Brevis Zero-Knowledge Ethereum Virtual Machine: 12 seconds verification 99.6% Block 10K TPS in sight

Ethereum scaling company Brevis launched Pico Prism on October 15, a new type of Zero-Knowledge Ethereum Virtual Machine (zkEVM) for instant proofs. This technology utilizes 64 Nvidia RTX 5090 consumer gaming GPUs, achieving 99.6% instant proof in under 12 seconds during tests in September, replacing previous solutions that required expensive supercomputers.

The roadmap of Brevis Zero-Knowledge Ethereum Virtual Machine indicates that within the next few months, 99% instant verification will be achieved using fewer than 16 RTX 5090 GPUs. This breakthrough allows the computationally expensive proofs to keep pace with block production speeds, making lightweight verification feasible for the first time, laying the groundwork for Ethereum's goal of 10,000 transactions per second (10K TPS), and opening up the future of mobile node verification.

The Technological Revolution from Supercomputers to Game GPUs

(Source: Brevis)

The core breakthrough of Brevis's Zero-Knowledge Ethereum Virtual Machine lies in the drastic reduction of hardware requirements. Traditionally, the cryptographic proofs required to generate blocks correctly were expensive supercomputers and specialized data centers. Pico Prism has changed this landscape, enabling instant proofs with just 64 Nvidia RTX 5090 graphics cards. The RTX 5090 is Nvidia's current flagship model for gaming, with a single card priced at about $2,000, making the total cost of 64 cards approximately $128,000, far lower than the millions of dollars typically required for professional supercomputers.

Real-Time Proving (RTP) refers to the speed of generating cryptographic proofs of correct block execution being faster than the speed of producing new blocks. The block time for Ethereum is approximately 12 seconds; if the proof generation time exceeds this threshold, the prover will never catch up with the latest state of the blockchain. Pico Prism achieved 99.6% real-time proving in under 12 seconds during tests in September, meaning that the vast majority of blocks can be proven before the next block is produced.

What's even more exciting is Brevis's roadmap. The company plans to achieve 99% instant verification using fewer than 16 RTX 5090 GPUs in the coming months. With a reduction from 64 to 16 nodes, the hardware cost will drop from $128,000 to $32,000, making this price level affordable for small and medium-sized validators and developers. The total power consumption of a 16 GPU cluster is less than 10kW, which is equivalent to 3-5 times that of an average household, far lower than the energy consumption of traditional data centers.

This hardware democratization has far-reaching implications. When verification becomes cheap and accessible, more nodes can participate in network validation, enhancing the level of decentralization. Currently, Ethereum validators need to re-execute every transaction to validate blocks, which requires expensive hardware and creates fundamental bottlenecks. Brevis Zero-Knowledge Ethereum Virtual Machine breaks this model: a prover generates a proof, and others can verify it in a matter of milliseconds, with verification costs being almost negligible.

Ethereum 10,000 TPS roadmap technical path

According to the Ethereum roadmap, validators will transition from re-executing transactions to simply validating zero-knowledge proofs, allowing the base layer to reach 10,000 transactions per second. Ryan Sean Adams from Bankless stated, “If it scales at a rate of 3 times per year, by April 2029, the TPS of Ethereum L1 will reach 10,000 TPS.” This prediction is based on the current pace of technological advancement and the upcoming protocol upgrades.

Ethereum Foundation researcher Justin Drake explained that the Fusaka upgrade for Ethereum is expected to take place in December, which will simplify instant proof. EIP-7825 limits the gas usage per transaction, allowing for more parallel proofs through sub-blocks. This parallelization is key to achieving high TPS, as it enables multiple provers to simultaneously process different parts of the same block, significantly enhancing the overall proof speed.

Justin Drake added, “By the end of this year, several teams will demonstrate each L1 EVM block on a 16-GPU cluster, with a total power consumption of less than 10kW.” This means that the Brevis Zero-Knowledge Ethereum Virtual Machine is not the only team working in this direction; competition will further drive technological advancement and cost reduction. The parallel development by multiple teams also lowers the risk of single points of failure, ensuring the reliability of the Ethereum scaling roadmap.

From a technical architecture perspective, achieving 10,000 TPS requires collaboration at three levels. The first level involves instant proof technologies such as Brevis Zero-Knowledge Ethereum Virtual Machine, ensuring that proof generation does not become a bottleneck. The second level is about optimizing at the protocol level, such as the parallelization support of EIP-7825. The third level focuses on the continuous improvement of hardware and network infrastructure, ensuring that the generation, dissemination, and verification of proofs can be completed efficiently. All three are essential, and Pico Prism's breakthrough addresses the critical challenges of the first level.

The Future Vision of Mobile as a Node

The Ethereum Foundation stated that this is “a significant step towards the future of Ethereum,” adding that “ZK technologies like Pico Prism will enable Ethereum to scale to meet global demand while still maintaining trust and decentralization.” Tech entrepreneur Mike Warner remarked, “The future of mobile phones as nodes is about to become a reality.” This vision sounds like science fiction, but the technological pathway is already clearly visible.

When verification only requires checking zero-knowledge proofs instead of re-executing all transactions, the computational requirements are significantly reduced. Modern smartphones have powerful processors and ample storage space, theoretically sufficient to verify zero-knowledge proofs. If this vision is realized, anyone can participate in Ethereum network verification using the smartphone in their pocket, which would be the ultimate form of decentralization.

Ryan Sean Adams stated that Ethereum is transforming into a zk chain. He explained that Layer 1 will run global DeFi with a TPS of 10,000 for its large blocks, and nodes will run on mobile phones, while Layer 2 will run everything else. This architecture positions Ethereum as a global settlement layer, handling the most critical financial transactions, while Layer-2 solutions handle high-frequency low-value transactions, such as applications in gaming, social media, and more.

The implementation of mobile nodes needs to overcome multiple challenges. The first is the bandwidth issue; mobile phones typically rely on mobile networks, which have less bandwidth and stability compared to fixed broadband. The second is battery life; continuous validation consumes a significant amount of power. Third is storage space; although validation does not require the complete blockchain history, a certain amount of state data is still needed. These challenges are not insurmountable, as lightweight client protocols and state expiration mechanisms are being developed, which will further lower the operational threshold for nodes.

The Holy Grail and Challenges of Blockchain Scalability

This is essentially the holy grail of blockchain: achieving massive scalability without sacrificing decentralization or security. Traditional blockchains face the “impossible triangle”: decentralization, security, and scalability can only choose two out of three. Bitcoin and Ethereum chose decentralization and security, sacrificing scalability. Many high TPS public blockchains chose scalability but compromised on decentralization and security.

Brevis Zero-Knowledge Ethereum Virtual Machine and Pico Prism provide a path to break through the “impossible triangle.” Zero-knowledge proof technology allows verifiers to confirm the correctness of blocks without re-executing transactions, fundamentally changing the economics of verification. When verification costs drop to almost zero, more verifiers can participate (enhancing decentralization) while handling more transactions (enhancing scalability), and the cryptographic guarantees of zero-knowledge proofs ensure security.

However, challenges still exist. The first is the centralized risk of proof generation. Although verification has become easier, proof generation still requires specialized hardware. If only a few entities can generate proofs, it may create a new centralized bottleneck. Brevis's goal of 16 GPUs is an important step in alleviating this issue, but further reductions in the threshold are still needed. Secondly, there is the storage and transmission of proof data. Zero-Knowledge proofs themselves need to occupy block space, which may become a new scalability bottleneck.

The third is the design of economic incentives. When validation becomes easy, how to incentivize enough nodes to participate in validation? Validation rewards may need to be adjusted to reflect the new cost structure. The fourth is the compatibility issue during the transition period; migrating from the current architecture to a zk-driven architecture requires coordinated upgrades, and any misstep could lead to network splits. These challenges are not insurmountable, but they require careful engineering design and community coordination.

Frequently Asked Questions Q&A

Q1: How does Brevis Zero-Knowledge Ethereum Virtual Machine achieve 99.6% instant proof within 12 seconds?

A: The core technology is Pico Prism zkEVM, using 64 Nvidia RTX 5090 gaming GPUs instead of supercomputers. Through optimized zero-knowledge proof algorithms and parallel processing, cryptographic proofs of block execution are generated in less than 12 seconds. Instant proof means that the proof generation speed is faster than the block production speed (12 seconds), with 99.6% indicating that the vast majority of blocks can be completed instantly. In the coming months, this will be reduced to 16 GPUs to achieve 99% proof, significantly lowering the hardware threshold.

Q2: How to reduce from 64 GPUs to 16 GPUs? Where is the technical bottleneck?

A: The optimization is mainly through three directions: algorithm improvements to reduce computational complexity, parallelization strategies to enhance single GPU efficiency, and protocol upgrades such as EIP-7825 which limit the gas for a single transaction to facilitate easier parallel proof. Currently, a configuration of 64 GPUs is conservative to ensure a 99.6% success rate, and with software optimization and the Fusaka upgrade in December, 16 GPUs can achieve 99%. The bottleneck lies in the complexity of the proof circuit and memory bandwidth, with the high bandwidth memory of the Nvidia RTX 5090 being crucial.

Q3: What is the feasibility of Ethereum reaching 10,000 TPS by 2029?

A: Based on the forecast of a 3x expansion rate per year, the technical path is clear, but there are execution risks. Key milestones include: by the end of 2025, multiple teams achieving 16-GPU cluster real-time proof, the Fusaka upgrade achieving parallelization, and validators transitioning from re-execution to verifying ZK proofs. If these are successfully achieved, 10,000 TPS is feasible. The risks lie in protocol upgrade coordination, economic incentive design, and unexpected technical hurdles. Layer-2 scaling can provide a buffer; even if L1 does not reach 10K, the overall ecosystem throughput can still be significantly increased.

Q4: When will mobile devices be able to function as nodes? How can ordinary users participate?

A: The technical foundation is in place, but full implementation will require 2-3 years. Current mobile hardware is sufficient to verify ZK proofs, and the challenges lie in bandwidth, battery, and storage optimization. The development of lightweight client protocols and state expiration mechanisms is underway, which will further lower the barriers to entry. Ordinary users can participate in verification by running lightweight node apps without needing the complete blockchain data. It is recommended to pay attention to the Ethereum Foundation's progress on lightweight client development and Brevis's mobile adaptation plans.

Q5: Will zero-knowledge proofs introduce new centralization risks?

A: The existence of centralized risk in proof generation is indeed proven, as it requires specialized hardware. However, Brevis's goal of 16 GPUs (costing approximately $32,000) makes it affordable for small and medium-sized participants, far superior to supercomputer solutions. The key is to diversify the proof ecosystem to avoid monopoly by a single entity. The Ethereum community is developing proof aggregation and decentralized proof generation protocols to further lower the threshold. Verification remains extremely decentralized (any device can verify), which is the most important security guarantee.