Sui sets a 2026–27 timeline for post-quantum signatures, with vaults first and accounts later

Sui says it will bolt two NIST-approved post-quantum signature schemes onto the network, splitting them by risk profile: ML-DSA-65 as a native protocol scheme for everyday accounts, and the hash-based SLH-DSA-SHA2-128s running inside Move smart contracts for high-value vaults. The plan was set out in the project's own announcement, reported by The Block on Aug. 6.

The two schemes map to separate NIST standards — FIPS 204 for ML-DSA, FIPS 205 for SLH-DSA — and rest on different underlying mathematics, which Sui says means a break in one would not compromise the other. Sui is taking ML-DSA at Level 3 rather than the cheaper Level 1 parameters, citing a July incident in which an AI model halved the effective key strength of a different post-quantum candidate. "While the finding does not affect ML-DSA, the speed of it is reason to carry margin rather than take the cheapest parameters on offer," the team wrote. Neither the candidate scheme, the model, nor the researchers are named in the announcement.

The threat model Sui invokes is harvest-now-forge-later: because onchain public keys are exposed the moment an account transacts, an attacker can collect them today and forge signatures later, once a machine capable of running Shor's algorithm exists. No timeline for that machine is offered.

On migration, Sui says users won't need new seed phrases — keys derive deterministically from a seed, so an existing recovery phrase can produce a quantum-safe key, and Sui's address aliases let an account swap its authorization key without moving assets. The trade-off it acknowledges: post-quantum signatures and public keys are substantially larger than Ed25519, so transactions get bigger. It gives no figures for how much bigger, or what that does to fees.

Key facts

• Schemes: ML-DSA-65 (NIST FIPS 204, Level 3) native; SLH-DSA-SHA2-128s (FIPS 205) in Move contracts — Sui announcement, via The Block
• Quantum-safe vaults targeted for mainnet in 2026; native ML-DSA-65 accounts on testnet by end of 2026; mainnet native account authentication targeted Q1 2027 — Sui announcement
• Rollout is opt-in and additive, following the path used for zkLogin and passkeys — Sui announcement
• SUI was down 1.10% on The Block's quote at publication

The real-world read

Every technical and scheduling claim here traces to one place: Sui's own post. There is no independent audit, no third-party benchmark, and no external confirmation of the July key-strength finding that Sui uses to justify its parameter choice — a finding it declines to identify. Treat it as a roadmap from an interested party, not a shipped feature.

The framing does work, too. Calling this "a routine protocol-feature update rather than a change to consensus" positions Sui as already-prepared while competitors face migrations. But opt-in cuts both ways: "nothing existing changes, and no application needs to do anything today" also means the exposed Ed25519 keys Sui warns about stay exposed for every user who never opts in. The soonest anything lands for ordinary accounts is testnet, roughly five months out.

Note also that The Block discloses Foresight Ventures as its majority investor, with Bitget an anchor LP.

Opinion, and whose

The claim that hash-based signatures kept in-contract preserve compatibility "with whichever post-quantum standards the wider industry settles on, without a core protocol upgrade" is Sui's engineering judgment, not an established outcome. So is its assertion that larger transactions are "the price the whole industry is paying." No outside cryptographer is quoted.

Sources

• Sui (primary): the project's announcement — scheme selection, security levels, migration mechanics, timelines, all direct quotes. This is the subject describing its own roadmap; read accordingly.
• The Block, Danny Park (Aug. 6, 2026): reporting on and quoting that announcement; supplied the SUI price move. The Block discloses Foresight Ventures as majority investor and Bitget as a Foresight anchor LP.

Nothing here is financial advice.