BNB Smart Chain has released a post-quantum cryptography migration report showing that it completed a test deployment of quantum-resistant changes across both transaction signing and consensus. In the trial, BSC replaced ECDSA secp256k1 with ML-DSA-44 (Dilithium2) for transaction signatures and switched consensus vote aggregation to pqSTARK. The headline result was mixed: wallet addresses stayed unchanged and current RPCs, SDKs, and wallets remained compatible, but average transaction size expanded from about 110 B to roughly 2.5 KB, close to a 37x increase, while TPS in cross-region native transfer tests dropped by 40%.
ML-DSA-44 replaces ECDSA in a forward-looking test
BSC described the exercise as preparation rather than a response to an immediate break in production cryptography. Current blockchain transaction signing still depends heavily on elliptic curve cryptography built on discrete logarithm assumptions, and the report said that model becomes vulnerable in a quantum setting. For that reason, BSC used ML-DSA-44 as the new signature scheme in its test environment.
According to the report, ML-DSA-44 is the only post-quantum digital signature algorithm standardized in FIPS 204 by the U.S. National Institute of Standards and Technology in August 2024. BSC said it did not choose a higher security variant because ML-DSA-44 already provides enough security margin for the next 10 to 20 years, while offering the fastest verification speed among the options considered and reducing the performance hit on-chain.
Address format stays at 20 bytes and existing wallets still work
The report said users would not need to replace their addresses or migrate assets. BSC derives addresses from the new ML-DSA-44 public keys through the keccak-256 hash function, preserving the standard 20-byte address format. That keeps the user-facing address structure intact and allows the current wallet stack, along with RPC and SDK infrastructure, to remain usable without changes.
At the consensus layer, the compression result was much stronger. Using pqSTARK, BSC reduced 14.5 KB of raw signatures from six validators to about 340 B after aggregation, a 43:1 compression ratio. Based on the figures in the report, this sharply lowered the on-chain data footprint of validator signatures and prevented the consensus layer from becoming the main source of overhead.
Data growth, not compute limits, becomes the main bottleneck
The larger problem appeared in transaction data. BSC said that after moving to ML-DSA-44, public key size grew by 20x, while signature size per transaction surged from 65 B to 2,420 B. That pushed a typical transaction from around 110 B to about 2.5 KB, and block size rose to roughly 2 MB, an 18x increase.
In cross-region network tests for native transfers, TPS fell 40%. The report said the network was not hitting a compute ceiling first. Instead, the byte budget of each block was exhausted earlier than the Gas budget because each transaction had become much heavier. BSC concluded that post-quantum signatures can be integrated into both transactions and consensus in technical terms, but the harder issue now sits in network and data-layer scaling. The report also said post-quantum upgrades for the P2P handshake protocol and KZG commitments related to EIP-4844 remain on the to-do list.