Making Ethereum ready for PQAI Day
The third post in our series on the road to a fully post-quantum Ethereum.
In Mission 2028, we argued that Ethereum needs to accelerate its transition to stronger cryptography. In our second post, we examined what could break across accounts, applications, consensus and data availability. This post turns to preparation: what can begin now, what requires a protocol upgrade, and what users and operators need to do once those changes arrive.
Borrowing the PQAI terminology from Ethlabs’ pqai27 proposal, we use PQAI Day as shorthand for a quantum or AI-assisted mathematical breakthrough that undermines the cryptography Ethereum relies on. This describes a possible threat, not a demonstrated break. Preparing for it means establishing a migration path while today’s authorization mechanisms can still be trusted. [1]
Three pieces explain much of the proposed path:
- Frame transactions let accounts define how transactions are authorized, making room for post-quantum signatures.
- SETCODEFROM and EIP-8151 let migrated accounts retire their old ECDSA authority, including authorization through contracts that use
ecrecover. - A validator pre-registry records replacement key material before the mechanisms authorizing that registration become vulnerable.
Their timing differs. Validator registration could begin before consensus switches to new signatures. The native account migration path discussed here requires protocol changes before users can complete it. These proposals remain under development, and their scope and scheduling are subject to Ethereum’s public upgrade process.
What does it mean to be ready?
Readiness has two levels.
Protocol readiness means Ethereum has the mechanisms needed to move away from vulnerable cryptography: replacement signatures, migration rules and a data-availability fallback. Those mechanisms need to be implemented and tested before the network can rely on them.
Participant readiness means the people and systems using Ethereum are prepared to make that transition. Validators have registered replacement material and tested their signing setups. Users have migrated their accounts, and applications have addressed the permissions and cryptographic dependencies that could otherwise leave them exposed.
Our goal is to reach a point where retiring vulnerable schemes requires as little remaining coordination as possible. A fork would still need an agreed activation time, but its rules, software and operational procedures should already be ready. The network may initially operate at reduced capacity.
For consensus, participation has a concrete threshold. Under Ethereum’s current finality rules, votes must represent at least two-thirds of the active effective stake. A transition should aim comfortably above that minimum, with evidence that operators can perform their duties under the new scheme. Counting registrations alone would not establish this. [2]
For users, there is no equivalent threshold that makes everyone safe. Ethereum can keep running while an unmigrated account remains exposed. Readiness therefore also requires understanding which accounts and applications still depend on vulnerable authorization and what options remain available to them.
The path to readiness
Preparation, migration and retirement of the old schemes happen on different schedules. The milestones below show what enables each stage and which work can proceed in parallel.
The path to PQAI readiness
opens
activate
switch
Consensus: register first, switch later
Preparation. The consensus paths under discussion are built around hash-based signatures such as leanXMSS. Ethereum also needs a practical way to handle signatures from many validators. Aggregation lets the network verify a proof covering many signatures, and leanVM is being developed to support that work. We will explore it in more detail in a future post. [3]
Alongside this, validators need to establish which material should replace their current keys. That is the purpose of a pre-registry.
The post-quantum material registration proposal, EIP-8483, allows a contract to accept requests before the fork that processes them into consensus state. Deploying the contract and accepting requests can happen earlier; making consensus recognize them still requires a protocol upgrade. [4]
This creates an opportunity to begin preparation before the new signing system is ready. The ecosystem first needs agreement on what operators should register and how they should generate and preserve it. Early preparation needs to remain usable at the eventual transition.
Migration. Registration requires an organized rollout. Solo validators, staking providers, custodians and staking protocols have different arrangements. The party operating a validator may not control its withdrawal address, so instructions and tooling must account for that coordination.
Progress should be public and measurable: the share of active stake with registered material, the share whose operators have tested the new signing setup, and the remaining operational blockers.
Registering replacement consensus material does not migrate the account controlling a validator’s withdrawals. That account needs its own preparation on the execution side.
Switch. At the consensus transition, validators would begin signing under the new rules and the old signatures would cease to authorize their duties. The transition also needs a clear cutoff for registration or replacement requests authorized through vulnerable mechanisms.
As we explained in our second post, pre-registration allows the network to establish replacement authority before an attacker can impersonate the party entitled to choose it.
Execution: make migration possible, then make it usable
Preparation. Execution needs a way to accept new signatures and a way for accounts to stop accepting their old keys.
Frame transactions, EIP-8141, address the first part by letting account code define transaction validation and payment. This gives wallets a native transaction format that can support post-quantum authorization. Frames provide the structure; a scheme such as leanSPHINCS supplies the signature. There is much more to say about frames, and we will cover them in a separate post. [5]
At LambdaClass, where we have been implementing frames in ethrex, our draft proposal for leanSPHINCS dependency frames would add native verification of these signatures on top of frames. Its immediate goal is to make PQ accounts usable without waiting for the full signature aggregation system. We return to that longer-term connection below. [6]
Existing accounts also need to remove their old authority. SETCODEFROM, EIP-8298, would let an account adopt regular wallet code at its existing address. Once migrated, the old ECDSA key could no longer originate ordinary transactions or replace the wallet through EIP-7702 delegation. The installed wallet code would determine control. [7]
EIP-8151 addresses a related gap: applications could otherwise continue accepting the old key through ecrecover, Ethereum’s built-in ECDSA recovery function. The proposal makes recovery fail for accounts with regular deployed code. Applications that implement their own ECDSA verification need separate attention. [8]
There is also compatibility work before anyone migrates. EIP-8151 makes ecrecover read account state and charge additional gas. A contract that forwards exactly 3,000 gas could fail even when checking an ordinary, unmigrated account. Raising the transaction’s overall gas limit would not fix a hardcoded internal limit. [8]
Application teams need time to inspect their use of the precompile, replay transactions under the proposed rules and report affected contracts. The recent EIP-8037/8038 testing shows why: changes to gas accounting exposed assumptions in deployed applications and delayed the Hoodi upgrade while their impact was investigated. This work belongs in preparation, before activation is scheduled. [9]
Migration. Once the native tools are available, wallets need to turn them into a process users can complete and understand.
That process will differ across ordinary externally owned accounts, accounts using EIP-7702 delegation, and smart accounts using ERC-4337. Some smart accounts can already support custom signature verification. The proposed upgrades provide a native route and, for existing EOAs, a way to retire the original authority while retaining the address.
Contract accounts need review too. They have no private key of their own, but their owners, administrators and recovery mechanisms may depend on vulnerable signatures. Being a contract does not make those dependencies safe. [10]
Migration guides should explain how to establish new authorization, retire old permissions and keep using applications afterward. Wallet providers and application teams will have to coordinate that experience.
Switch. Execution does not necessarily have the same single transition point as consensus. Once the required changes are active, individual accounts can migrate and retire their ECDSA authority independently.
A network-wide cutoff for remaining legacy authorization is a separate decision. It would require a plan for accounts and applications that have not migrated. Vitalik Buterin’s 2024 recovery-fork proposal, which we discussed in our second post, sketches one such plan for a quantum emergency: disabling traditional account transactions and letting users recover control by proving knowledge of secrets that a key-recovery attack would not reveal. The fork enabling migration does not, by itself, settle that question. [11]
Data availability: prepare a fallback
Blobs need their own transition because their current KZG commitments depend on cryptography vulnerable to quantum attacks. Work on post-quantum data availability, including leanDA, aims to replace that dependency. [12]
An earlier readiness plan could include disabling the affected blob path and having rollups publish data through calldata in post-quantum-authorized transactions. That fallback is also discussed in the Ethlabs proposal. [1]
It would come with higher costs and lower available throughput. Rollup teams would need to test their publishing systems, estimate the capacity they could sustain and coordinate the transition with Ethereum’s upgrade. Their proof systems, bridges and administrative keys would still need their own migration.
The ecosystem can work directly with rollup teams on this fallback. That preparation should happen before anyone has to rely on it.
What can begin now?
The protocol changes are only part of the work. Each group has preparation it can begin while the designs and implementations develop:
| Who | Preparation that can begin now | What depends on the migration tools |
|---|---|---|
| Users | Identify the wallets, applications and custody services they rely on, and follow their migration plans. | Complete a supported migration and verify that the old authority has been retired. |
| Wallet teams | Develop and test PQ signing, recovery and migration flows against evolving proposals. | Offer the native migration path once the required protocol support is active. |
| Application teams | Review signature checks, administrative permissions, recovery mechanisms and gas assumptions. | Integrate new authorization methods and migrate affected permissions or contracts. |
| Validator operators and staking providers | Coordinate who generates, holds and registers replacement material; test signing setups. | Register when an agreed registry is available and activate the new signing rules at the transition. |
| L2 teams | Review proofs, bridge permissions and publishing infrastructure; test a calldata fallback. | Coordinate changes with Ethereum’s upgrades and migrate each remaining dependency. |
Let each migration start as soon as possible
The different schedules are one reason to consider separate execution and consensus upgrades. If execution’s migration tools are ready first, shipping them earlier would give wallets and users more time to move while consensus work continues. Ethlabs has suggested splitting the upgrades this way; it remains a roadmap discussion. [1]
The work between activation and adoption will be substantial. Ethereum needs sustained outreach, guides for different account and staking setups, test environments, support channels and public measurements of progress.
Those measurements should distinguish installing support from successfully using it. For validators, that means tracking both registered stake and demonstrated signing capability. For execution, it means observing migrated accounts, compatible applications and remaining dependencies on old authorization, while acknowledging what cannot be measured reliably.
There also need to be agreed responses to slow adoption, incompatible contracts, operators unable to migrate, or a cryptographic break arriving earlier than expected. These are part of making the transition operationally possible.
Where we want to get to
Readiness establishes the minimum needed to continue operating safely through a transition. Ethereum should also make post-quantum operation affordable and scalable.
EIP-8288 addresses that next step through signature and proof aggregation. Hash-based signatures are much larger than today’s signatures. Aggregation would let a block carry a shared proof that its required signatures and other cryptographic claims are valid, reducing the data each node must retain and verify individually. [13]
The broader design is still evolving. Vitalik Buterin recently described using leanPCS as an earlier route to general proof support, before standardizing the more extensive leanSTARK interface. The published EIP still describes leanSTARK; leanPCS is a direction under discussion. [14]
Our leanSPHINCS proposal is designed with this destination in mind. Initially, nodes would verify individual signatures. Later, an aggregate proof could replace those signatures without requiring users to change their keys or accounts for that reason. The aim is to let migration begin while preserving a path to EIP-8288. [6]
Other roadmap decisions remain open. FOCIL, which strengthens censorship resistance, needs a schedule compatible with the PQ work. For decoupled consensus, the question includes whether Ethereum should reach PQ consensus by adapting today’s design or moving directly to the newer one. What can proceed together, and what would delay readiness, still requires assessment. [1]
Our position remains the one we set out in the previous post: preparation that can safely begin without a hard fork should begin as soon as its design is agreed. The protocol changes needed to let users migrate should be prioritized for the next fork, with a scope focused on getting those tools into use.
Deploying the tools begins the migration. Wallet support, application updates, validator coordination and public tracking need to follow. By the time Ethereum retires its vulnerable cryptography, the ecosystem should already have tested how it will continue operating.
What comes next: the last mile
The next post will follow the path from a user’s post-quantum signature to the proof included in an Ethereum block. We will look more closely at leanSPHINCS, leanVM, leanPCS and leanMempool, and explain how our proposal provides a starting point for that broader design.
We will also examine how FOCIL and decoupled consensus fit into the transition, including decisions that may affect the earliest path to readiness. The focus will be the last mile: what remains to move from having a workable migration route to a fully post-quantum Ethereum that can operate at scale.
Sources
- An Ethlabs view on post-quantum and post-AI timelines: pqai27 — Caspar Schwarz-Schilling, November 11, 2026.
- Proof-of-stake: finality — ethereum.org.
- leanVM — development of a hash-based zkVM, including leanXMSS and leanSPHINCS aggregation.
- EIP-8483: Post-Quantum Material Registration — proposal and discussion in PR #12609.
- EIP-8141: Frame Transaction.
- Dependency frames for leanSPHINCS signatures — LambdaClass’s draft proposal, PR #12672.
- EIP-8298: SETCODEFROM Code Reuse Instruction.
- EIP-8151: Account Code Restricted ecRecover.
- EIP-8037/EIP-8038 Glamsterdam — Ethereum Foundation, November 6, 2026, with subsequent updates.
- Ethereum accounts — ethereum.org.
- How to hard-fork to save most users’ funds in a quantum emergency — Vitalik Buterin, March 2024.
- EIP-8479: Post-Quantum PeerDAS and Variable-Size Blobs — proposal and discussion.
- EIP-8288: In-mempool signature and proof aggregation.
- leanSPHINCS, leanSTARK and leanPCS in EIP-8288 — Vitalik Buterin, November 16, 2026.