The bull market is here. Total value locked in DeFi has surged past $120 billion, a new all-time high. Yet, when I pulled the latest on-chain data for ZK rollup operators last night, the revenue numbers told a different story. Proving costs are devouring margins at a rate that makes the bull case for most L2 tokens look like a speculative mirage.
Let me walk you through the mechanics. Ethereum’s L1 gas is hovering around 30 gwei – not the 200+ gwei of 2021, but enough to make batch submission non-trivial. For a ZK rollup like zkSync Era or Scroll, the cost structure is split: L2 transaction fees collected from users, minus L1 data posting costs, minus the proving cost (the computation to generate validity proofs). The first two are well understood. The third is the silent killer.
I spent the last two weeks auditing the economic models of three major ZK rollup operators as part of our firm’s sector review. The numbers are stark. A single validity proof on a modern GPU cluster can cost between $0.50 and $2.00, depending on circuit complexity. For a rollup processing 500,000 transactions per day, that’s $250,000 to $1,000,000 daily in proving costs alone. Meanwhile, the average transaction fee on these rollups is roughly $0.05. At 500,000 tx/day, gross revenue is $25,000. The bleed is obvious: proving costs are 10x to 40x revenue.
Contrast this with optimistic rollups like Arbitrum and Optimism, where fraud proofs are rarely computed. Their cost structure is dominated by L1 calldata. In a bull market, calldata costs rise, but they are still fractions of a cent per transaction. The result is that optimistic rollups are marginally profitable, while ZK rollups are hemorrhaging capital. The common narrative is that ZK is the holy grail – faster finality, better security, and scalability. But the economic reality is that ZK rollups are currently subsidized by token emissions and VC funding, not by sustainable unit economics.
This is not a new observation. Back in 2022, during my deep dive into liquidity fragility in DeFi, I noticed that many L2 tokens were priced based on future potential rather than current cash flow. The 2024 ETF approval accelerated the bull market, but it also masked structural flaws. The market is euphoric, and investors are buying the narrative of “ZK is the endgame” without doing the math on proving costs. I’ve seen this pattern before – in the ICO boom, where whitepaper promises were enough to raise millions. Today, the same naivety is being applied to L2 economics.
Let me clarify the counterpoint. Proponents argue that hardware improvements will reduce proving costs. They point to custom ASICs and recursive proofs that batch multiple transactions into one. That is true – but the timeline is uncertain. Ethereum’s own roadmap includes L1 improvements, but the EIP-4844 proto-danksharding (blob data) is already live and has reduced calldata costs for L1 posting. However, that does not help ZK proving. The bottleneck is the proof generation, not data availability. Even with blobs, a ZK rollup still needs to compute a proof every few minutes. The cost per proof is not dropping as fast as the bull market is inflating transaction volume.
Here is the contrarian angle: The decoupling thesis – that crypto assets will move independently of traditional markets in a bull run – is often applied to Bitcoin vs. equities. But I see a different decoupling happening within crypto. ZK rollup tokens are decoupling from their underlying network usage. Total value locked on zkSync Era has grown, yet the token price is stagnant. Why? Because the market is slowly realizing that the token’s value accrual is broken. The token is used for governance, not for paying fees. The real economic value is captured by the infrastructure (GPU providers, sequencers) not by token holders. This is a classic “value extraction” problem that I’ve written about in my post-mortem on liquidity contraction.
During the 2022 bear market, I audited three lending protocols and discovered hidden correlated exposures. The same pattern is repeating here: ZK rollup operators are exposed to proving cost spikes that are correlated with L1 congestion. When L1 is congested, Ethereum gas rises, which increases the cost of posting data, but also increases the incentive for users to use L2, which increases transaction volume, which increases proving costs. The operator is caught in a squeeze. The token holders are left holding a governance token that has no claim on the operator’s revenue (if any).
I’ve been in this industry for 17 years. I’ve seen the 2017 ICO idealism collapse under the weight of reality. I’ve seen DeFi summer turn into a liquidity trap. The current bull market is no different. The euphoria is real, but it masks the technical and economic fragility of the scaling solutions we are betting on. Emotion is the asset; discipline is the hedge.
What does this mean for cycle positioning? In a bull market, the tendency is to chase the highest beta narratives. ZK rollups are that narrative. But if you look at the cash flow, the only sustainable L2s right now are optimistic rollups with low proving costs, or application-specific chains (like dYdX’s L1) that circumvent the problem entirely. The takeaway is not to abandon ZK, but to be realistic about the timeline. The technology is sound, but the economics are not yet ready for prime time. Investors should ask: who is paying for the proving costs? If the answer is “tokens and VCs,” then the real value is being created, and extracted, elsewhere.
The next 12 months will be critical. If proving costs do not drop by at least 50%, the ZK rollup ecosystem will face a consolidation wave. Operators will merge, tokens will be restructured, and the market will learn the hard way that technology without sustainable economics is just a lottery ticket. I am watching the proving cost curves closely. That is where the signal is, not in the TVL headline.
