Written by: Yue Xiaoyu
Web3Caff founder Gootor raised a very important point; here are some further thoughts:
Ethereum is being ZKified, and ZK provers require hardware acceleration; just like Bitcoin mining has hardware dependencies, ZK mining may become a new form of PoW.
Ethereum's economic model may ultimately evolve into a coexistence of PoS and PoW, forming a new 'hybrid consensus' model.
Next, let's look at the underlying logic of the above assumptions.
1. Ethereum is being ZKified
The basic principle of ZK (Zero-Knowledge Proof) is: one party can prove to another that a statement is true without revealing any other information.
This is a very promising technology.
ZK technology is currently mainly used in the blockchain field for scalability.
Vitalik Buterin has been actively promoting ZK technology, and due to his willingness and inclination, the Ethereum Foundation has also invested in various ZK projects.
However, ZK projects in this infrastructure layer have a relatively long cycle, so what ordinary users can currently directly perceive are various ZK Rollups, i.e., ZK-based Ethereum Layer2 solutions.
Typical representatives include zkSync, Starknet, Polygon zkEVM, Scroll, Linea, etc.
These are all about integrating ZK technology into Ethereum's execution layer.
Ethereum itself is also gradually ZKifying.
ZKification of the consensus layer (such as the Beam Chain proposal): By ZKifying the verification process of consensus operations, the consensus process can be simplified and accelerated.
ZKification of the data layer: Only storing proofs without storing all transaction data can significantly reduce on-chain storage requirements.
Through these means, Ethereum can significantly enhance the performance of its network, including transaction processing speed, costs, and privacy protection while maintaining its decentralization and security.
The introduction of ZK technology is seen as part of Ethereum's long-term expansion strategy, capable of addressing higher user demand and application complexity.
2. Why does ZK need to be combined with hardware?
The numerous advantages of ZK mentioned above face the biggest obstacle, which is the performance issue of generating ZKPs (Zero-Knowledge Proofs).
The operational mechanism of ZK is to enable validators to combine multiple transactions into a single, concise proof and then submit the transaction to the mainnet, ultimately achieving scalability for the mainnet.
However, ZK also has limitations, as the process of generating proofs is resource-intensive in terms of time and effort.
Due to the need for many complex mathematical operations, such as exponentiation, inversion, and bilinear pairing calculations, the creation of proofs is typically slowed down.
Simply relying on software optimization cannot further enhance performance, which necessitates combining hardware. By using dedicated hardware for acceleration, the proof generation speed can be improved from hours to seconds, which would be a significant boost.
Therefore, ZK has hardware dependencies and requires dedicated chips.
As the application of ZKP technology in privacy computing, blockchain scalability, and other areas gradually increases, the speed and efficiency of generating ZKPs have become bottlenecks, and the market's demand for hardware is also growing.
3. How can PoS and ZK mining be combined?
If hardware acceleration for ZK is adopted, it could lead to hardware mining, resulting in a new economic model.
What is ZK mining?
Traditional PoW (Proof of Work) mining relies on solving complex mathematical problems, while ZK hardware mining focuses on the ability to quickly generate ZKPs.
Similar to PoW mining, participants can compete for the speed and efficiency of generating ZK proofs, with rewards going to the participant who generates valid proofs first.
How exactly can PoS and ZK be combined?
In PoS systems, validators (i.e., stakers) are responsible for validating transactions and proposing new blocks. ZK proofs can be used to demonstrate the correctness of a block or a series of transactions, enhancing security and reducing validation costs.
Validators generate ZKPs as additional proof of their work while performing their PoS duties. Validators who successfully generate and submit valid ZKPs can receive additional rewards beyond the existing block rewards.
This could be ETH or some form of reward within the network (such as a specific token or a portion of gas fees).
The mainnet block rewards can be divided into two parts, forming a layered reward system: one part of the rewards is used for traditional PoS validation, while the other part is distributed based on the node's contribution in generating ZK proofs.
It can thus be seen that ZK can enhance PoS validation, while ZK mining can serve as an extension of PoS staking.
Stakers can earn additional rewards by participating in ZK-related work in addition to staking ETH for PoS rewards. For example, stakers can run ZK proof generation nodes, providing computing resources to support the application of ZK technology.
However, this only presents a possibility; to integrate ZK technology into Ethereum's existing PoS mechanism would require modifications or expansions to the current PoS protocol to accommodate the role of ZK proofs.
Moreover, a fair incentive model must be designed to ensure that it neither undermines the incentive effect of PoS nor fails to promote the application of ZK technology.
4. Will ZK mining introduce new positive externalities?
After Ethereum transitioned from PoW to PoS, one of the most criticized points was the lack of externalities introduced into this 'closed' economic system.
PoW has externalities, requiring energy and hardware, thereby extending the blockchain's economic activities to outside the blockchain, impacting the real world.
In PoS systems, the eligibility to validate blocks is primarily based on the number of tokens held (i.e., staking), and economic activities are more concentrated on staking and transactions within the network, rather than involving extensive mining production, electricity markets, and other external industries like PoW.
ZK acceleration hardware, like PoW mining machines, requires dedicated hardware, but the problem of 'inefficient power consumption' would be reduced because PoW is about solving mathematical problems to compete for block rights, while ZK mining focuses on generating valid proofs to validate transactions, eliminating the need for different nodes to perform repetitive, ineffective labor.
5. AI and ZK shared computing power
Furthermore, a ZK computing power market will emerge, and AI computing power and ZK computing power can even be shared.
There may be integrated platforms in the market providing comprehensive computing services that support the computational needs of both AI and ZK tasks.
Users can handle high-performance computing AI tasks and tasks requiring ZK proof generation on the same platform.
Protocol-level integration can also be conducted: creating smart contracts or protocols that allow AI computing power providers to participate in the generation of ZK proofs or utilize ZK technology to validate AI computation results.
Computing power can also be dynamically allocated: dynamically allocating computing resources based on task requirements. For example, when AI task demands are low, these resources can be used to generate ZK proofs, and vice versa.
6. Leading players in ZK hardware acceleration
Currently, the leading project in the ZK hardware acceleration track is Cysic.
Cysic focuses on designing and producing application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), and GPU acceleration solutions to speed up the ZKP generation process.
Cysic has designed two hardware products:
ZK Air: A lightweight ZK device, roughly the size of a laptop power adapter, with computing power equivalent to 10 RTX 4090 GPUs, suitable for individual users.
ZK Pro: More powerful devices, with performance equivalent to 50 RTX 4090 GPUs, designed for large ZK projects such as zkRollup and zkML.
In addition to hardware, Cysic also provides a set of CUDA libraries that support various ZKP systems such as Plonky2, Gnark, Halo2, RapidSnark, etc. These libraries can significantly enhance performance, improving by at least 50% compared to open-source frameworks.
At the same time, Cysic has an ecosystem, Cysic Network, primarily aimed at integrating and managing community computing resources, connecting ZK project parties, computing power providers, and community validators to form a reliable computing power network.
In simple terms, Cysic Network is a ZK mining pool and SaaS platform.
Currently, Cysic has already collaborated with multiple zkRollup projects, demonstrating the practicality and market demand for its products.
In addition to Cysic, there is another notable project to mention: Scroll
Unexpectedly, the 'abandoned' and 'notorious' Scroll is actually making significant advancements in ZK hardware acceleration.
Compared to other ZK Rollup projects, Scroll can be said to take a different approach by starting with hardware, developing dedicated ZK acceleration hardware GUP, and establishing its own competitive barrier, which is Scroll's greatest differentiating competitive advantage.
Looking at this point alone, Scroll is indeed a pretty hardcore project.
However, Scroll and Cysic have different positioning: for Scroll, hardware acceleration is a part of its overall Layer2 solution; while Cysic is dedicated to providing hardware acceleration services for the entire ZKP ecosystem.
7. Conclusion
Finally, let's summarize the content of this article:
Ethereum is being ZKified: Ethereum is using zero-knowledge proof (ZK) technology to achieve scalability and improve privacy, having applied ZK technology at the consensus and data layers to enhance network performance.
ZK requires hardware acceleration: Due to the significant computational resources required to generate zero-knowledge proofs, hardware acceleration (such as ASICs, FPGAs) is seen as a necessary means to improve efficiency.
The combination of PoS and ZK mining: proposing a new economic model where ZK mining is an extension of PoS, allowing stakers to earn additional rewards by providing ZK computing power, forming a model where PoS and PoW coexist.
ZK mining introduces new positive externalities: compared to PoW, ZK mining may reduce inefficient energy consumption, but it still requires dedicated hardware, fostering incentives for hardware production and technological innovation.
AI and ZK computing power sharing: AI and ZK computing tasks may share hardware resources, forming a comprehensive computing service market.
Market leading projects: Cysic and Scroll are leaders in the field of ZK hardware acceleration, with the former focusing on providing hardware for the ZK ecosystem and the latter incorporating it as part of its Layer2 solution.
Finally, the ZK hardware acceleration track has already begun to take shape. As Ethereum further ZKifies, ZK hardware acceleration will become increasingly important and may even reverse the economic model of Ethereum, warranting our continued attention to this track.
