This is our published view
The bench stands behind it
For your facts, in confidence, put the question to the firm.
The bench stands behind it
The position
The honest answer is that it depends where you sit, and that even the strongest position carries a question the rulebooks do not finally answer.
The honest answer is that it depends where you sit, and that even the strongest position carries a question the rulebooks do not finally answer.
The framework does assert that client assets are protected when a firm fails, and it builds that protection highest around custody. There it says so in the plainest terms the framework contains: assets held in custody are not the firm’s assets or liabilities at all (Custody Services Rulebook, Rule III.B.1, current as at 1 July 2026). Where a firm instead uses client assets under the held-on-behalf default, the assets are still treated as the client’s, unless the Client Agreement has reversed that default (VA Management, Rule II.D.3; Lending, Rule II.A.5). So a client’s starting position on a failure runs from strong, in custody, to conditional, under the default, to weakest of all where the default has been switched off and the client is merely owed value.
What the framework cannot do by itself is bind an insolvency. VARA’s rulebooks are regulatory instruments. Whether a ring-fence they assert survives contact with an insolvency is, in the first place, a question of the insolvency law that governs the proceedings, not of the rulebook that asserts the ring-fence. The framework can say client assets are not the firm’s; it cannot, on its own authority, guarantee that a court winding the firm up will treat them that way. That gap, between what the framework asserts and what an insolvency would deliver, is the subject of this article, and it is a question the framework leaves genuinely open.
The clearest sign the gap is real comes from the framework’s own drafting. Where it makes its strongest promise, that certain assets are bankruptcy-remote, it qualifies that promise in the same breath, by reference to what applicable law permits. Section III reads that qualifier closely, because it is the framework marking the limit of its own reach.
So this article sets out what the framework asserts on a failure, reads the qualifier that bounds it, places both against the federal insolvency layer that ultimately governs, and then says, position by position, what a client can and cannot safely assume. It does not claim to resolve what is genuinely unresolved, because here the value is in seeing the question clearly.
What the framework asserts on a failure is a strong protection of client assets at the front end, set, in the very same Part, beside a deference to the insolvency process that would test it. Both halves have to be read, because the gap between them is the article.
The front-end protection is real and specific. In custody, the assets are not the firm’s and cannot be made so; they are segregated and controlled (Custody Services Rulebook, Rules III.B.1 to B.4). Where a firm uses client assets under the held-on-behalf default, they remain the client’s unless the Client Agreement has reversed it (VA Management, Rule II.D.3; Lending, Rule II.A.5). And above the activity rules, every VASP, whatever it does, must maintain a Wind Down Plan (Company Rulebook, Rule VII.A.1, current as at 1 July 2026). That plan must provide for the safekeeping and prompt return of clients’ assets (Rule VII.A.1.c); it must ensure that “the sale of Client Money and/or Client VAs is explicitly excluded from” implementing the plan, any scheme directed by an insolvency appointee, or a going-concern sale of the business (Rule VII.A.1.k); and it must preserve VARA’s power to step in and take control of client assets at its election (Rule VII.A.1.l). Read on its own, this is a framework instructing firms to build a ring-fence designed to hold even against an insolvency appointee.
Then the same Part turns outward. The framework’s insolvency rule provides that, once a firm is in insolvency proceedings, it must cooperate with the Insolvency Appointee to implement the Wind Down Plan, or “any other plan, procedures or scheme as the Insolvency Appointee deems to be commensurate” with the duties imposed by the relevant insolvency proceedings (Rule VII.B.1). The rulebook does not assert that its wind-down plan overrides the insolvency. It requires the firm to work through the appointee, and it accepts that the appointee acts under the insolvency proceedings, not under the rulebook.
So the two rules sit side by side and face in different directions. One tells the firm to design its wind-down so client assets are excluded from any sale, including a scheme run by an insolvency appointee. The other tells the firm that, in an actual insolvency, that appointee proceeds under the duties the insolvency law imposes. Whether the exclusion the framework demands is one the governing insolvency law will honour is the question the rulebook raises and does not answer. It asserts the ring-fence; it concedes the forum. Section III shows the framework marking that same limit inside its strongest promise of all.
The clearest sign the gap is real comes from the framework’s own words, at the point where it makes its strongest promise. For stablecoin reserves the framework goes further than it does anywhere else in protecting client value, and it qualifies that promise in the very same clause.
For issuers of fiat-referenced virtual assets, the reserve rule sets out precisely the outcome a holder would want on a failure. The reserves are to be legally segregated and held remote from the issuer’s own assets; they are not to form part of the issuer’s estate; they are not to be rehypothecated, pledged, encumbered or subject to set-off; and they are not to be subject to recourse by the issuer’s creditors, in particular if the issuer becomes insolvent (Virtual Asset Issuance Rulebook, Annex 1, Rule III.B.4, current as at 1 July 2026). This is the framework naming bankruptcy-remoteness directly, and building the reserve regime around delivering it.
The whole of that obligation opens with one qualifier: the issuer must achieve it “to the furthest extent permitted by applicable laws” (Rule III.B.4). The same phrase governs the issuer’s duty to give VARA priority access to the reserves (Rule III.B.5). Those words are not throat-clearing, and reading them as filler misses the most important thing the clause says. A rule that could deliver bankruptcy-remoteness on its own authority would not need to defer to anything. The deference is the framework acknowledging, in the sentence that makes its strongest promise, that whether the promise holds is decided elsewhere, by the applicable law, and not by the rule.
It is the same move as the insolvency rule in Section II. There, the framework required client assets to be excluded from any sale, then routed the actual insolvency through the appointee acting under the governing proceedings. Here, it specifies estate-remoteness, then bounds it by what applicable law permits. Two of the framework’s strongest client-asset promises, the wind-down exclusion and the reserve remoteness, carry the same outward reference to a law the rulebook does not itself supply.
Which makes one question unavoidable, and the article can defer it no further. If the framework’s protection reaches only as far as applicable law permits, what does the applicable law permit? For a failure resolved in the Emirate, the applicable law is federal insolvency law. Section IV places the ring-fence against it.
The applicable law is federal, and it is the reason the framework qualifies its promises. A VASP’s insolvency in the Emirate is not conducted under VARA’s rulebooks. It is conducted under the UAE’s federal bankruptcy law, before a federal court, and that is the law the rulebooks are pointing to when they defer to “applicable laws.”
The current statute is Federal Decree-Law No. 51 of 2023 on Financial Restructuring and Bankruptcy, in force since 1 May 2024, which replaced the 2016 bankruptcy law; insolvency and restructuring matters are now heard by a dedicated federal Bankruptcy Court established in 2025 (verified against the UAE’s official legislation register, current as at 1 July 2026). It governs corporate insolvency for entities onshore in the UAE. Institutions licensed by the Central Bank sit under a separate regime, and the financial free zones, such as the DIFC, operate their own insolvency laws, which is part of why the VARA framework excludes the DIFC. A firm licensed by VARA and incorporated onshore in Dubai is neither a Central Bank institution nor within a free zone, so when it fails it falls to this federal law.
That is what makes the ring-fence a question rather than a guarantee. An insolvency is a proceeding under the federal statute, run by the court and the appointee it recognises. VARA’s rulebooks bind the firm as a matter of regulation: they require it to segregate client assets, to keep them estate-remote, to exclude them from any sale. But whether those assets in fact sit outside the estate, once the firm is in the proceeding, is decided by the insolvency law and the court applying it, not by the regulatory instruction that they should. A rule stating that client assets are not the firm’s does the necessary groundwork. It does not, on its own, create an interest a bankruptcy court is bound to recognise. This is exactly why the framework’s strongest promises are written to reach only as far as applicable law permits, and why its insolvency rule sends the firm to the appointee under the governing proceedings. The framework was drafted by people who understood where its own authority ends.
What can be said with confidence, then, is the forum and the statute: a VARA firm that fails onshore is wound up under Federal Decree-Law No. 51 of 2023, before the federal Bankruptcy Court. What cannot be said with the same confidence is how that statute, applied to segregated client virtual assets, resolves the estate question in practice. The regime is recent, the asset class is novel, and neither the rulebooks nor the federal statute settles the interaction on its face. So the position is precisely this: the ring-fence is asserted by VARA, bounded by VARA to what federal law allows, and not confirmed by federal law in this setting. That is an open question in the exact sense, one the current materials do not answer, rather than one whose answer is merely hard to find.
Which is why certainty, where a client or a counterparty needs it, comes from how a particular arrangement is structured and from advice on that structure, not from the rulebook’s assertion. And it is why what a client can safely assume differs so sharply by where it sits. Section V sets that out, position by position.
The article’s distinctions now resolve into a set of positions, and the honest account of each has the same shape: a front-end protection the client can rely on, and an estate question no position escapes. Where the positions differ is in how strong the front-end protection is, and how much of the client’s recovery is already contingent before the estate question is even reached.
A custody client holds the strongest position. It can assume that its assets are not the firm’s and never were, that they were segregated in its own wallets and kept under the firm’s control, and that they could not be rehypothecated with or without its consent (Custody Rules III.B.1 to B.4). Of every position in the framework, this is the one with the firmest groundwork for the assets to sit outside the estate, because the rule does not merely ring-fence them, it denies they were ever the firm’s to begin with. What a custody client still cannot assume is that this regulatory status has been confirmed, as a matter of decided law, to bind the federal insolvency (Section IV). It holds the best hand the framework deals; whether that hand is honoured in full is the open question, not a certainty.
A client whose assets are used under the held-on-behalf default, where the Client Agreement has not reversed it, holds a genuine but weaker position. It can assume the framework treats the assets as its own rather than the firm’s (VA Management II.D.3; Lending II.A.5), and that the firm’s wind-down plan must exclude client assets from any sale (Company VII.A.1.k). What it cannot assume is that the assets are sitting intact and reachable. By definition they have been used or moved, and in lending they have been deployed to counterparties who must perform for the client to be made whole (Lending II.A.3 and II.A.6). So even before the estate question, a lending client’s recovery is contingent on third parties. The held-on-behalf default protects the character of the claim, not the availability of the asset.
A client whose Client Agreement has reversed the default holds the weakest position of all, and often without registering it. The assets are no longer held on the client’s behalf, so the client is owed value rather than holding property, and on a failure it stands as a creditor in the federal proceeding, ranking with others under the distribution the insolvency law sets. For this client the estate question barely arises, because it is not asserting that specific assets are its own; it is asserting a claim. The single contractual term examined in the companion article “What can a VASP do with client virtual assets?” is what placed it here, and the failure is where that term does its work.
A holder of a fiat-referenced stablecoin sits on a different but parallel footing. It can assume the reserve regime is built to protect it: backing at full value, reserves segregated and held remote from the issuer’s estate, insulated from the issuer’s creditors, with a redemption right and VARA priority access over the reserves (Annex 1 III.B.1 and III.B.4). What it cannot assume is that the estate-remoteness holds against an insolvency any more firmly than anywhere else, because the framework itself conditions that protection on what applicable law permits (Section III). The stablecoin holder relies on the reserves being there and being reachable, and on the same open question as everyone else.
Across all four, two things hold. The front-end protection is real: the framework segregates, excludes and reserves, and gives VARA power to intervene. And the estate question is unresolved for every position, strongest to weakest alike, because it is answered by the federal insolvency law and not by the rulebook. What this leaves a client able to do is precise and limited. Read the Client Agreement to find which position you are in, because that, and not the name of the service or the fact of a licence, is what sets your exposure. And where the answer matters, have the arrangement structured and advised with the estate question in view, rather than relying on the rulebook to have closed it. Section VI closes the article on what is settled, what is open, and what can be done about it.
What is settled here is everything except the ending. The framework’s protections are real, specific, and on the current rulebooks not in doubt. In custody, client assets are not the firm’s, are segregated and controlled, and cannot be rehypothecated even with consent (Custody Rules III.B.1 to B.4). Where assets are used, the held-on-behalf default keeps them the client’s unless the Client Agreement reverses it (VA Management II.D.3; Lending II.A.5). Every VASP must maintain a wind-down plan that excludes client assets from any sale, including a scheme run by an insolvency appointee (Company VII.A.1.k), and must preserve VARA’s power to step in and take control of client assets (VII.A.1.l). Stablecoin reserves must be fully backed, segregated and held estate-remote (Annex 1 III.B.1 and III.B.4). Each of these was read at source for this article. As front-end architecture, the ring-fence is built.
What is open is whether the architecture holds when the firm actually fails. That question is answered by federal insolvency law, now Federal Decree-Law No. 51 of 2023, before the federal Bankruptcy Court (Section IV), and the framework itself concedes as much: it bounds its strongest promises by what applicable law permits, and routes the insolvency to the appointee under the governing proceedings. The statute and the forum are settled. How that statute treats segregated client virtual assets is not, and the framework does not close the gap because it cannot. This is the one place in the series where the honest answer is that the answer is not yet fixed.
That is not a counsel of despair, and it is worth being plain about what can be done, because the openness sits alongside levers that are real. The Client Agreement is the lever a client controls most directly: through the held-on-behalf term it decides whether the client is an owner or a creditor, and it deserves to be read for that term rather than skimmed as boilerplate. The choice of activity and of counterparty sets the starting position before any agreement is drafted, and custody starts furthest forward. Because whether assets sit outside the estate turns on whether the arrangement creates an interest the insolvency law will recognise, the useful work is done in how the holding is structured, and in advice aimed squarely at that question in the governing law, before a failure rather than during one. None of this converts an open question into a closed one. All of it improves the position a client occupies while the question stays open.
There is also a currency point that belongs to this article in particular. The federal insolvency layer is the moving part. The regime is recent, a dedicated Bankruptcy Court is newly at work, and the treatment of virtual-asset client holdings is the kind of question answered over time, through decided cases and further regulation. An answer that is open on 1 July 2026 may not stay open, in either direction. This is the article in the set most likely to need revisiting, and it should be revisited as the federal position develops.
So the throughline of the piece is also its ending. The framework tells a client what it will do with and for client assets on a failure, and it does so in real and specific terms. The federal insolvency law decides whether that holds. The distance between the two is not a flaw to be papered over with the word “regulated”; it is the precise thing to understand and to plan around. The companion article “What can a VASP do with client virtual assets?” set out what a client holds going into a failure. This article has placed that holding against the law that governs the failure, and has marked, without flinching, where the framework’s authority ends and the open question begins.
For your facts, in confidence, put the question to the firm.
The bench stands behind it