Ethereum’s proving technology has reached a point where zkEVM security deserves more attention than another speed record, according to co-founder Vitalik Buterin.
His comments followed ZKsync’s introduction of Airbender Prime, a prover that the team says can generate Ethereum execution proofs in real time using a single GPU. Buterin congratulated the developers while arguing that the next priority should be establishing stronger guarantees about what those proofs actually demonstrate.
The distinction is consequential: producing a proof quickly and proving the correct rules are being enforced are separate engineering challenges.
The benchmark behind the reaction
ZKsync’s October 13 announcement describes an evaluation covering 7,200 consecutive Ethereum mainnet blocks.
Benchmark component | ZKsync’s reported result |
Hardware | One NVIDIA RTX PRO 6000 Blackwell GPU |
Sample | 7,200 consecutive Ethereum mainnet blocks |
Proving time for 99% of blocks | Under 10 seconds per block |
Proving time for every sampled block | Under 12 seconds |
Included work | Witness preparation |
These are results reported by ZKsync from its own evaluation.
Witness preparation supplies the execution information needed to construct a proof. Including it makes the reported timing more informative than a figure covering only part of the proving process.
ZKsync also describes Airbender Prime as based on the open RISC-V standard, with a standalone verifier intended to support independent verification.
The benchmark establishes performance for the tested sample and configuration. It does not establish that every possible future Ethereum block will meet the same timing.
Buterin’s message: three security guarantees
In his X post responding to the milestone, Buterin described the speed battle as “definitively won” and said:
“the top priority for zkEVMs should shift fully to security.”
He identified three areas for formal verification: consistency with Ethereum’s specification, an upper bound on prover cost relative to block gas, and the security of the proof system itself.
Put plainly, those questions are:
Does the system enforce Ethereum’s actual rules? A proof is only useful if the computation being proved corresponds to the intended protocol.
Can an unusually difficult block make proving prohibitively expensive? Typical performance and worst-case behaviour need separate examination.
Can the proof system reject false claims reliably? Faster generation does not replace the need for sound cryptography.
The broader architectural context is explored in Bitnxt’s coverage of Buterin’s cryptographic vision for Ethereum.
Why formal verification matters
Ethereum Foundation technical material explains the central risk: a circuit can implement the wrong rules. In that situation, demonstrating that the circuit ran correctly would not establish that the resulting state transition complies with Ethereum.
The same overview identifies emulator correctness, circuit correctness and the behaviour of the complete proving pipeline as security concerns.
Formal verification aims to establish mathematical properties against specified models. It complements testing, audits and other scrutiny; its usefulness depends on the property and assumptions being verified.
The Foundation’s overview also discusses diversity across execution implementations and proving systems. Requiring different implementations to support a block’s validity can reduce dependence on one implementation’s correctness.
This concern about hidden implementation errors has parallels elsewhere in blockchain infrastructure. Bitnxt’s report on the XRP Ledger supply-cap bug and its emergency patch examines a separate case in which safeguards and code behaviour required close scrutiny.
What this changes and what still needs to happen
Bitnxt analysis: Airbender Prime strengthens the performance case for proof-based verification. Lower hardware requirements could broaden the range of operators able to participate in proving, provided the economics and operational requirements remain practical.
But the announcement does not itself change Ethereum’s consensus rules, activate a protocol upgrade or establish a timetable for adopting this particular prover.
It also should not be presented as a guarantee of transaction privacy. Ethereum Foundation material cautions that systems commonly grouped under “ZK” can provide succinct validity proofs without providing zero-knowledge privacy.
For readers assessing the technology, the next useful evidence is reproducibility: independent benchmark runs, clearly documented test conditions, security reviews and the scope of formal-verification work.
You might also like: EU regulators’ warning about quantum-computing risks to blockchains—related coverage of another long-term cryptographic challenge, distinct from proving speed.
Bitnxt view: the next milestone needs a different scorecard
The headline performance figures are impressive on their stated terms. Buterin’s response raises the standard for what should count as progress from here.
Bitnxt analysis: The strongest next announcement would explain which correctness properties have been verified, which assumptions remain and how the system behaves under adversarial workloads. Another reduction in average proving time would answer only part of that question.
For Ethereum, the goal is dependable verification at scale. Speed makes that goal more practical; security determines whether the infrastructure can be trusted.
