An MPC wallet can be custodial or non-custodial, because MPC describes signing architecture, not who has legal or practical control over your funds. The custody test is who can approve a transfer, who can block access, who can replace a lost share, and whether the user can leave with recoverable signing control.
This is where custodial vs. non-custodial labels can mislead. A non-custodial wallet gives the user control over signing authority, but an MPC setup may still rely on a provider share, account login, cloud backup, or recovery service. Calling a wallet “self-custodial” while a provider holds a required share is a partial truth at best.
User-Held Shares
A user-held share model can support genuine self-custody when the user can approve transactions and recover funds without a company being able to unilaterally move or block them. The wallet may still use software, cloud backup, or a co-signer in the process, but the provider should not be able to spend alone.
Strong user-controlled designs answer these questions clearly:
- Can the provider move funds without the user approving?
- Can the user recover if the provider shuts down?
- Can shares be exported, rotated, or migrated to another wallet?
- Can an account restriction block signing or recovery?
Users comparing self-custody options should start with self-custodial wallet choices and then narrow by seed phrase, MPC, hardware, or smart wallet design based on their risk tolerance.
Provider-Held Shares
Provider-held shares make setup and recovery easier, but they add a trust boundary. If a company holds a required share, the user may depend on that company for recovery availability, fraud checks, account status, and policy enforcement. That company becomes a single point of failure, just a different one than a seed phrase.
That model can still be appropriate. A custodial setup works for small balances, active trading, or users who need customer support to recover access. It fits the centralized custodial wallet list, where the main acknowledged risk is counterparty control.
Hybrid Recovery
Hybrid recovery sits between pure self-custody and full custody. The user controls one share, the provider controls one share, and a backup method involves passkeys, cloud storage, social recovery, or identity verification. Each of those backup channels is its own potential weak point.
The label “non-custodial” should be backed by specifics, not marketing copy. If the provider can block recovery, refuse co-signing, or close an account in a way that strands funds, the wallet has a provider dependency, even if no single party ever holds the complete private key.