A huge thanks to Amin, Cooper, Hannes, Jacob, Michael, Norbert, Omer, and Teemu for sharing their feedback on the model and the article (this doesn’t mean they agree with the presented numbers!).

Zero-knowledge proofs are entering a period of rapid growth and widespread adoption. The core technology has been battle-tested, and we have begun to see the emergence of new services and more advanced use cases. These include outsourcing of proof computation from centralized servers, which opens the door to new revenue-generating opportunities for crypto infrastructure providers.

How significant could this revenue become? This article explores the proving ecosystem and estimates the market size in the coming years. But first, let’s start by revisiting the fundamentals.

Proving ABC

ZK proofs are cryptographic tools that prove a computation's results are correct without revealing the underlying data or re-running the computation. 

There are two main types of zk proofs:

1. Elliptic Curve-based SNARKs: Slow to generate but have a fixed proof size, regardless of computation size.
2. Hash-based STARKs: Can be faster to generate but produce larger proofs, making verification on L1s costly.

A zk proof needs to be generated and verified. Typically, a prover contract sends the proof and the computation result to a verifier contract, which outputs a "yes" or "no" to confirm validity. While verification is easy and cheap, generating proofs is compute-intensive.

Proving is expensive because it needs significant computing power to 1) translate programs into polynomials and 2) run the programs expressed as polynomials, which requires performing complex mathematical operations.

ZK Ecosystem

This section overviews the current zk landscape, focusing on project types and their influence on proof generation demand.

Demand Side

• zk-Rollups: The demand for proving currency mostly comes from zk-rollups. In 2024, the main zk-rollups (zkSync Era, Linea, Starknet, and Scroll) generated 580K transactions. Each transaction requires multiple proofs to be generated. 
• zkVMs: Developers can write zk circuits on their own using or use a zkVM to abstract away the zero-knowledge part and use just a high-level language like Rust to write applications. This democratizes access to zk-proofs as devs no longer need to learn domain-specific languages to write verifiable code. zkVMs will not drive demand by themselves but will instead facilitate one coming from rollups, apps, and infra projects.
• Apps and Infrastructure: Any apps and infra projects using zk, including privacy apps, oracles, bridges, or zkTLS.
• Aggregators minimize verification costs by batching multiple proofs from various sources. Instead of sending proofs directly to an L1, rollups, apps, or zkVMs can route them to an aggregator. The aggregator validates these proofs off-chain and submits a single consolidated proof to the L1. Since L1 verification incurs high gas costs on Ethereum (400-500k for SNARKs, up to 5 million for STARKs), it is the most expensive aspect of the current zk pipeline. 

Supply Side

• Infrastructure Providers: The main limitation in proof generation is hardware. Thus, anyone with powerful hardware will be incentivized to generate proofs. In blockchain, companies with extensive hardware expertise operate validators, making zk-proving a natural next step for them.
• Centralized Proving: The demand side can independently generate proofs, e.g., at the sequencer level for a rollup, or outsource them. Currently, rollups utilize centralized provers, but there is an incentive to offload proving to improve decentralization and liveness.
• Client-Side Proving (on user device): Shifting proving to user browsers reduces trust assumptions in zk applications by eliminating the need to send user data to proving servers. Performance constraints currently limit proof generation on consumer devices and will likely remain so for some time.

For the privacy-focused rollup Aztec, only one proof per transaction will be generated in the browser, as depicted in the proving tree below. A similar dynamic is expected with other projects.

• Hardware and Accelerators: Companies build specialized hardware and software-based hardware accelerator platforms. While these projects do not directly generate proof demand, they enhance proof delivery speed.
• Proving Marketplaces: Networks that connect proof demand with computing power. They will not generate proofs by themselves.

Monetization

Monetization strategies will include fees and token incentives.

The primary revenue model will rely on charging base fees. These should cover the compute costs of proof generation. Prioritization of proving work will likely require paying optional priority fees.

The demand side and proving marketplaces will offer native token incentives to provers. These incentives are expected to be substantial and initially exceed the market size of proving fees.

Proving Market Opportunity

Market Dynamics

To understand the proving market, we can draw analogies with the proof-of-stake (PoS) and proof-of-work (PoW) markets. Let’s examine how these comparisons hold up.

At the beginning of 2025, the PoS market is worth $16.3 billion, with the overall crypto market cap around $3.2 trillion. Assuming validators earn 5% of staking rewards, the staking market would represent approximately $815 million. This excludes priority fees and MEV rewards, which can be a significant part of validator revenues. 

PoS characteristics have some similarities to zk-proving:

• Both prioritize accuracy, speed, and reliability in computation.
• They could use similar economic tools, such as posting bonds and slashing.

The PoW market can be roughly gauged using Bitcoin’s inflation rate, which is expected to be 0.84% in 2025. With a $2 trillion BTC market cap, this amounts to around $16.8 billion annually, excluding priority fees.

Both zk-proving and PoW rely on hardware, but they take different approaches. While PoW uses a “winner-takes-all” model, zk-proving creates a steady stream of proofs, resulting in more predictable earnings. This makes zk-proving less dependent on highly specialized hardware compared to Bitcoin mining.

The adoption of specialized hardware, like ASICs and FPGAs, for zk-proving will largely depend on the crypto market’s volume. Higher volumes are likely to encourage more investment in these technologies.

With these dynamics in mind, we can explore the revenue potential zk-proving represents.

Methodology

Our analysis will be based on the Analyzing and Benchmarking ZK-Rollups paper, which benchmarks zkSync and Polygon zkEVM on various metrics, including proving time.

While the paper benchmarks zkSync Era and Polygon zkEVM, our analysis will focus on zkSync due to its more significant transaction volumes (230M per year vs. 5.5M for Polygon zkEVM). At higher transaction volumes, Polygon zkEVM has comparable costs to zkSync ($0.004 per transaction).

Approach

• Measure the proving time of groups of different transaction types (e.g., ERC token transfers, ETH transfers, contract deployments, hash function computations) in various quantities. This data is based on the benchmarks available in the paper.
• Create a batch of roughly 4,000 transactions, which matches the average batch on zkSync.
• Calculate the proving time for the batch, including the STARK to Groth16 compression time. 
• To calculate the costs, use cloud-based hardware offering:
  
  1. Hardware: 32 vCPUs, 1 NVIDIA L4 GPU.
  2. Cloud Cost: $1.87/hour.

Results

A single Nvidia L4 GPU can prove a batch of ~4,000 transactions on zkSync in 9.5 hours. Given that zkSync submits a new batch to L1 every 10 minutes, around 57 NVIDIA L4 GPUs are required to keep up with this pace.

Proof Generation Cost

Knowing the compute time, we can calculate proving costs per batch, proof, and transaction:

• Batch Size: 3,985 transactions.
• Cost per batch: $17.97.
• Cost per proof: $0.0423.
• Cost per transaction: $0.0045.

The above calculations can be followed in detail in Proving Market Estimate(rows 1-29).

Proving Costs Estimates

Proving costs depend on the efficiency of hardware and proof systems. The hardware costs can be optimized by, for example, using bare metal machines.

2024: Current Costs

• zkSync: $0.0045 per transaction.
• Other zk-Rollups: Since smaller and less optimized rollups have higher costs, a 40% premium is applied. This brings their proving cost to $0.0063 per transaction.

2025: Optimizations Begin

• zkSync: Proving costs remain at $0.0045 per transaction.
• Other zk-Rollups: Optimizations reduce costs down to $0.0059 per transaction.

2030:  Proving costs fall to $0.001 per transaction across all rollups.

Transaction Volume Estimates

2024: Real Data

The number of transactions generated by rollups and other demand sources:

• zk-Rollups: Virtually the only demand driver with 580M transactions. No rollup opened provers in 2024, but this will change starting in 2025.
• Optimistic Rollups: None added zk-proving in 2024, but transaction volumes are a baseline for future estimates: 2.3B transactions.
• Apps and Infrastructure: negligible.

2025: Market Takes Off

The proving market begins to gain momentum. Estimated number of transactions: ~4.4B, including: 

• zk-Rollups: The primary driver with 2.46B transactions.
• Apps and Infra: Demand starts to grow with 490M transactions.
• Aggregators: Smaller share. For simplicity, one batch equals one transaction in this analysis. Add 12M transactions.
• Other Blockchains: Aleo, now on mainnet, will contribute significantly. With zk-compression on Solana and Celestia’s zk initiatives in the early stages, the impact is 366M transactions.
• Multi-proofs: Optimism implements zk-proofs to improve finality time, adding 1.09B transactions.

2030: zk-Proving at Scale

Proving will have reached widespread adoption. Estimated number of transactions: ~600B

• zk-Rollups transactions volume grows to 17B.
• Optimistic Rollups will switch to validity proofs, increasing transaction volumes and driving demand for 69B transactions.
• Apps and Infra: New ideas and legacy solutions add 15B transactions.
• Aggregators are crucial but do not drive significant transaction volumes with 151M.
• Other Blockchains: Solana, Celestia, and various L1 platforms have significantly advanced their zk efforts. Ethereum Beam Chain is live, bringing the total transaction count to 108B.
• Unknown Opportunities: zk-proving expands into the real world, with use cases like Worldcoin adding 76B transactions.
• Multi-proofs: At least one redundant proof system will be integrated across almost all ecosystem projects, adding 315B transactions.
• Client-side Proving: Required by privacy-preserving solutions substracts around 3.5B transactions from the market.

Market size estimate

We estimate the proving surplus based on previously estimated proving costs. This surplus is revenue from base and priority fees minus hardware costs. As the market matures, base fees and proving costs decrease, but priority fees will be a significant revenue driver. 

Token incentives add further value boost, While it’s difficult to foresee the size of these investments, the estimate is based on the information collected from the projects.

2024: Early Market

• zk-Rollups processed 590M transactions for $3.26M in hardware costs.
• There are no token incentives or proving fees.

2025: Expanding Demand

The total market is projected at $97M, including: 

• The total cost for all zk-proofs of $24M.
• A 30% proving surplus results in a market size of $32M.
• Projects offer significant token incentives alongside regular fees, boosting the market size by an additional $65M.

2030: Almost a Two-Billion-Dollar Market

The total zk-proving market opportunity is estimated at $1.34B.

• Proving costs are $813M.
• With priority fees increasing, the proving surplus rises to 60%, bringing the market to $1.3B.
• As the market matures, token incentives decrease, adding only $40M.

A detailed analysis supporting the calculations is available in Proving Market Estimate(rows 32-57).

Sensitivity Analysis

The estimates with so many variables and for such a long term will always have a margin of error. To support the main conclusion, we include a sensitivity analysis that presents other potential outcomes in 2025 and 2030 based on different transaction volumes and proving surplus. For the sake of simplicity, we left the proving costs intact at $0.059 and $0.001 per transaction in 2025 and 2030, respectively.

In 2025, the most pessimistic scenario estimates a total market value of just $12.5M, with less than a 10% proving surplus and 2B transactions. Conversely, the ultra-optimistic scenario imagines the market at $55M, based on a 50% surplus and 6B transactions.

In 2030, if things don’t go well, we could see a proving market of roughly $300M, from 10% proving surplus and 300B transactions. The best outcome assumes a $1.7B market based on a 90% surplus and 900B transactions.

Risks

Estimating so far into the future comes with inherent uncertainties. Below are potential error factors categorized into downside and upside scenarios:

Downside 

1. Broader blockchain adoption may not occur as quickly as anticipated, slowing transaction growth across the ecosystem participants.
2. The dynamics of priority fee markets may not follow the same path as those of today’s blockchains, which can lead to overestimating the proving surplus.
3. Multi-proofs significantly increase transaction volumes in the estimates. However, projects might stick with single proving systems supported by Trusted Execution Environments (TEEs), which offer similar functionality on a hardware rather than software level.
4. Without major security breaches, optimistic rollups may not feel pressure to switch to zk-proving beyond adding a single proof system for reduced finality.
5. Advancements in proving tech could drastically reduce costs, leading to commoditization. Profit margins will be compressed as proving services become broadly available at lower prices.

Upside

1. Breakthroughs in software, especially in apps and zkVMs, could accelerate adoption across and beyond blockchains, leading to faster growth than projected.
2. Priority fees significantly boost revenue for validators on Ethereum and Solana. If zk-proving follows suit, proving fees could exceed the estimates.

Conclusions

After PoW and PoS, zk is the next-generation crypto technology that complements its predecessors. Comparing proving revenue opportunities with PoW or PoS is tricky because they serve different purposes. Still, for context:

• The PoS market is valued at $16.3B, with roughly $800M going to validators (minus priority fees and MEV rewards).
• The PoW opportunity is about $16.8B annually, excluding priority fees. Of course, Bitcoin mining’s cost structure and competition differ significantly from zk-proving or PoS.

We estimated that the zk-proving market could grow to $97M by 2025 and $1.34B by 2030. While these estimates are more of an educated guess, they’re meant to point out the trends and factors anyone interested in this space should monitor. These factors include:

• Proof generation costs, driven by advancements in software and hardware.
• Demand for zk-proofs represented in transaction volumes.
• Base and priority fees, which influence the economic incentives for proving.

Let’s revisit these forecasts a year from now.

‍