The numbers don't lie: ZK rollup operating margins are bleeding red.
I spent this morning digging into the latest batch of on-chain data for the top five ZK-rollup networks. The headline is stark — every single one is burning cash at a rate that would terrify a traditional venture capitalist. Not because of hack or exploit. Because the core economic assumption behind these chains — high gas fees on Ethereum — has quietly evaporated.
Context: The Bull Market’s Hidden Tax
During the 2021-2024 cycle, Ethereum L1 fees often spiked above $50 per simple swap. That made ZK rollups look like the only viable path to mass adoption. Projects raised billions based on the premise that they could offer near-zero fees by batching transactions off-chain and submitting succinct proofs. The economic model was simple: Charge users a fraction of L1 cost, collect the difference, and scale.
But here’s the catch no one wanted to discuss in those pitch decks: the math only works if L1 gas stays above a certain threshold. When Ethereum fees dropped to $1-3 during the 2022-2023 bear market, rollups kept their fee structures artificially low to attract users. They subsidized the gap with treasury funds and token incentives. That was fine during a bull run when TVL was growing. Now, in 2025, with spot Ethereum ETF inflows compressing volatility and L1 fees averaging below $2, the subsidy model is breaking.
Core: Forensic Dissection of ZK Rollup Profitability
Let me walk through the numbers using a representative mid-tier ZK rollup — let's call it Chain X. Based on my audit of its publicly disclosed data (and several private conversations with its operations team), the average cost to submit a batch of proofs to Ethereum L1 is currently 0.8 ETH per batch. At current ETH price of ~$3,500, that’s $2,800. Chain X processes roughly 200 batches per day, so daily L1 settlement cost = $560,000.
Now, what does Chain X earn in fees? Its users pay an average of $0.02 per transaction. Daily transaction volume is approximately 1.5 million. That’s $30,000 in daily revenue. Even if I add sequencer MEV extraction (which is minimal on this chain due to low volume and simple order types), total revenue might reach $50,000. The gap: $560,000 - $50,000 = $510,000 daily deficit. That’s $186 million annually.
Where does that money come from? Token issuance. Chain X inflates its native token supply by roughly 12% annually, and the newly minted tokens are sold by the foundation to cover operating costs. This is effectively a tax on existing holders. The same pattern repeats across Arbitrum, Optimism, zkSync, and others — though some have larger treasuries and can sustain the burn longer.
I recall a similar dynamic from my 2020 DeFi Summer audit days, when I modeled yield farming strategies and discovered that high APYs were simply the protocol selling its own token to cover returns. The math was unsustainable then. It’s worse now because the revenue base is smaller.
The Critical Variable: Proving Costs
The real killer is not just L1 settlement. It’s the cost of generating zk-proofs. For a ZK rollup, each batch requires a proof that must be computed by specialized hardware (GPU or FPGA). The proving cost per batch adds another $500-1,500 depending on circuit complexity. That brings total daily cost above $700,000 for Chain X. Meanwhile, competitors like Mantle or Blast are using simpler OP-rollup technology with no proving costs. They can undercut fees even more, putting ZK rollups in a race to zero they cannot win.
Emotion is the asset; discipline is the hedge. Right now, the market is full of emotion about “ZK proving efficiency improvements” and “parallel circuit optimization.” But discipline tells me that these improvements are marginal — maybe 2x over the next year. They need 10x to break even. And that assumes L1 gas doesn’t fall further. If it does, the subsidy hole deepens.
Contrarian Angle: The Decoupling Thesis That Never Materialized
The bull narrative for ZK rollups has always been that they would decouple from Ethereum’s fee economics — that volume would generate enough fee revenue to stand alone. That thesis hasn’t played out. In fact, the opposite has happened: as Ethereum becomes more efficient (thanks to blob data and proto-danksharding), L1 fees are compressing, not expanding. The 2025 spot ETF-driven bull market has actually reduced on-chain speculation because institutions buy the ETF, not the underlying asset. Base-layer transaction volumes are flat compared to 2023. The demand that was supposed to flow to rollups never arrived in sufficient quantity.
What’s more, the market is now flooded with alternative L1s that offer similar throughput without the extra complexity. Solana, Sui, and Aptos all process millions of transactions for pennies per transaction — and they don’t need to pay L1 settlement costs. ZK rollups were supposed to be the Ethereum ecosystem’s scaling solution, but they’ve become a premium product in a world that demands discount.
Takeaway: The Clock Is Ticking
If you hold ZK rollup tokens hoping for a narrative revival, look at the cash flows first. At current burn rates, most of these projects have two to four years of treasury runway. But the clock isn’t just about treasury — it’s about token price. Inflation from token sales will drip-feed sell pressure. The only way out is a sharp increase in L1 fees (which would require a speculative mania on Ethereum) or a reduction in proving costs by 10x. Neither is guaranteed.
So where does that leave us? ZK rollups will not die — the technology has real value for privacy and interoperability. But as yield generation vehicles or speculative assets, they are operating on borrowed time. The liquidity mirage will fade when the next bear market strips away the subsidy. By then, the survivors will be those that pivot to a real revenue model — perhaps by selling proving power to enterprise users or focusing on specific verticals (e.g., gaming).
Until then, I’m watching the flows, not the foam. And the flows are pointing one direction: red.