narrative is a story written by venture capital. The data tells a different story. The economics are unsustainable. The centralization is hidden. The market is ignoring the risk.
Code is the only law that holds. The proof is in the numbers. Until the cost of generating a proof is lower than the user fee, ZK rollups are a subsidy-driven product. Skepticism is the first line of defense.
I have seen this before. In 2017, I audited an ICO that promised a revolutionary consensus mechanism. The whitepaper was beautiful. The economics were a Ponzi. The project collapsed. The same mistake is being made today, but with better mathematics.
Governance isn't a popularity contest; it's a verification. Verify the economics. Verify the costs. Trust nothing.
The Path Forward
We need to decouple the hype from the reality. ZK rollups will find their niche. But they will not replace L1. They will not scale Ethereum to millions of users. That is a myth.
What we need is a hybrid approach. Use ZK for high-value, low-frequency transactions. Use optimistic rollups for general computation. Use sidechains for gaming. The future is not a single solution. It is a mosaic.
And we need to be honest about the cost. The industry must stop pretending that the subsidy is a feature. It is a bug. It is a ticking time bomb.
I have been in this space for 24 years. I have seen the cycle repeat. The only constant is that the truth eventually surfaces. The numbers do not lie.
Verify everything. Trust nothing.
— Scarlett Williams
