Institutional Crypto Custody: How Big Money Stores Coins

Institutional crypto custody in 2026: how the big money stores coins

When an ETF holds billions in Bitcoin, or a corporation like Bitmine sits on 6,016,414 ETH (about 4.9% of all Ether), a practical question arises: where exactly are those coins, and who can move them? The answer is crypto custody — the systems, technologies, and procedures institutions use to store digital assets securely. As of October 19, 2026, custody has become one of the industry’s most consequential and least understood battlegrounds, deciding which institutions can participate in crypto at all.

This article explains how institutional custody works, the technologies involved, what goes wrong (including a $6 million lesson from October 2026), and how to evaluate custody when choosing where to put your own money.

Why custody is crypto’s hardest problem

Crypto custody is uniquely difficult because digital assets combine two properties nothing else has: bearer-instrument finality (whoever controls the private key owns the coins, like cash) with digital transferability (keys can be copied and funds moved globally in seconds, like data). Cash can be locked in a vault; data can be backed up. Crypto private keys must be simultaneously usable (to sign transactions) and unguessable, uncopyable, and indestructible — a combination with no perfect solution.

For individuals, the stakes are personal. For institutions, they’re existential and regulatory: a fund managing other people’s money must demonstrate that assets can’t be stolen by insiders, lost to technical failures, or seized — while remaining accessible enough for daily operations. Every custody design is a trade-off between security and operability, and the history of crypto is largely the history of getting that trade-off wrong.

The numbers make the stakes concrete. September 2026 saw ~$766 million stolen, including the $387.5 million Bitget hot-wallet breach. The difference between Bitget’s hot wallets (emptied) and cold wallets (untouched, per the exchange) was custody design working as intended — the breach was catastrophic but not existential precisely because cold storage held.

The custody spectrum: from phone wallet to qualified custodian

As requirements scale — more assets, more regulation, more people involved — institutions move down this spectrum toward specialized solutions. Each step trades some sovereignty for some safety.

Multisig: security through shared control

Multisignature (multisig) wallets require M-of-N authorized parties to approve a transaction — for example, 3-of-5, where any three of five keyholders must sign. No single person can move funds alone, which eliminates the single-point-of-failure that plagues individual wallets.

Multisig is the workhorse of institutional crypto: team treasuries, DAO funds, and exchange cold storage all rely on it. The security properties are intuitive — an attacker must compromise multiple independent keyholders, ideally in different locations with different devices. But multisig has failure modes of its own: coordination risk (keyholders unavailable when urgent action is needed), keyholder compromise (insider threats scale with the number of keyholders), and — as October 2026 demonstrated — access-control failures in the software around the multisig.

MPC: the key that never exists

Multi-party computation (MPC) takes a different approach: the private key is split into encrypted shares distributed across parties or devices, and transactions are signed collaboratively without the full key ever being assembled in one place. There’s no single key to steal because no single key exists.

MPC’s advantages for institutions are operational: signing policies can be complex (time locks, amount limits, approver hierarchies) without the rigidity of multisig’s fixed M-of-N, and key shares can be refreshed periodically without changing the underlying wallet address. The trade-off is complexity — MPC cryptography is subtle, implementations vary in quality, and auditing them requires rare expertise. In 2026, MPC underpins most institutional-grade wallet infrastructure, usually invisibly to the end user.

Cold vs. hot storage at institutional scale

The hot/cold distinction scales up dramatically for institutions:

• Hot wallets hold operational funds — enough for daily withdrawals and trading, typically a small fraction of total assets. They’re online, fast, and exposed. The Bitget breach hit hot and warm wallets; the design assumption is that hot wallets can be breached, so their contents are capped at survivable levels.
• Warm wallets stage funds between hot and cold — online but with additional controls and approval steps.
• Cold storage holds the bulk of assets on devices that have never touched the internet — often in geographically distributed vaults with physical security rivaling bank vaults. Moving funds requires coordinated human ceremonies: multiple people, multiple locations, deliberate friction.

The institutional discipline is minimizing hot balances and automating the refill process so human intervention (and human error) stays minimal. Every dollar in hot storage is a dollar of accepted risk; the art is keeping that number as small as operations allow.

The $6M lesson: the Base vault incident

October 2026 delivered a custody case study that every institution should memorize. In the Base vault exploit, an attacker added a malicious contract to the whitelist of a vault, then borrowed its receipt tokens and redeemed them through Aave V3 for approximately 1,783 wstETH — about $6 million.

The critical detail: neither Base’s nor Aave’s core contracts were compromised. The vault was controlled by a 3-of-7 multisig — textbook institutional setup — and the losses came from an access-control failure on that multisig. Someone with sufficient authority added the attacker’s contract to the whitelist, whether through compromised credentials, social engineering, or insider action, and the multisig’s own rules permitted it.

The lesson is profound and general: custody is only as strong as its governance. The cryptography worked. The multisig math worked. What failed was the human-and-software layer deciding what the multisig was allowed to approve. Institutions spend millions on key management and then lose funds through admin panels, whitelists, and operational procedures — the unglamorous plumbing around the cryptography. In 2026, the attack surface isn’t the vault door; it’s the list of who’s allowed through it.

How ETF custody works: billions under professional lock and key

Spot crypto ETFs — now holding billions across BTC, ETH, XRP, and SOL funds — run on qualified custodians: regulated entities specializing in digital-asset safekeeping. The model:

• Segregated cold storage: each fund’s assets held separately from the custodian’s own assets and other clients’, in offline wallets.
• Institutional key ceremonies: key generation and storage follow audited procedures with multiple independent participants.
• Insurance: commercial crime and specie insurance policies cover defined loss scenarios (with meaningful exclusions — read them).
• Regulatory oversight: custodians operate under trust charters or similar frameworks, with examinations and capital requirements.
• Daily reconciliation: fund administrators verify holdings against fund shares outstanding — the mechanism that makes ETF arbitrage and creation/redemption work.

This is why ETF investors accept counterparty risk: in exchange for trusting the custodian, they get professional-grade security, insurance, and regulatory recourse that self-custody can’t offer. The trade is rational for most traditional investors — but it concentrates enormous assets in a handful of custodians, creating systemic concentration risk the industry is only beginning to discuss.

Proof of reserves: trust, but verify

After every major exchange failure, the industry rediscovers proof of reserves — cryptographic demonstrations that a custodian actually holds the assets it claims. The concept: the custodian publishes Merkle-tree proofs letting any user verify their balance is included in the total, without revealing other users’ data.

Proof of reserves is valuable but limited. It proves assets exist at a point in time; it doesn’t prove liabilities don’t exceed them (that’s proof of solvency, much harder), nor does it prevent the custodian from moving funds afterward. It’s a snapshot, not a guarantee. Still, in a post-Bitget world, the absence of proof of reserves is itself information: custodians who won’t demonstrate holdings are asking for trust they’ve done nothing to earn.

The direction of travel is toward real-time attestations and proof of solvency — harder problems, but the ones that actually prevent the next insolvency from hiding behind a reserves snapshot.

Insurance: what it covers and what it doesn’t

Custody insurance exists but is narrower than most assume. Typical coverage includes:

• Covered: theft by external attackers, insider theft (with sublimits), physical loss of key material, some operational errors.
• Often excluded: losses from smart-contract exploits in DeFi positions, market losses, regulatory seizures, and — critically — losses above policy sublimits that are small relative to assets under custody.
• The fine print matters most: “hot wallet” coverage is typically a fraction of cold-storage coverage, and policies often exclude the exact novel attack vectors that cause the biggest losses.

Insurance is a backstop, not a strategy. No policy covers everything, and the industry’s $2.68 billion in 2026 losses dwarfs total available coverage. Treat insurance as one layer among many — never the plan.

Evaluating custody: what to ask before trusting anyone with coins

Whether you’re choosing an exchange, a fund, or a custodian, ask these questions:

1. Who holds the keys, and how? Multisig? MPC? Single key? (If they won’t say, that’s your answer.)
2. What’s the hot/cold split? What fraction sits in online wallets? Smaller is better.
3. Is there proof of reserves? Published, recent, and verifiable — or just claimed?
4. What insurance exists, specifically? Coverage amounts, exclusions, and whether hot wallets are meaningfully covered.
5. Who governs the governance? The Base vault lesson: ask about admin controls, whitelists, and who can change the rules — not just the cryptography.
6. What’s the incident history? Past breaches handled transparently build more confidence than unblemished claims.
7. What happens in bankruptcy? Are client assets legally segregated and bankruptcy-remote? This question mattered enormously in previous exchange failures.

The future of custody: where it’s heading

Three trends will define custody’s next phase. First, regulation is formalizing it: the SEC’s October 2026 proposals on funds holding crypto and the advancing CLARITY Act point toward standardized custody requirements — good for user protection, costly for smaller players, and likely to accelerate consolidation among custodians. Second, technology is distributing it: MPC advances and emerging threshold-signature schemes keep pushing the frontier of “secure without centralizing.” Third, transparency is becoming competitive: after September 2026’s $766M lesson, proof of reserves, real-time attestations, and published security practices are shifting from nice-to-have to expected — and institutions that resist will lose business to those that don’t.

The end state is probably bifurcated: a regulated, insured, transparent custody layer for institutions and mainstream users — and self-custody remaining for those who value sovereignty over convenience. Both are legitimate. What’s not legitimate anymore, after 2026, is opacity: any custodian asking for your trust without evidence is selling the past, not the future.

There’s a final irony worth sitting with. Crypto was invented to eliminate trusted third parties — and its institutional future is being built on the most trusted third parties imaginable: regulated custodians, insured vaults, audited key ceremonies. The technology didn’t eliminate trust; it relocated it, priced it, and insured it. Whether that’s a betrayal of the original vision or its maturation into something that can actually hold the world’s money is the defining philosophical question of custody in 2026. The market, voting with trillions, seems to have chosen an answer.

Custody lessons for individuals: what to copy from the big money

You don’t need billions to benefit from institutional custody thinking. Several big-money practices scale down perfectly:

• Minimize hot balances. Institutions keep most assets cold; you should too. A hardware wallet is personal cold storage — use it for anything you can’t afford to lose, and keep only spending money in hot wallets.
• Separate roles. Institutions split duties across people; you can split across devices and locations. One hardware wallet at home, a backup seed in a different physical location, and a hot wallet for daily use mirrors the institutional hot/warm/cold tiering.
• Document procedures. Institutions run key ceremonies from checklists; you should write down your own recovery procedure — where seeds are, how to restore, in what order — and store it separately from the seeds themselves.
• Test recovery before you need it. Institutions drill. You should verify that your seed phrase actually restores your wallet before you depend on it — with a small test amount, on a clean device.
• Govern your own governance. The Base vault lesson applies personally: review what your wallets are allowed to do. Revoke old token approvals, remove unused connected sites, and audit your own whitelist the way an institution audits its admin panel.
• Plan for incapacitation. Institutions have succession procedures. Individuals need them too — which brings us to the hardest problem in custody.

The inheritance problem: crypto’s unsolved custody challenge

Here is the darkest corner of custody, institutional and personal alike: what happens to the coins when the keyholder dies? With traditional assets, probate courts and banks manage the transfer. With crypto, if nobody else can access the keys, the assets are permanently lost — not transferred, not escheated to the state, just frozen forever on the blockchain. Uncounted billions are already estimated to be lost this way across crypto’s history.

Institutions handle this with legal structures: keys held by corporate entities survive individual deaths, with succession procedures defining who assumes control. But even institutions struggle with the details — authorized-person lists go stale, procedures aren’t tested, and the exact scenario (founder dies unexpectedly) is the one least likely to have been drilled.

For individuals, the options are all imperfect:

• Shared multisig with heirs (e.g., 2-of-3 with a lawyer holding one key) — robust but complex to set up and maintain.
• Documented instructions in a will or safe deposit box — simple but creates a single point of compromise: whoever reads the document gets everything.
• Time-locked dead-man’s switches — smart contracts that transfer funds if the owner doesn’t check in periodically. Elegant in theory; risky if the owner is merely incapacitated or loses access temporarily.
• Specialized inheritance services — an emerging category, but young, unproven, and itself a counterparty risk.

The honest advice: any inheritance plan beats none. Write down where your assets are and how to access them, store that information securely but accessibly to your heirs (not with the keys themselves), and tell at least one trusted person that the plan exists. The biggest risk isn’t choosing the wrong method — it’s choosing nothing and letting the blockchain become a mausoleum.

Frequently Asked Questions

What is crypto custody?

Crypto custody is the secure storage and management of digital assets’ private keys by individuals or institutions. It ranges from personal hardware wallets through multisig and MPC setups to regulated “qualified custodians” that safekeep billions for ETFs and funds. Good custody balances security against operational accessibility — and in 2026, getting that balance right has become a competitive advantage, not just a cost center.

What is multisig and why do institutions use it?

Multisignature (multisig) wallets require M-of-N authorized parties to approve transactions (e.g., 3-of-5). Institutions use it because no single person can move funds alone, eliminating single points of failure. But multisig’s security depends on its governance — as the October 2026 Base vault incident showed, an access-control failure on a 3-of-7 multisig cost ~$6M despite the cryptography working perfectly.

What happened in the Base vault exploit?

An attacker got a malicious contract added to a vault’s whitelist, then drained about 1,783 wstETH (~$6 million) through Aave V3. The vault used a 3-of-7 multisig, but the loss came from an access-control failure — not broken cryptography. Neither Base’s nor Aave’s core contracts were compromised. It’s 2026’s clearest lesson that custody fails at the governance layer, not the math layer.

How do Bitcoin ETFs store their Bitcoin?

Through regulated qualified custodians using segregated cold storage, institutional key ceremonies, commercial insurance, regulatory oversight, and daily reconciliation against fund shares. Investors trade counterparty risk for professional security and recourse — a rational trade for most, but one that concentrates assets in a handful of custodians.

What is proof of reserves?

Cryptographic proof (typically Merkle-tree based) that a custodian holds the assets it claims, letting users verify their balances are included. It’s valuable but limited: it proves assets exist at a moment in time, not that liabilities don’t exceed them, and it can’t prevent subsequent movement. Absence of proof of reserves is itself a red flag.

Is self-custody or institutional custody safer?

Neither is categorically safer — they fail differently. Self-custody eliminates counterparty risk but concentrates everything on you: loss, error, and no recourse. Institutional custody adds professional security, insurance, and recourse but introduces counterparty and concentration risk. Many sophisticated holders use both: self-custody for sovereignty, institutions for convenience and scale.

This article is for informational purposes only and is not investment advice.

Related Reading

• DeFi in 2026: The State of Decentralized Finance
• Crypto Scams in 2026: How to Spot and Avoid Them
• Who’s Buying Crypto in 2026? Market Structure Map