KoinlyticsMulti-Party Computation splits a secret across multiple parties in a way that lets them compute functions of it (like signing) without ever reconstructing the whole thing in one place. In crypto custody, that means a private key that never fully exists.
Multisig vs MPC vs Threshold signatures
- Multisig. A smart contract requires m-of-n signatures from n distinct keys. Each key exists. Simple. On-chain visibility. Every party knows a full key.
- MPC / Threshold signatures. A single key is mathematically split into n shares. Any m of them can produce a signature via a protocol run, and the final signature looks like a normal single signature on-chain. No party ever holds a full key.
Why MPC won institutional custody
- Signatures are compact and chain-agnostic. No custom contract needed.
- Off-chain policy: which m-of-n signs depends on your rules (amount limits, whitelists, KYC steps).
- Key rotation is possible without moving funds.
- Recovery via re-sharing: if one share is lost, redistribute among remaining parties.
The catch
The signing protocol itself is complex. Bugs in threshold ECDSA implementations have leaked keys before (see BitForge, TSSHOCK research). Auditing MPC libraries is much harder than auditing Solidity. Custodians build in defense-in-depth: HSM per share, jurisdictional separation, per-share hardware attestation.
Where you meet MPC
- Fireblocks (custody + wallet infrastructure).
- Coinbase Institutional (custody).
- Web3Auth / Torus (consumer login, split between user device and network).
- Zengo, Yope, and similar consumer wallets (no seed phrase, MPC recovery).
MPC vs smart-contract multisig for regular users
For a single user with a $1M portfolio: hardware wallet + on-chain multisig (Safe) is cheaper and more transparent. For a company managing hundreds of wallets and needing granular policy: MPC. For consumer apps that want no seed phrase: MPC.