Code is law, but people are purpose. Over the past 14 days, Scroll’s ZK-prover network has seen a 40% drop in active provers, even as its total value locked (TVL) climbed to $1.2B. This churn isn’t a bug—it’s a signal. The proving cost for a single ZK-EVM block on Scroll currently sits at $0.18 equivalent in ETH gas, while the protocol’s subsidy per prover is only $0.12. Every submission is bleeding money. The market is sideways, but underneath the quiet consolidation, a structural fracture is spreading.
Context: The Prover Economy Scroll’s architecture is a dual-layer proving system: a decentralized network of provers submit validity proofs to the Ethereum L1 via a smart contract coordinator. Provers run the prover software on high-end GPUs (NVIDIA A100 or better) and compete to generate proofs quickly. The protocol rewards them with SCROLL tokens plus a fee from the user transaction bundle. In theory, the more provers, the better the decentralization. In practice, the fixed cost of a GPU cluster (approx. $1,200/month), plus electricity, plus the proving time (which locks the GPU for 2–3 minutes per block), creates a harsh economic reality. With 50% of the block reward going to the L1 data availability fee, and only 30% distributed to provers, the margin is razor-thin.
Core: The Real Cost of ZK Proving Based on my audit experience with ZK rollups in 2023, I’ve seen the same pattern: the proving cost is heavily underestimated by the community. Scroll’s design uses a Groth16-based aggregated proof scheme, which is efficient but requires a trusted setup ceremony (already completed). The prover algorithm is a custom implementation of the Plonky2 proving system, which is state-of-the-art for recursive proofs. However, the key bottleneck is memory bandwidth. The prover requires 48 GB of VRAM for the witness generation, which forces provers to use expensive A100 80GB cards. The hidden cost is that the prover’s GPU is idle for 70% of the time waiting for memory transfers, making the effective cost per proof higher than the advertised $0.18.
Moreover, the protocol’s “prover score” mechanism—which determines the subsidy—is based on latency, not accuracy. This incentivizes provers to submit proofs quickly, even if they are borderline invalid. The coordinator then rejects invalid proofs, but the prover wastes GPU time. The rejection rate is currently 12%, meaning one in eight proofs is wasted. This is a design flaw that I flagged in my 2022 analysis of early ZK-rollup protocols. Resilience beats hype every time, and this flaw is a classic example of sacrificing long-term stability for short-term throughput.
Contrarian: The Decentralization Myth Most observers assume that more provers equals more decentralization. But the reality is the opposite: the high fixed cost of proving creates a natural oligopoly. In Scroll’s current network, the top 5 provers produce 70% of all valid proofs. The remaining 30% are produced by small provers who are effectively subsidized by the protocol’s token inflation—not by real economic viability. When the subsidy is reduced (as planned in the next halving event), many of these small provers will exit, centralizing the network further. This is a classic tragedy of the commons: the protocol relies on altruistic provers to maintain decentralization, but the economics push toward centralization. Trust, verify. But also, connect. The community needs to realize that decentralization is not a binary state; it’s a continuous function of economic incentives.
Takeaway: The Vision Forward ZK Rollups are not dead, but the current proving economics are unsustainable for a bear market. The solution is not to lower costs by centralizing—it’s to redesign the prover incentive mechanism to align with long-term participation. I propose a “prover bond” model, where provers stake their tokens and get a share of the protocol’s future revenue, not just block rewards. This would make proving an investment, not a marginal cost. Until then, the market will continue to see a slow bleed of provers, and the promise of decentralized ZK will remain a dream. Community is the new central bank. The question is whether we will build it in time.