Ethereum still has a long engineering and governance road ahead before it can be “quantum-safe,” but the search for ways to handle the data burden of post-quantum cryptography is accelerating. Ethereum co-founder Vitalik Buterin publicly changed his view on leanPCS on November 16, saying that in just a few weeks he had gone from “completely skeptical” to “completely optimistic.” He believes that, around EIP-8288, leanPCS could offer Ethereum a way to process post-quantum signatures, STARK proofs, and other verifiable objects faster.
This is not a network upgrade that has already been completed, nor does it mean Ethereum is already protected against quantum-computing attacks. The key question is whether Ethereum can process large amounts of cryptographic data off-chain and then compress the verifiable results into smaller on-chain commitments, given that post-quantum signatures are larger than those used today and their proofs are more complex to verify. The debate over leanPCS centers on how to balance data compression, generality, engineering complexity, and governance costs.
Quantum security is first and foremost a data-size problem
Existing blockchain account systems rely on public-key cryptography such as elliptic-curve cryptography. If quantum computers become powerful enough in the future, related algorithms could threaten this type of cryptography, so Ethereum needs to gradually support quantum-resistant signatures. But quantum-resistant schemes often bring larger signatures and greater proving and verification burdens. If every transaction writes its complete data directly to a block, nodes must propagate, store, and verify these objects one by one, putting pressure on block space and execution costs.
When explaining EIP-8288, Vitalik summed up the problem as performing verifiable computation on data. According to his explanation on X, users can first generate a STARK, a SPHINCS signature, or a PCS opening and publish the object to the mempool. Nodes then collect new objects, include statements such as “public key P signed message M,” “program X ran on data D and produced result Y,” or “polynomial P evaluates to Y at a certain point” in a new STARK, and prove that they actually saw and verified the corresponding objects.
Under this process, the network does not need to repeatedly broadcast every complete signature or proof. Instead, it mainly retains the hashes of inputs and outputs, along with one aggregated STARK for each time interval. Vitalik’s public explanation notes that the hash information for each statement is about 64 to 96 bytes, and that aggregated proofs can be further aggregated recursively. This figure describes the scale of the information in the scheme; it is not a performance result from deployment on Ethereum mainnet.
Why leanPCS changed Buterin’s mind
EIP-8288 is not about a single signature algorithm, but about providing a transaction framework for quantum-resistant signatures and STARK aggregation. The discussion includes different approaches such as leanSPHINCS and leanSTARK. leanSPHINCS targets the more clearly defined use case of hash-based signatures. Its approach is relatively direct, but its scope is also more limited. leanSTARK is more general-purpose and can support different signature schemes, privacy protocols, or hidden multisig logic.
The challenge is that the more general-purpose a system is, the more design questions it must resolve upfront. As Decrypt relayed Buterin’s view, for leanSTARK to become infrastructure that Ethereum users can widely use, the community may first need to reach consensus on the programming language used by user programs. This means it is not just a cryptographic implementation problem; it also touches on virtual machine boundaries, developer tools, clients, and governance processes.
leanPCS offers a third trade-off. Buterin believes it could retain capabilities close to those of leanSTARK while potentially being implemented faster in theory, since it requires fewer design choices. The approach he described is to start with a smaller, more focused mathematical checking tool around which different proof systems can work, then add a more complete leanSTARK to the system in the future. leanPCS is therefore appealing not only for its cryptographic performance, but also because it may avoid the governance challenge of designing a complete “new execution language” all at once.
However, “faster” is still an assessment of the proposal, not a confirmed delivery timeline. EIP-8288 remains in the draft and discussion stage, and leanPCS has no mainnet activation date. Buterin’s public support can influence research and development priorities, but it cannot replace EIP review, client implementation, testnet validation, or hard-fork governance.
From mempool aggregation to on-chain verification: who bears the cost?
The core change introduced by leanPCS and EIP-8288 is a redistribution of verification costs, not their elimination. Users or their clients need to generate signatures, STARKs, or PCS openings; nodes in the mempool need to listen for, organize, and generate aggregated proofs; and block builders need to replace the corresponding dependencies with recursive proofs that can be verified on-chain. This may reduce on-chain data and verification burdens, but it increases off-chain computation, network coordination, and the complexity of proof generation.
This arrangement also raises new engineering challenges. First, the mempool is not a globally consistent database: different nodes may see transactions in different orders and at different times. The aggregation process needs to define object lifetimes, dependencies, duplicate submissions, ordering, and failure-retry mechanisms. Second, if proof generation is too expensive or takes too long, ordinary users may depend on a small number of intermediaries with the necessary hardware and services, which could affect the network’s openness.
The transition to quantum resistance cannot be judged by signature size alone. How accounts migrate from existing keys to new ones, how wallets display and safeguard more complex key material, how smart contracts verify different types of signatures, and how exchanges and custodians upgrade are all questions that must be answered before mainnet adoption. Aggregated proofs address how to efficiently carry and verify large numbers of cryptographic objects; they do not automatically determine who changes which keys, or when.
What this means for Ethereum’s roadmap
If the leanPCS approach is eventually adopted, it could change the order in which Ethereum’s quantum-resistant roadmap advances: the network would not need to wait for a complete proof language covering every use case to mature, but could first establish an aggregation layer capable of carrying multiple types of verifiable objects. For ordinary users, the ideal outcome is not having to learn new cryptographic terminology, but being able to use quantum-resistant accounts after an upgrade without bearing all the on-chain costs associated with full signature data.
The benefits could be more apparent for developers. A unified aggregation interface could allow account signatures, zero-knowledge proofs, L2 proofs, and other verifiable computations to share some infrastructure. Quantum-resistant security would therefore no longer be a one-off task for wallet teams; it would connect with Ethereum’s execution layer, transaction propagation, block building, and proof markets.
But the ecosystem also needs to avoid mistaking “general-purpose aggregation” for “immediate compatibility with every application.” Different proof systems have different security assumptions, trusted setup requirements, verification times, and levels of implementation maturity. If the protocol enshrines one approach as a core standard too early, it may face migration costs in the future; if the standard is too broad, it may make client implementation and auditing more difficult. The value of leanPCS will ultimately need to be validated through reference implementations, public testing, performance measurements, and collaboration among multiple clients.
What to watch next
The most important thing to watch now is not whether the market immediately prices in “quantum security,” but whether EIP-8288 can move from conceptual discussion to reproducible engineering validation. Specifically: whether leanPCS’s definition and security model are clear; the costs of generating and verifying aggregated proofs on different hardware; whether mempool aggregation could create a new point of centralization; whether clients and block builders can support dependency replacement without sacrificing usability; and whether account migration plans can be integrated with the aggregation framework.
Vitalik’s shift from skepticism to support suggests growing confidence in the idea, within Ethereum’s research and development discussions, of “addressing urgent scalability problems first, then gradually expanding generality.” But this is still an early-stage protocol design, not an upgrade commitment that has already been fulfilled. The real test of quantum security for Ethereum is not coming up with an acronym that sounds efficient, but finding a solution that can operate over the long term while balancing security, cost, decentralization, and governance feasibility.
