Crypto for Private Wealth: Custody, Tax, Structures

For a family of significant wealth, crypto has become an ordinary asset class with its own infrastructure: custodians, banks, OTC desks, auditors. It is managed under the same rules as a securities portfolio or real estate, except that the price of an operational mistake is higher — a lost key cannot be recovered, and an account frozen by compliance is not quickly unfrozen either.

The era of anonymity has ended in parallel. The OECD extended automatic exchange of tax information to crypto-assets: the first exchanges under the CARF standard take place in 2027. The EU built a single authorisation regime for providers under MiCA, Hong Kong and Singapore closed their own regimes, and the United States passed a federal statute on payment stablecoins. In the theory of five flags the digital flag survives, but its content has changed: it is now a choice of provider jurisdiction, custody quality, and reporting discipline.

Section map: which question to start with and where the answer is.

Concept

Crypto in private wealth comes down to five decisions.

1. Custody: who controls the private keys and what happens on a hack, a custodian insolvency, or the owner's death.
2. Banking: which bank will accept funds of crypto origin and on what file.
3. Regulatory perimeter: in which jurisdictions the exchange, the custodian, and the OTC desk sit.
4. Tax: which events are taxable and where the holder's residency delivers a favourable regime.
5. Succession: how access to the assets outlives the owner.

The decisions are interlocked: the choice of custodian shapes the banking options, the provider's jurisdiction determines where reporting goes, and the ownership structure governs what heirs receive and in what order.

Custody: the full spectrum

At one end sits self-custody: the keys are with the owner. For amounts that would hurt to lose, a single seed phrase in a safe has long counted as poor practice. The standard is multisig, where a transaction requires, say, two signatures out of three with keys distributed across people and locations, or MPC: cryptographic splitting of the key into shares such that the complete key never exists on any single device. Institutional MPC infrastructure comes from platforms such as Fireblocks; a family's operational security is covered in a separate piece.

At the other end sit qualified custodians and the banks that have added crypto custody to classic private banking. Between them the axes that matter are the same four: which entity holds the licence, who may be its client, what protects the fiat side, and what happens to the coins if the holder fails.

Client scope, fiat protection and what happens to the coins in the provider's insolvency differ by provider and are compared, with entry thresholds, custody models, fiat rails, published fees and the choice by holder profile, on the crypto banks and custodians map. For trusts and funds an independent custodian is usually mandatory: a fiduciary may not keep client keys on its own laptop, and a trust charter delivers segregation without insurance.

Banking: source of funds

The weak link in any crypto strategy is the fiat account. A sale on a regulated exchange guarantees nothing by itself: the proceeds must be accepted by a bank whose compliance function is prepared to work with crypto origin, and most classic private banks still decline. The specialist segment consists of Swiss banks AMINA and Sygnum, built around digital assets and reading on-chain history as a matter of routine; part of the flow is handled by EMIs — payment institutions with crypto-tolerant onboarding, but without a banking licence and without deposit protection.

The decisive document at the gate is proof of source of funds and proof of source of wealth. For crypto that means addresses and transaction hashes, exchange statements going back years, contracts for the original purchases, and tax returns where the income has already been declared. The bank will run the addresses through blockchain forensics, and the declared history must reconcile with the on-chain data: gaps in the chain have to be explained with documents. The longer the asset lived in self-custody, the harder the file is to assemble, so the evidence pack is prepared in advance, before onboarding.

The holder's regulatory map

Licences are held by the provider, but it is the holder who picks it, and the provider's jurisdiction settles three questions: which law protects the assets on its insolvency, which regulator to complain to, and which tax authority the reporting reaches.

The licensing regimes behind these jurisdictions, from MiCA authorisation to the AIFC, are compared on the crypto licence map. For the holder the choice splits across three questions, each with its own page: where the tax on a disposal falls (crypto tax by country), which provider in that jurisdiction will actually take the assets and the fiat (crypto banks and custodians map), and which reporting wave the jurisdiction sits in (CARF in practice).

Tax: events and favourable jurisdictions

In most systems crypto is characterised as property for tax purposes. A taxable event arises on a sale for fiat, on a token-for-token swap, on paying for goods, and often on receiving staking rewards and airdrops. Buying with fiat and moving between one's own wallets generally create no event but still require records: without a fixed acquisition cost, any future disposal turns into a dispute over the tax base.

Which country may tax is decided by the owner's tax residency, and the spread between jurisdictions remains wide.

Rates and characterisation across the rest of the world are collected in the country tax comparison.

CARF: the transparency timetable

Residency and disposals are worth planning against what tax authorities will see on their own. CARF is the OECD standard for automatic exchange of information on crypto-assets: exchanges, custodians, and brokers report client and transaction data to each client's country of tax residence, mirroring the mechanics of CRS for bank accounts. Under the Global Forum commitment list (in its edition of 14 September 2026), 77 jurisdictions have committed. The timetable for first exchanges splits into three groups.

The data is already being collected: jurisdictions in the 2027 group required providers to record transactions from the start of 2026, and in the United Kingdom, for instance, the first report for 2026 goes to HMRC by 31 May 2027 — with penalties of up to £300 for each user whose data is missing or inaccurate.

Structures: personal, trust, fund

Personal ownership remains the default: minimum cost, a clean tax picture. A structure is justified for succession, asset protection, joint ownership, or institutional access; as a tool for concealment it is useless — the beneficial owner is disclosed under UBO rules, and without real economic substance both the tax authority and the bank will look straight through the arrangement.

Six wrappers cover the field, and a seventh — the exchange-traded product — removes the keys from the family altogether. The axis that separates them in practice is who holds the keys once the wrapper is in place, and who answers for them.

The exchange-traded row trades control for simplicity. iShares Bitcoin Trust charges a 0.25% sponsor fee and states that it is not an investment company registered under the Investment Company Act of 1940, so its holders do not get mutual-fund protections. Tax follows the note rather than the coin in most residences, with one notable exception: under the German Finance Ministry letter of 6 March 2025 (paras 84–85) a note that gives only a claim to delivery of deposited crypto-assets or to their sale proceeds is a claim in kind, so its sale falls under the private-disposal rules with the one-year holding period instead of the flat tax on capital income. For succession the product is an ordinary securities position: heirs inherit a brokerage account, not keys, and cannot take the coins on-chain.

The independent custodian is the constant across five of the six structure rows: a trustee, a foundation board, a fund depositary and an insurer all contract custody out to a licensed provider, so the wrapper sits on top of the custody line and never replaces it. Which trustees genuinely accept digital assets, and where the statute lets the settlor keep investment powers over them, is the crypto axis of the trust jurisdictions map; the providers a fiduciary can actually contract with are on the crypto banks and custodians map.

Tax rarely decides the choice. Attribution rules — CFC for companies, grantor and settlor-interested rules for trusts, look-through for foundations — push the income back to the person who funded the structure in most residences, so a wrapper earns its keep through succession, protection and institutional access. Reporting follows the same logic: a wrapper changes the account holder, not the visibility, because the trust, foundation or company becomes the account holder and its controlling persons are reported behind it.

Custody dominates the cost line: custodians charge a percentage of the assets they hold, so the fee grows with the position, and the published schedules are compared on the crypto banks and custodians map; trustee, board or administrator fees sit above that number. A structure that does not solve a specific succession, protection or access problem therefore adds cost without changing anything the holder can point to.

A trust holding crypto faces a specific problem: custody on the trustee's side. A trustee carries a fiduciary duty to safeguard the trust property, yet professional trust companies are not prepared to hold private keys themselves. Practice has produced three answers.

1. An account with a qualified custodian opened in the name of the trust.
2. A private trust company into which custody competence is hired.
3. Delegation of investment management to a specialist manager with assets held by an independent custodian.

Key authority and liability for a compromise are written into the trust deed as a separate section — otherwise a dispute over who was responsible for the hack is guaranteed.

Families that trade actively or bring in outside co-investors wrap the position into a fund: an administrator, an auditor, and an independent custodian turn a set of wallets into verifiable NAV reporting that banks and tax authorities understand. Running costs exceed those of a trust, but institutional access opens up — prime brokerage, lending against the position, subscriptions from external investors.

Key succession

A private key known only to its owner makes the asset permanently inaccessible after death: a will transfers the right, but a right without the key is worth nothing. A succession protocol is assembled from three parts.

1. The technical part is multisig or MPC with shares distributed among heirs, trusted persons, and a custodian; this also covers the dead man's switch — a mechanism that releases access automatically after a long period of owner inactivity, useful as a backstop and carrying no legal force of its own.
2. The documentary part consists of sealed access instructions held by a notary or trustee, stored separately from the will: during probate the will becomes a public document, and a seed phrase has no place in its text.
3. The organisational part is a succession plan in which the executor knows the digital assets exist and the heirs know whom to approach for access.

Rehearsing recovery during the owner's lifetime exposes gaps in the protocol far more cheaply than a probate file does.

Stablecoins in the family treasury

Stablecoins serve a family treasury as settlement cash: transfers between jurisdictions in minutes, parking proceeds between deals, denominating obligations in dollar equivalent without opening an account in every country. In exchange, the holder takes on the issuer's credit risk and the quality of its reserves.

Regulation has shifted the choice towards licensed instruments. In the EU, MiCA governs the issuance of e-money tokens; in the United States the GENIUS Act — signed on 18 July 2025 as Public Law 119-27 — admits only approved issuers with segregated reserves to payment stablecoin issuance and leaves the tokens outside federal deposit insurance; in Hong Kong the HKMA licenses issuers. An issuer for treasury balances is picked on the law of issue, reserve composition, and audit; long-term savings are not held in stablecoins.

Reporting: CARF or the amended CRS

Digital dollars come in four legal forms — a tokenised bank deposit, a payment stablecoin, a MiCA e-money token and a tokenised money market fund — and they do not all travel the same reporting channel. In the EU the split is written into Directive (EU) 2023/2226, which added the crypto rules and at the same time widened the CRS: a Reportable Crypto-Asset is any crypto-asset other than a central bank digital currency, electronic money, or an asset the provider has determined cannot be used for payment or investment. Whatever falls out of that definition does not disappear from reporting — it lands in the CRS instead, whose Depository Account now expressly covers an account representing e-money held for a customer.

The dividing line is a five-part definition of electronic money, applied to the token's own terms: a digital representation of a single fiat currency, issued on receipt of funds, a claim on the issuer in the same currency, accepted in payment by third parties, and — the decisive limb — redeemable at any time at par by virtue of regulatory requirements. A euro e-money token clears that last condition, because MiCA gives its holder the right to redeem at par at any moment and at no charge, so it is reported as a balance once a year. A dollar token whose issuer sets a redemption minimum and charges a fee for it does not clear it, and its whole transaction history goes into the crypto channel instead.

For a family treasury the practical consequence is granularity rather than secrecy. Both channels reach the same tax authority, but the CRS return shows a balance and interest while the crypto return shows every disposal and every transfer out, including transfers to the family's own self-hosted addresses. EU providers have been collecting that data since 1 January 2026 for the first reports, so a treasury that moves balances between an e-money token, a stablecoin and a tokenised fund needs a transaction log that reconciles with two returns instead of one.

Exiting into fiat

A large sale through an exchange order book moves the price and fragments into dozens of fills. Positions in six figures and above are sold through an OTC desk: the price is fixed before the trade, settlement runs as a single transfer, and the KYC file is agreed in advance. The proceeds are received by a bank already familiar with the source-of-funds file — a specialist crypto bank or a classic private bank with the evidence pack assembled. Readiness on each side matters more than speed here: settling into an unprepared account ends in a freeze.

A sale is not the only way to reach fiat. A private bank that holds the position in custody will lend against it: Lombard lending against a BTC or ETH position keeps the position and its acquisition cost intact and creates no disposal to tax — the source of funds for the next bank is a loan agreement and a drawdown notice rather than a settlement statement. The trade-offs are the margin call if the collateral falls, the interest cost against an unrealised gain, and the short list of banks that take crypto as collateral at all. Most private banks set loan-to-value case by case and do not publish it; the OTC desks page compares four lenders on size, liquidation threshold and what happens to the collateral, and the providers that offer the facility are marked on the crypto banks and custodians map.

Risks

The crypto-specific risks for a holder concentrate in five places.

Crypto in private wealth rests on five decisions: keys (multisig, MPC, or a qualified custodian), the bank (who will accept the source of funds), the provider perimeter (MiCA, Hong Kong and Singapore licensing, BitLicense), tax (residency and realisation events), and succession (a documented access protocol). Secrecy is no longer part of the design: 46 jurisdictions begin CARF exchanges by 2027, and the configuration that holds is the one that withstands questions from the bank, the tax authority, and the heirs at the same time.