Vitalik praised ZKsync, which has long created the fastest zkVM

Original Author: Eric, Foresight News

On November 1, Vitalik quoted a tweet from the founder of ZKsync regarding the ZKsync Atlas upgrade and praised ZKsync for doing a lot of "underestimated but valuable work for the Ethereum ecosystem."

The market quickly reacted to Vitalik's words, with the price of ZK rising more than 2.5 times over the weekend. Tokens in the ZK ecosystem, including ALT (AltLayer), STRK (Starknet), SCR (Scroll), and MINA (Mina), also saw significant increases.

After learning about the ZKsync Atlas upgrade, we found that what ZKsync has done may indeed be underestimated.

Fast, Small but Expensive ZKP

The Ethereum Foundation has been promoting ZKP (Zero-Knowledge Proof) for a long time, essentially aiming to solve the problems of slow verification speed and large amounts of data for verification.

ZKP is essentially a mathematical probability problem. To illustrate its principle with an imperfect example: suppose someone claims to have solved the "Four Color Theorem." How can we determine that this person has indeed solved it without fully disclosing their solution? The solution of zero-knowledge proof is to select some parts of the entire graph and prove that there are no adjacent regions with the same color. When the number of selected parts reaches a certain value, it can be proven that the probability of this person having solved the Four Color Theorem has reached 99.99…%. At this point, we have achieved proof of "indeed solved the Four Color Theorem" without understanding the whole picture.

This is what people often hear about: "proving that something has indeed been done without understanding how it was done" in zero-knowledge proof. The reason for vigorously promoting ZKP in the Ethereum ecosystem is that the theoretical speed limit of ZKP is much faster than that of transaction-by-transaction proof, and the generated proof itself has a very small data size.

The speed is fast because ZKP does not need to understand the whole picture; it only needs to challenge. For example, to verify an Ethereum block, the current method is for each node to verify whether each transaction's execution address has sufficient balance and other basic issues. However, if only one node verifies each transaction through ZKP and then generates a "proof," other nodes only need to verify that the "proof" itself is reliable. More importantly, the data size of this "proof" is very small, so the speed of its transmission and verification is extremely fast, and the cost of storing data is lower.

As for why this all-beneficial technology is not widely used, it is simply because it is too expensive.

Although ZKP does not require reproducing all processes, the challenges themselves consume a lot of computing power. If one were to crazily stack GPUs like in an AI arms race, faster speeds could be achieved, but not everyone can bear such costs. However, if it is possible to reduce the required computing power and the time to generate proofs under low computing power to a certain extent through algorithmic and engineering innovations, achieving a balance between "price increases driven by technological innovation introducing more applications" and "the cost of building nodes to purchase GPUs," then this can be done.

Therefore, many ZK concept projects or open-source developers in the Ethereum ecosystem focus on combining ZKP with Ethereum primarily on generating ZK proofs at lower costs and faster speeds under low costs. Recently, the Brevis team achieved an average proof time of 6.9 seconds for Ethereum blocks using only half the cost of the SP1 Hypercube solution (64 RTX 5090 GPUs), with 99.6% of proof times being less than the current average block time of Ethereum (within 12 seconds), which is why it received applause from the Ethereum community.

Although the cost of GPUs still exceeds $100,000, at least the proof speed has now dropped to a level comparable to that without ZKP, and the next task for everyone is to reduce costs.

Atlas Upgrade Achieves 1 Second ZK Finality

Perhaps many people do not know that the open-source zkVM ZKsync Airbender launched by ZKsync is the fastest single GPU verification zkVM. According to data from Ethproofs, using a single 4090, ZKsync Airbender has an average verification time of 51 seconds, costing less than a cent, both of which are the best results among zkVMs.

According to data provided by ZKsync itself, excluding recursion, Airbender's average time to verify the Ethereum mainnet using a single H100 and ZKsync OS storage model is 17 seconds. Even including recursion, the total average time is only about 35 seconds. ZKsync believes this is much better than needing dozens of GPUs to achieve verification within 12 seconds. However, since the current average time with only two GPUs is 22.2 seconds, the actual quality has yet to be determined.

All of this is not solely due to Airbender; algorithmic and engineering optimizations are just one aspect. The deep integration with the ZKsync tech stack is key to maximizing the effect. More importantly, it indicates that achieving real-time proof of the Ethereum mainnet using a single GPU is possible.

At the end of June, ZKsync launched Airbender, and on the second-to-last day of the National Day holiday, the Atlas upgrade went live. This upgrade, which integrated Airbender, significantly improved ZKsync's throughput, confirmation speed, and cost.

In terms of throughput, ZKsync optimized the sorter through engineering: it maximized the reduction of consumption caused by synchronization through independent asynchronous components; it separated the states required by the virtual machine, the states required by the API, and the states required to generate zero-knowledge proofs or verify zero-knowledge proofs on the L1 layer, thereby reducing unnecessary overhead of components.

Through ZKsync's field tests, the TPS for high-frequency price updates, stablecoin transfers in payment scenarios, and native ETH transfers reached 23k, 15k, and 43k, respectively.

Another significant qualitative change comes from Airbender, which helps ZKsync achieve 1-second block confirmation and a cost of $0.0001 per transaction. Unlike verifying mainnet blocks, ZKsync only verifies the validity of state transitions, so the computational load is much smaller than that of verifying mainnet blocks. Although transactions achieving ZK finality still need to be verified on the mainnet to achieve L1 finality, having ZK verification indicates the validity of the transaction, while L1 finality is more like a procedural guarantee.

In other words, transactions executed on ZKsync only need ZKP verification to be fully confirmed as valid, and with significantly reduced costs, ZKsync has achieved, in their own words, application scenarios that only Airbender can bring:

First, naturally, are applications such as on-chain order books, payment systems, exchanges, and automated market makers. Airbender enables the system to verify and settle at extremely fast speeds, reducing the risk of rollbacks for these applications on-chain.

The second point, which many current L2s cannot achieve, is that public systems and private systems (such as ZKsync's Prividiums) can interoperate without third parties. Prividiums is the infrastructure launched by ZKsync to help enterprises establish private chains. For enterprises, the requirements for blockchain are fast settlement and privacy. Fast settlement goes without saying, and the inherent privacy of ZKP allows enterprise private chains to verify transaction validity without exposing the ledger information of the chain when interoperating with public chains. The combination of the two even meets the compliance requirements for settlement times in on-chain securities and foreign exchange trading.

This may also be the reason why ZKsync has become the second-largest tokenized RWA asset issuance network after Ethereum.

ZKsync proudly states that all of this can only be achieved under the Atlas upgrade: the sorter provides low-latency transaction packaging, Airbender generates proofs within a second, and then the Gateway verifies and coordinates cross-chain messages.

Bridging L1 and L2

As Vitalik retweeted in a tweet, ZKsync founder Alex believes that after the Atlas upgrade, ZKsync has truly bridged the Ethereum mainnet.

Now, ZKsync's final confirmation time for transactions (about 1 second) is shorter than the Ethereum mainnet's block time (average 12 seconds), which means that institutional and RWA transactions conducted on ZKsync are essentially consistent with those on the Ethereum mainnet, waiting only for confirmation from the Ethereum mainnet. This means that ZKsync does not need to repeatedly establish liquidity centers on L2 and can directly use the liquidity of the mainnet, as the cross-chain interaction between ZK Rollup and the mainnet does not require a 7-day challenge period like OP Rollup, and the Atlas upgrade further accelerates the speed on this basis.

This has improved the recent discussions in the Ethereum community regarding L2 fragmentation; L2 and L1 are no longer two disconnected chains but are connected through rapid confirmation and verification, making L2 truly capable of being called an "expansion network" for the first time.

I remember when ZKsync and Scroll first launched their mainnets, the transaction confirmation speed and gas fees were no different from or even higher than the mainnet, essentially because there had not yet been systematic optimizations for the algorithms and engineering of ZKP, resulting in slow verification speeds and high costs, which at the time triggered a crisis of trust in ZK Rollup. Today, Optimism and Arbitrum are slowly transitioning from OP Rollup to ZK Rollup (or a combination of both), and the further improvements in cost and speed of ZK Rollups like ZKsync, along with Scroll's decentralized ZKP, have transformed from "nonsense" into results worth looking forward to.

From being universally criticized to becoming highly sought after, ZK has welcomed a dawn. After the multi-signature implementation of sorters and cross-chain bridges is fully decentralized, it may truly achieve what Dragonfly managing partner Hasseb Qureshi said: "can't be evil."