
GUIDE
Crypto Custody Insurance: What Policies Actually Exclude
Crypto Bank Editorial Team · November 17, 2026
A custodian stating “insured” on its homepage is making a claim with a lot of room inside it. Insurance policies covering crypto custody are built from the same building blocks insurers have used for decades — crime cover, specie cover, cyber cover — adapted to digital assets, and each of those building blocks excludes specific things by design. The exclusions are not a defect in the policy; they are the policy. Reading them is the only way to know what “insured” is actually promising.
This matters more in crypto than in most other insured industries because there is no deposit-insurance backstop sitting underneath it the way FDIC cover sits underneath a US bank account, or FSCS cover sits underneath a UK one. Whatever private insurance a custodian has arranged is, in practice, the entire safety net beyond the custodian’s own balance sheet — which is exactly why the specific terms of that net are worth more scrutiny than the single word “insured” usually gets.
There is no single “crypto insurance” product
What gets marketed under one word is usually a combination of distinct policy types, each covering a different peril:
- Crime insurance. Typically covers theft by a third party and, importantly, employee dishonesty — an insider stealing or misappropriating assets. This is generally the policy type covering “hot wallet” exposure, where assets are more operationally active and therefore seen as higher risk.
- Specie insurance. The traditional product for insuring physical valuables in storage (originally built for bullion and fine art in vaults), adapted to cover crypto held in cold storage. Cold storage is generally treated as lower risk than hot wallets, so specie cover is often available at higher limits relative to the same premium.
- Cyber insurance. Covers certain losses arising from cyber incidents — breach response costs, some forms of business interruption — which may or may not extend to the underlying asset loss itself depending on policy wording.
A custodian may carry one, two, or all three, often from different underwriters, each with its own limit, its own retention (the amount the custodian itself absorbs before the policy pays), and its own exclusions. “We are insured” without specifying which of these, for which part of the operation, is not yet a specific claim.
It is also common for a custodian to insure only part of its total holdings — a tier of “excess” cover sitting on top of a smaller primary layer, written by a different syndicate, with its own trigger conditions for when it actually responds. A headline figure quoted publicly is often the sum of these layers, which is one more reason the figure alone, without the layer structure behind it, tells a reader less than it appears to.
What actually gets excluded, in practice
- The client’s own device or credentials. Custody insurance policies generally cover theft of assets from the custodian’s own infrastructure — the vault, the hot wallet, the signing systems it controls. They do not typically cover a loss caused by a client’s own compromised device, phished credentials, or a client voluntarily sending funds to the wrong address. If the custodian never had custody of the specific failure point, the custody policy usually does not reach it.
- Market loss. None of these policy types compensate for a decline in the market value of the asset. Insurance responds to a defined peril — theft, destruction, dishonesty — not to the client simply losing money because the price fell.
- Key-shard and multi-party arrangements. Where a key is split so the client holds one share and the custodian holds the rest (common in multi-party computation and multisig setups — see our multisig vs. MPC guide), whether the policy responds to a loss at all can depend on exact wording about what counts as “the custodian’s” asset versus a jointly controlled one. This is a genuinely contested area in policy drafting, not a settled convention.
- Sub-custodian exposure. If a custodian in turn holds assets through another firm — a sub-custodian — the policy may or may not extend to losses occurring at that second layer, depending on how the policy defines covered locations and covered custodians. See sub-custodians and the hidden chain of custody for why this layer is often invisible from the outside.
- Acts the custodian characterizes as authorized. A transaction the custodian’s own systems treated as properly authorized, even if a client later disputes that characterization, can fall outside a theft-focused crime policy, which is generally written around unauthorized taking, not disputed authorization.
- War, sanctions and regulatory-seizure exclusions. Standard boilerplate across most commercial insurance, these clauses also apply to crypto policies and can be relevant where a government action freezes or seizes assets.
Limits, and who the policy actually names
A headline coverage figure attached to a custodian is usually an aggregate limit — the maximum the policy will pay out in total, across all clients, for the entire period, not a guarantee per client. If losses exceed that aggregate, clients share the payout pro rata, or some claims simply go unpaid. It is also worth checking who the named insured actually is: many custody policies name the custodian itself as the policyholder and beneficiary, with the custodian then responsible for making its own clients whole out of the proceeds, rather than clients having a direct claim against the insurer. That distinction affects what happens if the custodian itself is insolvent at the time a claim would be paid — see client assets in insolvency.
Who actually underwrites this, and why it is concentrated
A large share of institutional crypto custody insurance capacity has historically come through the Lloyd’s of London market, where multiple syndicates can each take on a slice of a single large policy, pooling enough capacity to insure amounts that a single insurer would be reluctant to carry alone. This concentration is itself a risk-relevant fact: a market event large enough to produce simultaneous claims across several custodians could strain the capacity available from a relatively small circle of specialist underwriters, which is part of why this insurance segment has historically been described by participants as cyclical — capacity and willingness to write new cover has expanded and contracted over time as underwriters reassess the risk.
What to ask a provider before relying on its insurance claim
- Which specific policy type covers which specific part of the operation — hot wallets, cold storage, or both?
- Who is the named underwriter, and is the policy, or a summary of it, available to review rather than taken on trust?
- Is the stated limit an aggregate across all clients, or something closer to a per-client guarantee? (In almost every case, it is the former.)
- Does the policy name the client as a beneficiary, or only the custodian, with the custodian responsible for passing proceeds through?
- What is explicitly excluded, in the policy’s own exclusions section, rather than in marketing language?
Where this sits in the Custody Question
This is the second row of the three-row check we run on every provider profile: who insures it. A useful answer names the underwriter, the policy type (crime, specie, cyber), the aggregate limit, and the major exclusions — not just the word “insured.” Where a provider has not disclosed any of that, “no disclosure found” is the honest entry, and it is a materially different situation from a named, specific policy, even though both might get marketed with similar confidence.
FAQ
- Does FDIC or FSCS cover apply to crypto held by a custodian?
- No. Deposit insurance schemes like the FDIC in the US or the FSCS in the UK cover specific categories of bank deposits and, in some cases, cash held incidental to regulated activities — they do not cover crypto-asset custody. Any custody insurance a provider carries is a private commercial policy, not a government-backed deposit guarantee.
- If a custodian names a specific dollar coverage figure, is that the amount I’m personally covered for?
- Usually not directly. That figure is typically the policy’s aggregate limit across all clients and incidents, not a per-client guarantee, and it is subject to the exclusions and retentions in the policy.
- Is self-insurance (a custodian’s own reserve fund) the same as a commercial policy?
- No. A self-funded reserve is the custodian’s own money, subject to the custodian’s own solvency; a third-party insurance policy is a claim against an external underwriter, which is a different kind of protection with its own exclusions and claims process.
- Does carrying a large insurance policy mean a custodian is safer than one without?
- Not automatically. A large aggregate limit is meaningful, but it says nothing about the retention the custodian absorbs first, the exclusions that apply, or whether the custodian’s broader operational and segregation practices (see our segregated vs. omnibus guide) are sound. Insurance is one input into the Custody Question, not a substitute for the other two rows.
Risk note
Insurance terms, limits and underwriters change and are specific to each provider and each policy period. Nothing in this guide describes any named provider’s current coverage; always ask for the policy type, limit, exclusions and named insured directly, or check the provider’s own disclosure.
Not a bank. Not a deposit-taking institution. No FDIC, FSCS or deposit-insurance cover. Editorial only.
Not financial advice.