pledgeCustody is a control decision before it's a brand decision.
Compare 11 bitcoin custody providers by who actually holds the keys, how recovery works, and what happens if the custodian fails. We group by custody model first — collaborative and self-custody versus institutional qualified custody — and never rank across models, because an 8.0 self-custody score and an 8.0 institutional score solve different jobs.
All 11 tracked providers carry an 8-factor score.
Collaborative & self-custody / institutional qualified.
Rendered next to every score — never hidden.
Who can move the Bitcoin if the provider fails?
We rank within a custody model — never across them.
A decentralized multisig you partly control and a single trust company holding your keys carry fundamentally different risks — who can move your Bitcoin, what happens in recovery, and what legal structure stands behind the claim. Comparing their scores head-to-head would pretend they answer the same question. They don't. So the desk separates the two families and ranks only within each.
You keep meaningful key control or the keys are split so no single firm — including the provider — can move your Bitcoin alone. Recovery and inheritance support are layered around your participation. These solve a different job than handing the whole thing to one custodian, so they are ranked separately — never against institutional custody.
A single regulated custodian (a trust company, bank, or qualified custodian) holds and signs for your Bitcoin in segregated cold storage. You trade key operations for institutional controls and legal structure. Different control tradeoffs from collaborative custody, so it is ranked separately — never against the self-custody family.
Collaborative & self-custody
You keep meaningful key control or the keys are split so no single firm — including the provider — can move your Bitcoin alone. Recovery and inheritance support are layered around your participation. These solve a different job than handing the whole thing to one custodian, so they are ranked separately — never against institutional custody.
Institutional qualified custody
A single regulated custodian (a trust company, bank, or qualified custodian) holds and signs for your Bitcoin in segregated cold storage. You trade key operations for institutional controls and legal structure. Different control tradeoffs from collaborative custody, so it is ranked separately — never against the self-custody family.
2 of 11 scores are fully verified. The rest are best-effort.
Every score ships with a confidence flag so you can weigh how hard to lean on it. An "estimated" score rests on a verified entity and core facts, but some sub-factors are inferred from public disclosure rather than confirmed at source. It will move if we get stronger evidence.
Sources independently confirmed; sub-factors data-driven. The score will not change without new public information. BitGo and Anchorage Digital are the only two fully verified today.
Entity verified, but some sub-factors inferred from public disclosure rather than primary attestation. The score will move as Pledge receives stronger evidence. Nine of eleven providers carry this flag at launch.
One score is a verdict. Eight show the trade-offs.
- !The Lloyd's policy covers named theft and key-loss scenarios — it has exclusions, so read the coverage scope, not just the word "insured."
- !A claim runs through AnchorWatch's claims process and operational setup, so you are still depending on the company, not only the policy.
- !You still need to understand the shared-control model well enough to operate it under stress.
- !Collaborative setups are more involved than leaving coins at a single institution.
- !The 2-of-3 model only helps to the extent that Onramp, BitGo, and Coincover are operationally independent — that read still needs deeper diligence.
- !Withdrawal flow is slower than single-custodian custody by design.
- !The product is only as good as the user’s willingness to understand and maintain a multi-key setup.
- !It is not the right answer for someone who wants a pure custodian to run everything.
- !Service-heavy products can hide more dependence on people and process than a clean wallet diagram suggests.
- !The exact control boundary can be harder to summarize quickly than straightforward multisig or qualified-custodian models.
- !Heavily institutional — the product is not aimed at individual holders.
- !Multi-asset platform means Bitcoin sits alongside other crypto exposures operationally.
- !Institutional custody can be the wrong mental model for a self-custody-first user.
- !Brand scale should not be confused with fit for a household or individual holder.
- !Institutional access and fit can diverge sharply from what an individual holder actually needs.
- !Big-brand familiarity can mask unresolved questions about relevance, access, and service scope.
- !Single-custodian model — concentration risk lives at one firm, even with strong legal framing.
- !Retail Coinbase consumer custody is a different product with different protections; the two are routinely confused.
- !Single-custodian concentration risk applies as with any qualified custodian.
- !Access tends to flow through institutional and advisor channels rather than direct retail.
- !Operational ties to the broader Gemini exchange business add counterparty considerations.
- !Earn-program history is a separate product but still sits in the same brand context — keep the lines clear.
We compare who controls the keys — not brand familiarity.
About → How Pledge makes money ↗All tracked providers carry an 8-factor score, each flagged verified or estimated.
Collaborative / self-custody and institutional qualified — never ranked across.
No custodian is FDIC/SIPC-insured. "Insured" without an amount + named underwriter is a marketing zero.
Published. No hidden formulas.
How Pledge reviews custody control models.
Each provider is graded 0–10 on eight weighted factors: custody model (18%), key management (16%), insurance (14%), regulatory (12%), recovery (12%), transparency (10%), track record (10%), and loss history (8%). The full rubric, weights, and evidence gates are published.
Not custody advice. Pledge provides custody research for educational purposes only. It does not recommend any custodian, self-custody tool, or key-management design.
Data checked Aug 23, 2026. Provider controls, agreements, insurance scope, and recovery assumptions can change. Scores are category-relative and carry confidence flags (verified/estimated).
No government insurance. No custodian is FDIC/SIPC-insured. An "insured" claim without a disclosed amount and named underwriter should be treated with skepticism.
Affiliate disclosure. Pledge currently earns no affiliate or referral fees from any provider listed. If that changes, monetized links will carry a clear ↗ AFFILIATE marker.