Bitcoin’s Progress Towards Post-Quantum Security

By Coinbase • 5min read

Quantum computers are not an immediate threat to Bitcoin, but upgrading a decentralized ecosystem takes time. Good news is that Bitcoin’s post-quantum conversation is advancing quickly. Researchers are now comparing specific output formats, implementing test vectors, measuring transaction costs, and proposing signature schemes designed around Bitcoin’s constraints. We discuss some of these recent developments in this post.


Moving Bitcoin’s Post-Quantum Readiness Forward

Two complementary approaches to PQ output types

An output type defines what a Bitcoin UTXO commits to and what a spender must reveal to unlock it. A recent Delving Bitcoin discussion compares several candidates, including P2TRv2 and P2MR.

P2TRv2 is currently a design concept rather than a numbered BIP. It retains Taproot’s familiar structure: an efficient elliptic-curve key path plus a tree of alternative script paths, extended with post-quantum verification. Ordinary spending could continue using a 64-byte Schnorr signature. If elliptic-curve cryptography later became unsafe, a consensus mechanism would disable the vulnerable paths and leave the post-quantum path available.

Its strength is deployability. P2TRv2 could preserve today’s fee profile, wallet architecture, descriptors, and key-management workflows. That could help move more bitcoin into outputs with a post-quantum recovery path.

Its tradeoff is that every output exposes an elliptic-curve point. Once a cryptographically relevant quantum computer exists, safety would depend on disabling elliptic-curve spending in time. Proposed mechanisms include an on-chain “tripwire” demonstrating an elliptic-curve break and miner signaling that triggers a predefined lockdown. P2TRv2 thus does not offer a pure PQ option for holders who want the security of their funds to not depend on future network actions.

P2MR, or Pay-to-Merkle-Root, takes a different approach. BIP 360 is a draft proposal for a SegWit version 2 output that commits directly to a script-tree root. Unlike Taproot, it has no internal public key and no privileged key-path spend. A spender reveals a selected script, its inputs, and the Merkle path.

This keeps elliptic-curve public keys off-chain until their corresponding path is used, assuming they have not been exposed through key reuse or descriptor sharing. With a future post-quantum verification opcode, a P2MR tree could contain both an efficient elliptic-curve path and a post-quantum path—or omit elliptic-curve spending entirely for stronger protection.

The cost is a less compact “happy path.” The comparison on Delving Bitcoin estimates approximately 128 witness bytes for a simple P2MR elliptic-curve spend, versus a 64-byte P2TRv2 key-path signature. However, P2MR is 32 bytes lighter than an equivalent Taproot script-path spend because it omits Taproot’s internal key. An independent regtest implementation confirmed the savings and demonstrated that the consensus delta is under a hundred lines of code by reusing existing Taproot machinery.

The emerging insight is that these designs may be complementary. P2TRv2 prioritizes broad, low-friction migration; P2MR offers greater control and a path to post-quantum-only security. Other discussions e.g. public-key recovery and new witness accounting are exploring how to narrow the efficiency gap.

SHRINCS makes the signature tradeoffs explicit

An output type still needs a post-quantum signature scheme. The new SHRINCS draft BIP proposes a hash-based construction designed specifically for Bitcoin.

SHRINCS uses SHA-256 and targets NIST security category 1. One 48-byte public key supports two signing paths:

  • A compact, stateful path based on FXMSS and WOTS+C, producing signatures from 548 to 4,619 bytes.

  • A stateless fallback based on the standardized SLH-DSA algorithms, with a Bitcoin-specific parameter set producing 5,777-byte signatures.

For comparison, a Schnorr signature is 64 bytes. Post-quantum signatures are simply less space-efficient today, which is why difficult tradeoffs around block space, verification, and operational complexity cannot be avoided.

The stateful path offers a promising compromise. Each one-time signing leaf may be used only once, so the signer maintains a counter to select the next leaf. An unbalanced tree can make early signatures especially small, which fits wallets that rarely reuse a key. Balanced trees provide fixed-size signatures and larger signing budgets for other applications.

The efficiency gains come with real operational requirements though since reusing a counter can enable forgery. Implementations therefore need durable, rollback-resistant storage; must prevent concurrent signing; and must persistently advance the counter before returning a signature. Restoring an old copy of the counter is unsafe.

SHRINCS’s stateless path helps with recovery. If the state is lost, restored from a static seed backup, or otherwise uncertain, the wallet must stop using the stateful path but can still spend through the larger stateless path. 

The proposal is in its early stages. A full security proof and comprehensive test vectors are listed as unfinished work. Separately, a proof-of-concept C implementation of WOTS+C (one of SHRINCS’s components) called libshrincs has been published, together with machine-checked proofs of functional correctness and security. That does not yet prove the complete SHRINCS scheme, but it is an encouraging foundation for rigorous review.

Bringing the ecosystem together

Coinbase continues to work with research organizations such as Localhost Research and Brink to organize technical events and working sessions that bring the ecosystem together. Earlier this year, Coinbase, ProfDan Boneh from Stanford, and Localhost Research hosted a Post-Quantum Bitcoin Workshop for developers, cryptographers, researchers, custodians, and hardware-wallet experts. 

This work builds on the recommendations of the Coinbase Quantum Advisory Council and our broader post-quantum readiness program.

Recent stories