As zero-knowledge proof (ZKP) technology becomes an indispensable cornerstone of the Web3 world, we are entering a new era prioritizing "computational trustworthiness." From validity proofs for Rollups to off-chain computation verification powered by zkVM, and privacy-AI fusion applications like zkEmail and ZKML, ZK is no longer just a cryptographic term but the execution and verification engine of the next-generation internet. According to Crypto.com's forecast, by 2030, Web3 services will need to execute nearly 90 billion zk proofs annually, with the market size for ZK computation expected to exceed $10 billion.

However, behind this explosive growth lies a harsh reality for ZK infrastructure: nearly all zk proof generation processes still heavily rely on centralized servers. Whether it’s zkRollup or zkVM, most mainstream projects currently depend on a handful of trusted entities to operate Prover nodes. While this architecture can勉强支撑 performance, it comes at the cost of high expenses, low resilience, censorship risks, and limited scalability—contradicting the decentralized ethos of blockchain.

Thus, building a ZK Prover Network has become a critical step in scaling zero-knowledge computation. By outsourcing proof tasks to community participants worldwide and constructing a permissionless, verifiable Prover network, we can achieve faster generation, stronger censorship resistance, lower costs, and true network-layer decentralization.

This article will unfold along two main threads:
·Part 1: Ethereum’s Snarkification Process and Vision for 3.0 — Ethereum is fully embracing ZK technology from the consensus layer (Beam Chain) to the execution layer (zkEVM/zkVM).

·Part 2: The Full Explosion of ZK Demand — With the surge in L2 networks, the rise of zkVM, and various ZK Dapps, the workload and diversity of ZK Provers are growing exponentially.

1. Ethereum’s Snarkification Process and Vision for 3.0

With the rapid development and application of zero-knowledge proof technology—including ZK-Rollups, zkVM, algorithmic optimization of proof systems, and hardware acceleration—Ethereum is gradually entering a new phase of "Snarkification." Reflecting on Ethereum’s evolution at EDCON 2024, Vitalik noted that in 2019, mainnet transaction confirmation times often stretched to tens of minutes, and zk-SNARK computation costs remained prohibitively high. Today, thanks to EIP-1559 optimizations and the widespread adoption of Layer 2, mainnet confirmation times have shrunk to 5–20 seconds, with Layer 2 achieving sub-second confirmations and transaction fees dropping below one cent. Looking ahead, as zk-SNARK performance continues to improve, Layer 2 TPS is expected to exceed 100,000, while significantly reducing computation and storage overhead.

Vitalik believes ZKP has evolved from a "theoretical technology" five years ago into a widely available core infrastructure for both on-chain and off-chain use today, with an especially urgent need for on-chain privacy and efficient interaction in non-financial scenarios like social applications. He also called on the Ethereum ecosystem to confront privacy issues—current solutions like the Railway privacy wallet still have high usage barriers, limiting the widespread adoption of privacy features. Fortunately, the evolution of zk-SNARK has given rise to more feasible solutions, such as local key backups and zk-encapsulated centralized identities (e.g., zk-email, ZKTLS, ZKDID), gradually expanding the boundaries of privacy applications. In the future, ZK is expected to become ubiquitous on everyone’s devices.

In October 2024, Vitalik proposed long-term goals for Ethereum block validation, aiming to achieve a more efficient and decentralized verification mechanism to support a broader range of clients (including light clients) and cross-chain scenarios. The specific goals include two aspects:

Minimalist block downloads: Through Data Availability Sampling (DAS) technology, nodes only need to download a tiny fraction of block data to verify its integrity.

Lightweight block validation: The verification process only requires checking a small proof, which can be quickly validated in resource-constrained environments—such as mobile clients, browser wallets, or even cross-chain validation on other blockchains—thereby enhancing Ethereum’s decentralization and scalability.

To realize this vision, improvements are needed at two core levels of Ethereum: the consensus layer (i.e., the proof-of-stake mechanism, currently implemented by the Beacon Chain) and the execution layer.

Consensus Layer Improvements: Beam Chain

Using zero-knowledge proofs to reduce the computational and communication burden on validators, lowering the entry threshold for becoming a validator, allowing more validators to join the network to enhance security, reduce computational redundancy, and improve overall network efficiency. At the same time, ZK-ifying the consensus layer requires encapsulating processes like state transitions and validator signature aggregation into SNARK or STARK proofs to simplify verification. This is a complex technical challenge in itself. Ethereum’s current Beacon Chain, designed between 2018 and 2020, carries technical debt, such as long block times (12 seconds) and finality delays (about 13 minutes). The Beam Chain proposal leverages zero-knowledge proofs as a core technology to optimize the consensus mechanism, boost efficiency, and prepare for future quantum resistance, including the following aspects:

1. SNARKification of state transitions: "ZK-ifying" the State Transition Function (STF) by using SNARKs to prove the correctness of state updates (e.g., account balances, validator registries, signature aggregation).

2. Achieving single-slot finality through ZK

3. Quantum-resistant ZK upgrades: Introducing quantum-resistant STARKs to address the threat of quantum computing.

4. Integration of zkVM: Verifying complex consensus logic through zkVM

5. Optimizing shard data verification with ZK: Each shard generates state updates and SNARK proofs, with the main chain only verifying them.

Execution Layer: EVM Validity Proofs

ZK proofs for EVM execution are already widely used in Layer 2 (L2) solutions, such as zk-Rollups (e.g., zkSync, StarkNet). These solutions ensure trustworthy computation results by generating validity proofs while submitting data to Layer 1. However, applying EVM validity proofs directly to Layer 1 still faces two major bottlenecks:

Insufficient security: Secure EVM validity proofs must ensure that SNARKs accurately verify EVM computations without errors. Proofs in L2 are typically tailored to specific scenarios with smaller computation scales, making security manageable. On L1, however, the EVM must handle more complex computations (e.g., DeFi protocols), increasing the attack surface of the proof system.

----Solution: Multi-prover and formal verification: A multi-prover approach means having multiple independently written validity proof implementations, similar to having multiple clients. If a sufficiently large subset of these implementations proves a block, the client accepts it. Formal verification involves using tools typically employed to prove mathematical theorems to ensure that validity proofs only accept inputs that correctly execute the underlying EVM specification.

Proof generation time too long: For instance, generating a SNARK proof for a complex smart contract execution currently takes seconds to minutes, while L1’s block production target is 4–12 seconds (with Beam Chain aiming to reduce it to 4 seconds). We are still far from this goal.

-----Solution: ① Parallelization: The fastest current EVM provers can prove a typical Ethereum block in about 15 seconds. To further reduce this time, parallelization techniques are needed to break EVM computation into multiple parts, generating proofs in parallel at the opcode level. This could be based on the EVM itself or a lower-level virtual machine (e.g., RISC-V). ② Proof system optimization: Emerging proof systems (e.g., Orion, Binius, GKR) promise to significantly reduce proof times for general computation and lower computational overhead. For example, Binius’s binary field optimization could boost proof generation speed by 2–3 times, while GKR’s layered proving further reduces the overhead of complex computations.

The Demand Side of ZK Proofs

If Ethereum 3.0 represents a structural demand for ZK from the ground up, the explosion on the demand side is a pressing need driven by real-world dynamics.

1. The Growing Number of L2 Networks: Surge in Quantity and Application Expansion

The ZK technology stack has successively released production-ready chain development toolkits (e.g., Polygon’s SDK, zkSync’s Elastic Chain, OP Stack), making it easier to deploy new validity Rollups or Validiums, significantly increasing the number of L2s. As Ethereum’s modularization accelerates, the computational demand for zk proofs will rise rapidly, for example:

Enterprise-grade L2s: For instance, Hong Kong’s HashKey launched its own L2, HashKey Chain, to expand application scenarios; Deutsche Bank adopted zkSync technology for Project Dama 2, enabling more efficient asset service transactions; Coinbase’s Base.

Scenario-specific L2/L3s: For example, Sophon (a Validium L2 based on Avail’s data availability layer) focuses on gaming scenarios.

RaaS (Rollup as a Service): For example, Zeeve supports zkSync with one-click deployment tools, offering over 40 integrated services (e.g., sequencers, block explorers), significantly lowering the development barrier.

Equilibrium Research predicts that by the end of 2025, the number of L2 solutions on Ethereum will exceed 2,000, with ZK L2s dominating. ZK-Rollups will surpass Optimistic Rollups, and with the development and adoption of zkVM—especially with the emergence of Succinct—we are already seeing Optimistic solutions gradually shifting toward ZK.

2. The Decentralization Trend of L2

ZK-Rollups aggregate large batches of transactions and submit a concise proof to Layer 1, significantly boosting transaction throughput and reducing costs.

However, generating these proofs is computationally intensive, and currently, it largely relies on centralized infrastructure (e.g., dedicated servers or cloud services). This centralized approach has the following issues:

Performance bottlenecks: Centralized Provers may struggle to handle high transaction volumes, causing delays.

Limited decentralization: Reliance on centralized infrastructure undermines the decentralized nature of ZK-Rollups, increasing censorship risks.

High costs: Rollup operators must pay steep hardware or cloud service fees, raising operational costs.

The key to solving this lies in outsourcing proof generation to community-operated commodity hardware, achieving decentralization through "Community Proving" while maintaining performance and security (from Matter Labs’ January 2025 paper "Crowd Prove: Community Proving for ZK Rollups," arxiv.org).

2.1 ZKsync Opens Prover Nodes to the Community

In June 2024, ZKsync announced a five-step plan to decentralize its Prover, aiming to transition proof generation from centralized control to community participation. According to ZKsync’s official announcement, the community was invited to join the ZKsync Era Prover network to enhance the censorship resistance of the proof process. The launch of this plan marks ZKsync’s commitment to decentralization, with the opening of Prover nodes being a core component. Given the high hardware requirements for ZKsync nodes (40GB of GPU memory), participants are mostly professional teams with GPU resources, leaving individuals or small players with limited resources unable to participate.

By September 2024, ZKsync introduced the Prover API, allowing anyone to generate proofs and verify them against ZKsync via endpoints. This permissionless approach enhances ZKsync’s security and helps build a more decentralized proof ecosystem. Several ZK proof providers, including Fermah, Gevulot, and Lagrange, have joined the effort. According to Lagrange’s blog, by the end of 2024, Lagrange demonstrated the technical and economic feasibility of replacing centralized ZK stack provers with the Lagrange Proof Network for the first time. This milestone was achieved with support from top Lagrange operators, including P2P, Nethermind, Black Sand, EigenYields, and Staked. Since then, the Lagrange Proof Network (LPN) has reduced costs, increased throughput, and improved the reliability of decentralized proofs in the ZKsync Rollup ecosystem, with Matter Labs ultimately outsourcing up to 75% of its proofs to LPN.

2.2 Aztec Opens Prover Nodes to the Community

In early August, Aztec announced a request for prover integration to test decentralized proof outsourcing. Projects interested in generating proofs for the Aztec Network were invited to join ProverNet, a permissioned testnet for provers, with rewards offered through a competition. The competition took place in early September, with about 15 teams participating. Gevulot stood out in the ProverNet contest, leading in proof speed and the number of proven blocks.

3. ZKVM

zkVM (Zero-Knowledge Virtual Machine) is a virtual machine that integrates zero-knowledge proofs (ZKP), capable of executing arbitrary programs while generating proofs to verify computation correctness without revealing sensitive data. It supports general-purpose computing, allowing developers to write programs in high-level languages (e.g., Rust, C) and generate proofs without needing deep knowledge of complex zero-knowledge circuit design. This lowers the development barrier while enabling developers to perform complex off-chain computations (e.g., smart contracts, data processing) and generate proofs to verify the correctness of the results.

Take Delphinus Labs as an example: its network has produced 571,036 proofs, sourced from zkWASM games within its network. Delphinus Labs’ zkWASM requires 15 seconds to prove 1 million instructions on an RTX 4090 GPU, consuming 64GB of memory, highlighting its high hardware performance demands.

4. Other ZK Dapps

ZKML: Combines zero-knowledge proofs with machine learning to verify ML model computation results without exposing sensitive data, addressing data privacy and computational trustworthiness issues.

ZK Co-Processor (Zero-Knowledge Co-Processor): ZK co-processors outsource complex computations to off-chain processors via zero-knowledge proofs and verify the results on-chain, addressing the limited computational capacity of blockchains.

ZKtls: ZKtls enhances TLS (Transport Layer Security) with zero-knowledge proofs, allowing users to prove communication content with a server without exposing sensitive data, suitable for privacy-preserving identity verification and data access.

ZKEmail: ZKEmail uses zero-knowledge proofs to verify the authenticity or content of emails without revealing specific details, applicable to privacy-preserving identity verification and communication.

$Conclusion：$

The demand for ZK proofs is steadily growing and is expected to increase exponentially in the coming years, driving the development of new infrastructure and operators capable of efficiently generating computationally intensive proofs. However, nearly all current proof infrastructure relies on permissioned systems, with many applications limited to a single prover, posing risks of single points of failure and censorship. In the future, more applications will shift toward decentralized proof generation for several reasons: First, multiple provers enhance protocol liveness, ensuring the system remains reliable even if some provers are unavailable; second, increasing the number of provers strengthens censorship resistance, preventing a few provers from refusing to prove specific transactions; finally, competition among more provers will lead to faster and cheaper proof generation, fostering market growth.

As zero-knowledge proof (ZKP) technology becomes an indispensable cornerstone of the Web3 world, we are entering a new era prioritizing "computational trustworthiness." From validity proofs for Rollups to off-chain computation verification powered by zkVM, and privacy-AI fusion applications like zkEmail and ZKML, ZK is no longer just a cryptographic term but the execution and verification engine of the next-generation internet. According to Crypto.com's forecast, by 2030, Web3 services will need to execute nearly 90 billion zk proofs annually, with the market size for ZK computation expected to exceed $10 billion.

However, behind this explosive growth lies a harsh reality for ZK infrastructure: nearly all zk proof generation processes still heavily rely on centralized servers. Whether it’s zkRollup or zkVM, most mainstream projects currently depend on a handful of trusted entities to operate Prover nodes. While this architecture can勉强支撑 performance, it comes at the cost of high expenses, low resilience, censorship risks, and limited scalability—contradicting the decentralized ethos of blockchain.

Thus, building a ZK Prover Network has become a critical step in scaling zero-knowledge computation. By outsourcing proof tasks to community participants worldwide and constructing a permissionless, verifiable Prover network, we can achieve faster generation, stronger censorship resistance, lower costs, and true network-layer decentralization.

This article will unfold along two main threads:
·Part 1: Ethereum’s Snarkification Process and Vision for 3.0 — Ethereum is fully embracing ZK technology from the consensus layer (Beam Chain) to the execution layer (zkEVM/zkVM).

·Part 2: The Full Explosion of ZK Demand — With the surge in L2 networks, the rise of zkVM, and various ZK Dapps, the workload and diversity of ZK Provers are growing exponentially.

1. Ethereum’s Snarkification Process and Vision for 3.0

With the rapid development and application of zero-knowledge proof technology—including ZK-Rollups, zkVM, algorithmic optimization of proof systems, and hardware acceleration—Ethereum is gradually entering a new phase of "Snarkification." Reflecting on Ethereum’s evolution at EDCON 2024, Vitalik noted that in 2019, mainnet transaction confirmation times often stretched to tens of minutes, and zk-SNARK computation costs remained prohibitively high. Today, thanks to EIP-1559 optimizations and the widespread adoption of Layer 2, mainnet confirmation times have shrunk to 5–20 seconds, with Layer 2 achieving sub-second confirmations and transaction fees dropping below one cent. Looking ahead, as zk-SNARK performance continues to improve, Layer 2 TPS is expected to exceed 100,000, while significantly reducing computation and storage overhead.

Vitalik believes ZKP has evolved from a "theoretical technology" five years ago into a widely available core infrastructure for both on-chain and off-chain use today, with an especially urgent need for on-chain privacy and efficient interaction in non-financial scenarios like social applications. He also called on the Ethereum ecosystem to confront privacy issues—current solutions like the Railway privacy wallet still have high usage barriers, limiting the widespread adoption of privacy features. Fortunately, the evolution of zk-SNARK has given rise to more feasible solutions, such as local key backups and zk-encapsulated centralized identities (e.g., zk-email, ZKTLS, ZKDID), gradually expanding the boundaries of privacy applications. In the future, ZK is expected to become ubiquitous on everyone’s devices.

In October 2024, Vitalik proposed long-term goals for Ethereum block validation, aiming to achieve a more efficient and decentralized verification mechanism to support a broader range of clients (including light clients) and cross-chain scenarios. The specific goals include two aspects:

Minimalist block downloads: Through Data Availability Sampling (DAS) technology, nodes only need to download a tiny fraction of block data to verify its integrity.

Lightweight block validation: The verification process only requires checking a small proof, which can be quickly validated in resource-constrained environments—such as mobile clients, browser wallets, or even cross-chain validation on other blockchains—thereby enhancing Ethereum’s decentralization and scalability.

To realize this vision, improvements are needed at two core levels of Ethereum: the consensus layer (i.e., the proof-of-stake mechanism, currently implemented by the Beacon Chain) and the execution layer.

Consensus Layer Improvements: Beam Chain

Using zero-knowledge proofs to reduce the computational and communication burden on validators, lowering the entry threshold for becoming a validator, allowing more validators to join the network to enhance security, reduce computational redundancy, and improve overall network efficiency. At the same time, ZK-ifying the consensus layer requires encapsulating processes like state transitions and validator signature aggregation into SNARK or STARK proofs to simplify verification. This is a complex technical challenge in itself. Ethereum’s current Beacon Chain, designed between 2018 and 2020, carries technical debt, such as long block times (12 seconds) and finality delays (about 13 minutes). The Beam Chain proposal leverages zero-knowledge proofs as a core technology to optimize the consensus mechanism, boost efficiency, and prepare for future quantum resistance, including the following aspects:

1. SNARKification of state transitions: "ZK-ifying" the State Transition Function (STF) by using SNARKs to prove the correctness of state updates (e.g., account balances, validator registries, signature aggregation).

2. Achieving single-slot finality through ZK

3. Quantum-resistant ZK upgrades: Introducing quantum-resistant STARKs to address the threat of quantum computing.

4. Integration of zkVM: Verifying complex consensus logic through zkVM

5. Optimizing shard data verification with ZK: Each shard generates state updates and SNARK proofs, with the main chain only verifying them.

Execution Layer: EVM Validity Proofs

ZK proofs for EVM execution are already widely used in Layer 2 (L2) solutions, such as zk-Rollups (e.g., zkSync, StarkNet). These solutions ensure trustworthy computation results by generating validity proofs while submitting data to Layer 1. However, applying EVM validity proofs directly to Layer 1 still faces two major bottlenecks:

Insufficient security: Secure EVM validity proofs must ensure that SNARKs accurately verify EVM computations without errors. Proofs in L2 are typically tailored to specific scenarios with smaller computation scales, making security manageable. On L1, however, the EVM must handle more complex computations (e.g., DeFi protocols), increasing the attack surface of the proof system.

----Solution: Multi-prover and formal verification: A multi-prover approach means having multiple independently written validity proof implementations, similar to having multiple clients. If a sufficiently large subset of these implementations proves a block, the client accepts it. Formal verification involves using tools typically employed to prove mathematical theorems to ensure that validity proofs only accept inputs that correctly execute the underlying EVM specification.

Proof generation time too long: For instance, generating a SNARK proof for a complex smart contract execution currently takes seconds to minutes, while L1’s block production target is 4–12 seconds (with Beam Chain aiming to reduce it to 4 seconds). We are still far from this goal.

-----Solution: ① Parallelization: The fastest current EVM provers can prove a typical Ethereum block in about 15 seconds. To further reduce this time, parallelization techniques are needed to break EVM computation into multiple parts, generating proofs in parallel at the opcode level. This could be based on the EVM itself or a lower-level virtual machine (e.g., RISC-V). ② Proof system optimization: Emerging proof systems (e.g., Orion, Binius, GKR) promise to significantly reduce proof times for general computation and lower computational overhead. For example, Binius’s binary field optimization could boost proof generation speed by 2–3 times, while GKR’s layered proving further reduces the overhead of complex computations.

The Demand Side of ZK Proofs

If Ethereum 3.0 represents a structural demand for ZK from the ground up, the explosion on the demand side is a pressing need driven by real-world dynamics.

1. The Growing Number of L2 Networks: Surge in Quantity and Application Expansion

The ZK technology stack has successively released production-ready chain development toolkits (e.g., Polygon’s SDK, zkSync’s Elastic Chain, OP Stack), making it easier to deploy new validity Rollups or Validiums, significantly increasing the number of L2s. As Ethereum’s modularization accelerates, the computational demand for zk proofs will rise rapidly, for example:

Enterprise-grade L2s: For instance, Hong Kong’s HashKey launched its own L2, HashKey Chain, to expand application scenarios; Deutsche Bank adopted zkSync technology for Project Dama 2, enabling more efficient asset service transactions; Coinbase’s Base.

Scenario-specific L2/L3s: For example, Sophon (a Validium L2 based on Avail’s data availability layer) focuses on gaming scenarios.

RaaS (Rollup as a Service): For example, Zeeve supports zkSync with one-click deployment tools, offering over 40 integrated services (e.g., sequencers, block explorers), significantly lowering the development barrier.

Equilibrium Research predicts that by the end of 2025, the number of L2 solutions on Ethereum will exceed 2,000, with ZK L2s dominating. ZK-Rollups will surpass Optimistic Rollups, and with the development and adoption of zkVM—especially with the emergence of Succinct—we are already seeing Optimistic solutions gradually shifting toward ZK.

2. The Decentralization Trend of L2

ZK-Rollups aggregate large batches of transactions and submit a concise proof to Layer 1, significantly boosting transaction throughput and reducing costs.

However, generating these proofs is computationally intensive, and currently, it largely relies on centralized infrastructure (e.g., dedicated servers or cloud services). This centralized approach has the following issues:

Performance bottlenecks: Centralized Provers may struggle to handle high transaction volumes, causing delays.

Limited decentralization: Reliance on centralized infrastructure undermines the decentralized nature of ZK-Rollups, increasing censorship risks.

High costs: Rollup operators must pay steep hardware or cloud service fees, raising operational costs.

The key to solving this lies in outsourcing proof generation to community-operated commodity hardware, achieving decentralization through "Community Proving" while maintaining performance and security (from Matter Labs’ January 2025 paper "Crowd Prove: Community Proving for ZK Rollups," arxiv.org).

2.1 ZKsync Opens Prover Nodes to the Community

In June 2024, ZKsync announced a five-step plan to decentralize its Prover, aiming to transition proof generation from centralized control to community participation. According to ZKsync’s official announcement, the community was invited to join the ZKsync Era Prover network to enhance the censorship resistance of the proof process. The launch of this plan marks ZKsync’s commitment to decentralization, with the opening of Prover nodes being a core component. Given the high hardware requirements for ZKsync nodes (40GB of GPU memory), participants are mostly professional teams with GPU resources, leaving individuals or small players with limited resources unable to participate.

By September 2024, ZKsync introduced the Prover API, allowing anyone to generate proofs and verify them against ZKsync via endpoints. This permissionless approach enhances ZKsync’s security and helps build a more decentralized proof ecosystem. Several ZK proof providers, including Fermah, Gevulot, and Lagrange, have joined the effort. According to Lagrange’s blog, by the end of 2024, Lagrange demonstrated the technical and economic feasibility of replacing centralized ZK stack provers with the Lagrange Proof Network for the first time. This milestone was achieved with support from top Lagrange operators, including P2P, Nethermind, Black Sand, EigenYields, and Staked. Since then, the Lagrange Proof Network (LPN) has reduced costs, increased throughput, and improved the reliability of decentralized proofs in the ZKsync Rollup ecosystem, with Matter Labs ultimately outsourcing up to 75% of its proofs to LPN.

2.2 Aztec Opens Prover Nodes to the Community

In early August, Aztec announced a request for prover integration to test decentralized proof outsourcing. Projects interested in generating proofs for the Aztec Network were invited to join ProverNet, a permissioned testnet for provers, with rewards offered through a competition. The competition took place in early September, with about 15 teams participating. Gevulot stood out in the ProverNet contest, leading in proof speed and the number of proven blocks.

3. ZKVM

zkVM (Zero-Knowledge Virtual Machine) is a virtual machine that integrates zero-knowledge proofs (ZKP), capable of executing arbitrary programs while generating proofs to verify computation correctness without revealing sensitive data. It supports general-purpose computing, allowing developers to write programs in high-level languages (e.g., Rust, C) and generate proofs without needing deep knowledge of complex zero-knowledge circuit design. This lowers the development barrier while enabling developers to perform complex off-chain computations (e.g., smart contracts, data processing) and generate proofs to verify the correctness of the results.

Take Delphinus Labs as an example: its network has produced 571,036 proofs, sourced from zkWASM games within its network. Delphinus Labs’ zkWASM requires 15 seconds to prove 1 million instructions on an RTX 4090 GPU, consuming 64GB of memory, highlighting its high hardware performance demands.

4. Other ZK Dapps

ZKML: Combines zero-knowledge proofs with machine learning to verify ML model computation results without exposing sensitive data, addressing data privacy and computational trustworthiness issues.

ZK Co-Processor (Zero-Knowledge Co-Processor): ZK co-processors outsource complex computations to off-chain processors via zero-knowledge proofs and verify the results on-chain, addressing the limited computational capacity of blockchains.

ZKtls: ZKtls enhances TLS (Transport Layer Security) with zero-knowledge proofs, allowing users to prove communication content with a server without exposing sensitive data, suitable for privacy-preserving identity verification and data access.

ZKEmail: ZKEmail uses zero-knowledge proofs to verify the authenticity or content of emails without revealing specific details, applicable to privacy-preserving identity verification and communication.

$Conclusion：$

The demand for ZK proofs is steadily growing and is expected to increase exponentially in the coming years, driving the development of new infrastructure and operators capable of efficiently generating computationally intensive proofs. However, nearly all current proof infrastructure relies on permissioned systems, with many applications limited to a single prover, posing risks of single points of failure and censorship. In the future, more applications will shift toward decentralized proof generation for several reasons: First, multiple provers enhance protocol liveness, ensuring the system remains reliable even if some provers are unavailable; second, increasing the number of provers strengthens censorship resistance, preventing a few provers from refusing to prove specific transactions; finally, competition among more provers will lead to faster and cheaper proof generation, fostering market growth.