Earlier this month, Arthur Breitman laid out why post-quantum, and why now. Now, there is something concrete to experiment with alongside that argument: Quantumnet, a running experimental Tezos network whose consensus, manager keys and Data Availability Layer are all post-quantum.
Stood up last week, Quantumnet is live in its first iteration and now open to community testing. It is not a preview of a finished post-quantum Tezos, nor a pre-production network. It is a place to experiment rigorously and in the open with what making Tezos post-quantum actually entails - across cryptography, algorithms, architecture, operations and, eventually, the transition itself.
Network dashboard: https://quantumnet.pqpark.dal.nomadic-labs.com/
Block explorer: https://quantumnet-tzkt.pqpark.dal.nomadic-labs.com/
Join it as a baker: https://quantumnet-tutorial.pqpark.dal.nomadic-labs.com
What is actually post-quantum here
Making Tezos post-quantum involves more than replacing user keys. Cryptography is used at several layers of the protocol, and replacing it has consequences for the algorithms and architecture built around it.
Quantumnet brings several of those changes together in a running network:
-
Manager keys are tz5 - ML-DSA-44, the lattice-based signature scheme standardized by NIST as FIPS 204. These keys sign operations such as transfers, delegation and staking.
-
Consensus keys are tz6 - XMSS, a stateful hash-based signature scheme. Bakers therefore run a tz5 manager key alongside a tz6 consensus key used for per-block consensus operations.
-
Consensus operation aggregation uses a post-quantum STARK-based approach, allowing the network to retain aggregation while moving away from the BLS signatures used on Mainnet today.
-
The DAL runs ZODA, a data-availability scheme based on hashes and erasure coding rather than the pairing-based KZG commitments used by the current DAL.
-
Baking rights no longer depend on protocol randomness. The VDF and lottery-based allocation have been replaced by Smooth Weighted Round Robin.
The tz6/tz5 split illustrates why this work is not simply a matter of swapping one cryptographic primitive for another.
Tezos aggregates consensus operations, and losing aggregation when moving to post-quantum cryptography would come with significant costs. ML-DSA does not provide the aggregation properties needed here, so Quantumnet uses XMSS signatures together with a post-quantum aggregation mechanism for consensus.
That introduces a new operational constraint: an XMSS key can only produce a finite number of signatures. It is a mutable state that a baker has to monitor and eventually rotate.
For this first iteration, the tutorial sizes the key for 131,071 signatures - roughly three days of baking - and provides the rotation commands. Rotation is still manual, and Octez does not currently warn the baker as the signature budget is depleted.
That is exactly the kind of constraint Quantumnet exists to expose and explore.
What works today
This first iteration is already a complete enough network to exercise the different components together.
Today you can:
-
run your own node and become a baker;
-
delegate and stake;
-
send transactions using post-quantum manager keys;
-
produce and attest DAL slots using the post-quantum DAL;
-
explore the chain through a block explorer;
-
use a faucet and public RPC endpoint;
-
inspect L1 and DAL metrics through the public network dashboard.
Blocks are 6 seconds and a cycle is 10 minutes, so you can start exercising the network quickly. Budget roughly 45 minutes to go through the complete baker tutorial, including the protocol-mandated wait for baking rights.
What is not there yet
Being explicit about this is part of the purpose of a first iteration.
-
No rollup is deployed yet. The plumbing for a smart rollup consuming data from the ZODA DAL exists and has been exercised, but no rollup is currently running on the public network.
-
DAL bandwidth is deliberately modest. Slot geometry and publication cadence are conservative while we measure the system. Later iterations will push this further.
-
The network topology is still highly controlled. This iteration is primarily about integrating and exercising the post-quantum components together, not reproducing a realistic Mainnet P2P topology.
-
Key-rotation UX is still an engineering workflow rather than a finished product. XMSS state management and rotation are among the operational questions this environment lets us investigate.
-
Hardware requirements are still being measured. This first iteration prioritizes experimentation over matching the requirements of the lightest Mainnet setups.
Most importantly, Quantumnet is not a preview of a post-quantum Tezos design that is already finished.
This is iteration 1. There will be further iterations, each giving us an opportunity to change assumptions, replace components, improve performance and explore the operational consequences of the design.
New iterations may start from a fresh chain, so participants should not expect balances, delegate registrations or history to persist between them.
Use throwaway keys. Nothing on this network has value.
What comes next
The purpose of Quantumnet is to give us time to learn before decisions need to become commitments.
Among the areas we expect future iterations to explore are:
-
higher DAL bandwidth;
-
more realistic network topology;
-
improved XMSS key management and operational UX;
-
stateful addresses and the user transition path;
-
a rollup supporting post-quantum encrypted private transactions;
-
continued performance, security and hardware-requirement evaluation.
We also intend to publish a series of engineering insights explaining what we learn along the way: the protocol changes, consensus aggregation, the XMSS state problem, ZODA in practice, performance trade-offs, and the broader transition challenges.
Back in 2014, the position paper recognised that Tezos would require new cryptographic primitives as the need developed. Every amendment so far has changed what Tezos does or how well it does it. Twenty-one upgrades later, this one reaches into the cryptographic foundations themselves.
The precise shape of a future post-quantum Tezos is not something Quantumnet assumes in advance. The point of running these experiments now is to have the time to discover it.
Come and experiment with it
Follow https://quantumnet-tutorial.pqpark.dal.nomadic-labs.com, join Quantumnet with your own node, bake a few cycles and tell us what you find.
Failure reports, performance measurements, operational pain points, and “this step of the tutorial makes no sense” are all useful results.
Quantumnet is an engineering playground in the open. This is the first iteration, not the destination. The faster we learn from it, the better informed the next iteration can be.
