Zksync recently launched Airbender, which is quite interesting. Airbender is a zero-knowledge virtual machine (zkVM) based on RISC-V, which is the direction that Vitalik mentioned recently (Vitalik believes that RISC-V could replace the existing EVM to improve Ethereum's efficiency).
What are its main features? Currently, according to Zkync's description, it mainly balances decentralized proof and cost reduction. Airbender can achieve an end-to-end proof of an Ethereum block in 35 seconds on a single GPU (such as the NVIDIA H100) (in streamlined mode, this can be reduced to 17 seconds). According to its official benchmarks, its basic proof speed reaches 21.8MHz, exceeding SP1 Turbo (3.45MHz) and RiscZero (1.1 MHz). Traditional systems generally require 50-160 GPUs to achieve similar results within 12 seconds. Airbender only requires a single GPU, which is beneficial for decentralized proof. Succinct's SP1 Hypercube zkVM demonstrated an average time of 10.3 seconds for an Ethereum block last month (10.8 seconds for a single block), better than Airbender's 35 seconds. However, Zksync's Airbender has advantages in single GPU performance and open-source characteristics: it better meets the needs of decentralized proof scenarios and is more cost-effective. In other words, Airbender can alleviate the dependence on large-scale GPU clusters and reduce the risk of proof centralization.
In terms of cost, Zksync claims that its Airbender will reduce proof costs to about $0.0001 per transaction. If this is true, it could greatly reduce costs for high-frequency trading, micropayments, gaming, and decentralized social applications. Perhaps this will open up possibilities for larger-scale on-chain application scenario exploration, such as on-chain gaming and payments between AI agents. Compared to other projects, it may have an advantage of more than ten times in terms of cost. If this is the case, then the cost advantage combined with the advantages of decentralized proof will be more attractive to adopters in certain scenarios, such as on-chain gaming, while in some high-value transactions and scenarios that are not sensitive to fees, other zkVMs will also have their appeal.
Overall, Zksync's launch of Airbender cannot be considered a disruptive innovation yet, but it can be regarded as an important development in the Ethereum ecosystem. The ultimate effect will depend on the adoption of the Ethereum ecosystem itself. Since there is currently no synergy, the specific outcome remains to be seen.