The numbers don't lie—but the narratives do.
On March 12, 2025, a single ZK proof for a batch of 1,000 transactions on Scroll cost 0.87 ETH in gas. At $3,400 ETH, that's $2,958. The batch fees collected? $2,100. Negative spread of $858 per batch. This isn't an outlier—it's the new baseline. I've been tracking this for months, and the trend is clear: ZK rollup operators are bleeding capital even as the bull market euphoria pushes TVL to new highs.
Context: The ZK Rollup Promise vs. Reality
ZK rollups were supposed to be the holy grail of Ethereum scaling: infinite throughput, immediate finality, and Ethereum-level security. The pitch was simple—batch thousands of transactions, generate a succinct proof, and pay a fraction of L1 gas. The market bought it. Scroll, zkSync, StarkNet, and Linea collectively hold over $12 billion in TVL. VCs poured billions into these teams. But there's a dirty secret that nobody in the marketing department wants to talk about: the cost of generating that proof, including the L1 verification fee, frequently exceeds the revenue from user fees.
In my years auditing protocol balance sheets—back when I was doing post-mortems on DeFi lending collapses—I learned one thing: revenue-per-user is a vanity metric if unit economics are negative. ZK rollups currently operate on a model where every additional batch processed increases the deficit. The L1 verification cost scales linearly with batch size, but user fee collection is capped by a competitive market. The result: a structural Ponzi-like subsidy from token treasuries to maintain growth.
Core: The Math Behind the Bleed
Let me walk you through the actual numbers from a recent week of on-chain data across the four major ZK rollups:
Scroll: 8,200 batches, average batch size 1,200 tx, proving cost (gas + hardware amortization) $3,200 per batch, average batch revenue $2,450. Weekly loss ~$6.15M. zkSync Era: 15,000 batches, average batch size 900 tx, proving cost $2,800 per batch, revenue $2,100. Weekly loss ~$10.5M. StarkNet: 6,500 batches, average batch size 1,800 tx, proving cost $4,100 per batch, revenue $3,200. Weekly loss ~$5.85M. Linea: 9,800 batches, average batch size 1,100 tx, proving cost $3,000 per batch, revenue $2,300. Weekly loss ~$6.86M.
Aggregate weekly loss across these four: ~$29.36M. Annualized: $1.53 billion. And that's just the proof costs—not including R&D salaries, marketing, or infrastructure.
These projects sustain this because they draw from multi-hundred-million-dollar treasuries raised during the 2021-2022 cycle. But treasuries are finite. At current burn rates, Scroll's treasury (estimated $200M) lasts about 8 months. zkSync's ($450M) lasts about 10 months. StarkNet's ($300M) lasts about 12 months. Linea's ($150M) lasts about 5 months.
Contrarian: Bull Market Masking Structural Fragility
The conventional wisdom says: as ETH gas prices rise during a bull market, ZK rollups become more profitable because users are willing to pay higher fees for fast execution. That's true—but only partially. The L1 verification cost is denominated in ETH gas, which rises with L1 congestion. So when L1 spikes, both revenue AND cost spike. The spread actually widens because proving costs increase proportionally more due to the fixed overhead of proof generation hardware. In a high-fee environment, the operator's cost-to-serve increases faster than user willingness to pay.
This creates a paradox: the bull market that boosts TVL and fee revenue also boosts the cost of the very infrastructure that enables that growth. It's a fragility loop. And when the market turns, TVL evaporates, fee revenue collapses, but proving costs remain sticky (hardware leases, team salaries). The result: a death spiral.
Most DAOs governing these rollups have no legal recourse—most are structured as Cayman foundations with no obligation to disclose real-time financials. If they go under, token holders are left with nothing. I've seen this playbook before: first the yield disappears, then the liquidity, then the silence.
Takeaway: Positioning for the Cycle
Emotion is the asset; discipline is the hedge. Right now, the market is pricing ZK rollups as if they are revenue-generating businesses with moats. They are not. They are subsidized experiments with negative unit economics. The catalyst for a repricing will come when one major rollup announces a treasury stress event—likely within the next 6-8 months.
Smart capital is already rotating into L1s with sustainable fee structures (SOL, AVAX) or into Bitcoin as the ultimate macro hedge. The ZK narrative is loud, but the signal is weak. Watch the cash flows, not the buzz. Volatility is the price of entry—but negative unit economics is the price of exit.
I'm not saying ZK rollups will die. I'm saying they will survive only if they pivot to a SaaS model—charging protocols for dedicated proving capacity rather than relying on retail user fees. Until then, this is a story about burning cash to buy growth in a market that rewards hype over hygiene.
Chaos is just unstructured order. The order here is simple: if the math doesn't work in a bull market, it will be catastrophic in a bear one. Position accordingly.
