The case against mass self-custody is easy to make. It asks ordinary people to become their own security team, backup administrator, inheritance planner and incident responder. A lost recovery phrase has no help desk. A compromised key has no chargeback. A flawless device still depends on a trustworthy purchase route, sound randomness, careful backups and correct use.
Our recent analysis of the CryptoBilis investigation and the self-custody trust chain reached an uncomfortable conclusion: controlling keys is not a product somebody buys. It is an operating practice that must remain intact for years.
That makes self-custody poorly suited to many holders. It does not make it unimportant. The opposite may be true.
Bitcoin's most valuable contribution to people who use exchanges, neobanks or investment products may be the existence of a minority that refuses to delegate. Those users preserve a credible right of exit. Every custodian knows that its customers can, at least in principle, leave the institution and hold the underlying asset directly.
Self-custody is therefore not only private risk management. It is market infrastructure.
Adoption does not require eight billion security experts
The Bitcoin white paper describes a system for direct electronic payments without a financial institution acting as the required intermediary. It does not say that every economic activity built around Bitcoin must exclude service providers.
That distinction matters. An open protocol can support both direct use and managed services. Email is an open protocol even though most people do not run their own mail server. Open-source software remains open when companies host it for customers. The key property is not universal self-operation. It is whether people retain a practical alternative to the operator.
Bitcoin faces the same scaling reality. Many users prefer an interface that handles account recovery, reporting, inheritance and security monitoring. The preference becomes stronger when the value at risk is life-changing. Delegation can be rational, especially when the user is more likely to lose a seed than a regulated provider is to fail.
The trade-off changes by person and jurisdiction. Someone facing capital controls, institutional collapse or targeted exclusion may value direct key control more than someone with access to stable banks and credible courts. The same hardware wallet that protects one holder from a fragile institution can expose another to theft, coercion or an unrecoverable operational mistake.
There is no universal custody answer because there is no universal threat model.
People often escape a currency before they escape intermediaries
Usage data also weakens the idea that economic need automatically produces mass self-custody.
Chainalysis estimates that centralised exchanges handled 64% of Latin American crypto activity in its 2025 regional study, more than in North America or Europe. That figure measures activity through a venue, not the share of people who ultimately control their own keys. It cannot be used as a direct custody statistic.
It still reveals something important. In regions where inflation, capital controls and currency instability create a strong reason to use crypto, centralised services remain the dominant entry point. Many users first want to leave a weak national currency, access dollar stablecoins or move money across borders. They do not necessarily want to eliminate every trusted relationship at the same time.
Convenience is not evidence that people misunderstand Bitcoin. It is evidence that sovereignty has costs. Users compare those costs with the risks already in their lives and choose the service layer that solves the immediate problem.
Crises produce exits, but not permanent conversion
Custodian failures can change that calculation quickly. When FTX collapsed in November 2022, Glassnode recorded a 72,900 BTC decline in aggregate exchange balances over seven days, one of the largest weekly falls in its dataset. Coins moved from venues into investor-controlled wallets at a historic rate.
That response demonstrated the value of the exit. Customers elsewhere did not need permission from the industry to react to a failure at one exchange. Those who could withdraw had an alternative that did not require finding another company first.
But a crisis-driven withdrawal wave does not prove that the whole market will remain in self-custody. Fear changes behaviour immediately; convenience tends to reclaim users gradually. New participants arrive without experiencing the previous failure, and existing holders rediscover trading, lending, recovery and reporting features offered by intermediaries.
Expecting the next bankruptcy to convert everyone permanently into a sovereign operator misunderstands how adoption works. A functioning market must assume that people will continue to delegate, then make that delegation contestable.
The right to leave is stronger than a promise to behave
A custodian normally competes with other custodians on price, interface, security and trust. Bitcoin adds a competitor that traditional financial products rarely face: the customer can remove the asset from the category of custodial services altogether.
That possibility changes incentives. An exchange that delays withdrawals, a bank that adds excessive fees or a platform that weakens privacy is not only asking customers to switch brands. It is reminding them why the protocol was built to work without the brand.
The effect resembles open-source software. A company may host the software far better than most customers could operate it themselves. Yet the ability to export data and run another instance limits how completely the company can capture the user. The exit option has value even for customers who never use it.
Hal Finney anticipated a layered Bitcoin economy in 2010. He argued that Bitcoin-backed banks could issue their own redeemable digital claims while the base asset served as reserve money. That vision did not require every payment to settle directly on-chain. It required the claims to remain anchored to an asset outside the issuer's control.
The modern market is not Finney's design. Exchanges, custodians and spot products make different promises, and an ETF share is not a Bitcoin withdrawal claim for a retail investor. The useful principle survives: intermediaries can sit above an open monetary layer without owning that layer.
Deterrence is not insurance
Calling self-custody a form of deterrence can be helpful, provided the metaphor is not stretched too far.
The ability to withdraw cannot recover coins after a custodian has lost or misappropriated them. It cannot help a customer after withdrawals are frozen. It does not force every provider to maintain full reserves, and it does not replace audits, regulation, governance or legal recourse.
Exit works before failure. It gives customers a response to deteriorating service, changing terms or rising concern while the asset is still available. It also creates a benchmark for custody: proof of control over the keys remains possible outside the institution, so providers must explain what users receive in exchange for surrendering that control.
The deterrent weakens when withdrawing becomes technically difficult, expensive, culturally marginal or legally restricted. If almost nobody can hold Bitcoin safely, the theoretical right to leave stops placing much pressure on the companies that hold it for everyone else.
This is why hardware security, wallet usability, multisignature standards, full-node accessibility and recovery design are public concerns. Better self-custody tools do more than protect their direct users. They preserve a credible alternative for the whole market.
Bitcoin needs both adoption and refusal
There is a real tension between Bitcoin as a mass reserve asset and Bitcoin as a purely self-custodied asset. Deep liquidity, retirement products, institutional balance sheets and regulated access can broaden ownership and reduce some forms of market friction. They also concentrate keys and rebuild layers of trust around an asset designed to make trust optional.
Rejecting every intermediary would keep that trust surface small, but it would also exclude many people who will never accept the operational burden of direct key management. Accepting total intermediation would improve convenience while allowing the asset's defining escape hatch to rust shut.
The durable position lies between those extremes. Most holders may delegate. Some must remain able and willing not to. The first group brings scale; the second preserves the conditions under which scale does not become capture.
Self-custody does not need to become the default for everyone to remain fundamental to Bitcoin. It needs to stay usable enough that a custodian knows its customers have somewhere else to go.
The few who keep their own keys are not only protecting themselves. They keep the exit open for everyone who chooses to stay inside.
This is market analysis, not a recommendation to buy or sell any security.