Can a Creditor Wind You Up in Singapore Over a Debt You Agreed to Arbitrate?

What the court decided in September 2026

A fund that had never signed an arbitration agreement was held, on a prima facie basis, to be bound by it, and its applications to put two companies into judicial management were stayed in favour of arbitration in Singapore. The decision is DYX v DYY and another matter [2026] SGHC(I) 19, delivered by the Singapore International Commercial Court on 17 September 2026.

A Cayman Islands fund had lent money to two operating companies in 2020 and 2022. It said the loans had matured and were unpaid. The borrowers said the loans had been released by a waiver and release deed signed in January 2025 as part of splitting a group in two, and that the deed sent any dispute to arbitration under the rules of the Singapore International Arbitration Centre. The fund was not named in that deed and had not signed it.

The sequence matters as much as the law. The fund had already tried to wind the first borrower up in Hong Kong, where the companies are incorporated. That petition was dismissed in November 2025 on the footing that there was a genuine dispute about the debt which should go to arbitration. The fund filed in Singapore on 24 December 2025, weeks later, seeking judicial managers instead.

Three things in the decision are worth an hour of any cross-border counsel's time: what the court refused to do, what it left open, and what the deed did to a fund that never signed it.

The starting point: the stay is mandatory, and it is asked for early

Where the International Arbitration Act applies, a Singapore court has no discretion to keep an arbitrable dispute in court because litigation would be more convenient. The obligation is in section 6 of the International Arbitration Act 1994, and subsection (2) carries the operative words.

Domestic arbitration is different, and the difference is deliberate. Under the Arbitration Act the court keeps a discretion to refuse a stay where there is sufficient reason why the matter should not go to arbitration, a point the Court of Appeal opened its judgment with in CSY v CSZ [2022] SGCA 43 at [1]. If your clause is international, you have the stronger position, and nothing in this article about insolvency changes that.

The timing rule is unforgiving and it is where the right is most often lost. The application must be made after filing and serving a notice of intention to contest or not contest, and before delivering any pleading "(other than a pleading asserting that the court does not have jurisdiction in the proceedings)" or taking any other step in the proceedings: section 6(1). A defence filed to be safe is a step. So is an interlocutory application that invites the court into the merits. The instruction to give your Singapore lawyers on day one is to apply for the stay before doing anything else in the action.

The threshold is a prima facie case, not proof

The party asking for the stay does not have to prove the arbitration agreement binds the other side. It has to show a prima facie case, and the tribunal then decides for itself. The three limbs come from Tomolugen Holdings Ltd v Silica Investors Ltd [2015] SGCA 57 at [63]:

• there is a valid arbitration agreement between the parties to the court proceedings;
• the dispute in the court proceedings, or any part of it, falls within the scope of that agreement; and
• the agreement is not null and void, inoperative or incapable of being performed.

Once those are shown the court stays the proceedings and leaves the jurisdictional question to the tribunal, subject to the court's later supervision: Tomolugen at [64], [66] and [67]. "Dispute" is read widely. A bare assertion of denial is enough, and the court does not ask whether the denial is any good: Tjong Very Sumito v Antig Investments Pte Ltd [2009] SGCA 41 at [49].

That is a low bar by design, and it has a consequence worth stating plainly. A creditor with what looks like an unanswerable debt does not get to have a Singapore judge say so. It gets an arbitration. Whether your clause even works to that extent is a separate question, and Singapore courts will usually save a clause that shows a clear intention to arbitrate.

Saying you never agreed to arbitrate does not raise the bar

A party that denies being a party to the arbitration agreement at all is met with the same prima facie standard, not a higher one. This was the fund's main argument in DYX and it was rejected. It said the question whether a non-signatory is bound is an "anterior issue" which has to be settled first, on the balance of probabilities where it turns on disputed facts, before the stay framework is engaged at all.

The authorities the fund relied on were all decided at the setting-aside stage, after a tribunal had already ruled on its own jurisdiction, which is a different procedural moment and a different standard: [71] to [73]. The same argument had already been rejected in Malini Ventura v Knight Capital Pte Ltd [2015] SGHC 225 at [36], where a party said her signature on the guarantee was a forgery and the court still applied the prima facie test, because anything higher would impose too heavy a burden on the party seeking to enforce the arbitration agreement.

Read practically, denying the agreement buys you nothing at the court door. It buys you an argument before the tribunal, and then a full review by the seat court afterwards if the tribunal gets it wrong.

Why a winding-up application cannot get around the clause

A creditor cannot improve its position by reframing a disputed debt as a winding-up application, and the Court of Appeal has now said so three times. The reasoning is that the standard of review would otherwise depend on nothing more than the creditor's choice of paper. In AnAn Group (Singapore) Pte Ltd v VTB Bank [2020] SGCA 33 at [63] the court said there was no principled basis for differing standards on what is essentially the same disputed debt, and that the alternative would leave the outcome turning on the creditor's "arbitrary or tactical choice".

But the mechanism is not section 6, and this is the point most summaries get wrong. A winding-up application is not stayed under the Act. The court applies the same prima facie test by analogy and then exercises its own discretion about what to do, which is ordinarily to dismiss. The Court of Appeal put it beyond doubt in Singapore Commodities Group Co, Pte Ltd v Founder Group (Hong Kong) Ltd (in liquidation) [2026] SGCA 24, decided on 8 May 2026, at [60] and [62]: the approach "aligned with" the section 6 stay, and dismissal or, exceptionally, a stay "was ultimately an exercise of discretion by the court, and not a case of the court ordering an automatic or mandatory stay or dismissal".

What the loss of standing means, and for how long

The creditor does not lose its debt. It loses its standing to present the application, and only until the arbitration decides the dispute. The distinction is easy to overstate in the creditor's favour and easy to overstate in the debtor's, so it is worth the Court of Appeal's own words.

Two qualifications travel with that. The displacement of the court's power to decide the debt for itself operates "as a general rule" rather than absolutely: [61]. And the whole of it is subject to the abuse of process safeguard, dealt with next. A debtor reading this as immunity is reading it wrongly, and so is a creditor reading it as the end of the debt.

There is also a middle course that creditors underuse. Where the creditor can show real concerns about the debtor's solvency as a going concern and no triable issue on the debt, the court can stay the application rather than dismiss it, and give liberty to restore it if the debtor turns out to have no genuine desire to arbitrate and is taking active steps to stifle the arbitration: AnAn at [111]. That was the shape of the order in DYX, which stayed the applications on conditions, kept them open and required the parties to report the outcome of the arbitration to the Registry: [143] and [144].

The safety valve, and why it is not a second look at the merits

The court will refuse to dismiss where the debtor is abusing the process, most obviously by disputing a debt it has already admitted, but the threshold is high and it is not a route back to the merits. Singapore Commodities sets out a two-stage inquiry at [85]: first, whether there was a clear and unequivocal admission of both liability and quantum; second, whether there is a clear and convincing reason for resiling from it.

The warning attached to it is the important part for anyone tempted to use it as a workaround. At [80] the Court said it would be incongruous to reject any inquiry into the merits under the prima facie standard "only to engage in a back-door review of the merits under the guise of considering if there has been an abuse of process". A defence that is merely weak is not an abuse. The creditor in that case took the point, lost it, and was ordered to pay costs on an indemnity basis.

Judicial management: the part that is not settled

No appellate court has decided whether the same approach governs a judicial management application, and the leading statement the other way is expressly an aside. This is the open question and it is worth being precise about how open it is, because it is often described as a settled difference and it is not.

In Gulf International Holding Pte Ltd v Delta Offshore Energy Pte Ltd [2023] SGHC 151 the General Division said the principle should not apply as strictly to judicial management, and that the court should instead make "a more holistic assessment of the facts and consider, inter alia, the interests of the other stakeholders of the debtor company and the wider public interest": [69]. The reasons were that judicial management is a rescue regime engaging a broader public interest, that an applicant must additionally satisfy the court that a statutory purpose is likely to be achieved under section 91(1)(b) of the Insolvency, Restructuring and Dissolution Act 2018, which leaves less room for a tactical filing, and that the consequences are less severe, an order running 180 days unless the court says otherwise.

But the court said all of that while saying it did not need to.

The case was decided on other grounds. The court found the debt had been admitted, found the debtor had acted in abuse of process by issuing shares that cut the claimant's security from 75 per cent of the company to less than 0.0015 per cent, and said at [70] that it would have dismissed the stay applications in any event. So the judicial management reasoning is considered obiter of a first-instance court, not a holding.

In DYX the Singapore International Commercial Court noticed the difficulty and went only as far as the sources allowed, saying at [87] that Singapore Commodities "raises some doubt" as to whether "the holistic approach" in Gulf International remains appropriate in the judicial management context. It then held that the holistic factors were absent anyway: the applications were a "ricochet response" to the dismissed Hong Kong petition at [90], and the creditors said to need protecting were aligned funds under the same control at [91] to [93]. The question was raised and not answered.

Do the two lines actually conflict? A reading, and its limits

There is a way of reading these two decisions on which they do not conflict at all. It is available on the text, it is the reading we would advance, and no court has yet adopted it. We set it out with its weaknesses because the honest position on an unsettled question is more useful to a reader than a confident one.

The reading is that the two cases are about two different discretions. Singapore Commodities displaces the court's discretion to decide the disputed debt itself as part of its general civil jurisdiction: [61]. It expressly preserves the court's discretion in its response, which is why dismissal is ordinary rather than automatic: [62] and [63]. What Gulf International was addressing is the response: its holistic assessment is directed to the court's "discretion to stay or dismiss the application", at [64] and [69]. On that reading the Court of Appeal took away a power Gulf International was never claiming.

The difficulty is that the only court to have held the two side by side did not read them that way. In DYX the Singapore International Commercial Court put them in tension at [84] to [87] and said the later decision raises some doubt about the earlier one. If the two-discretions reading were plainly right, no doubt would arise. So the competing reading has to be stated: that the holistic approach simply does not survive Singapore Commodities, which appears to be where the SICC's own instinct pointed.

There is also a limit on how far the reconciliation gets you, and it is about standing rather than about merits. Singapore Commodities frames the whole doctrine as a question of standing: a prima facie arbitrable dispute leaves the creditor without standing as a creditor until the arbitration resolves it, at [64]. Sections 90 and 91(1) of the IRDA give standing to "any creditor (including any contingent or prospective creditor)". If that analysis transposes to judicial management, a creditor whose only qualifying debt is the disputed one has a standing problem even where the company's insolvency can be proved from other evidence entirely, because independent insolvency answers the separate requirement in section 91(1)(a) and does nothing for standing. Gulf International did not have to face that, because the debt there was found to have been admitted. Whether the standing analysis transposes at all is itself part of what is open.

One further loose end will be argued. The stay in DYX was granted under section 6 of the International Arbitration Act itself: [139]. The respondents asked for it on that basis, or alternatively as a case management stay: [56]. Nobody contested whether section 6 reaches a judicial management application, as distinct from the discretionary route the winding-up cases use, and the court did not analyse it. An assumption nobody contested is not authority for the point. The same judgment also reached for the abuse of process safeguard inside a section 6 application, saying at [94] that the stay might be defeated to the extent that the AnAn framework is engaged, although section 6(2) contains no such limb. Two provisions point the other way, towards the discretionary route: section 91(7) of the IRDA lets the court make any order it thinks fit on hearing the application, and section 91(10)(a) lets it make an order where the public interest requires one.

For a business planning around this, the practical upshot does not depend on which reading wins. Assume that a Singapore court will not decide your disputed debt for you, and that whether the rescue procedure gives a creditor a way around that is currently unpredictable.

Singapore has taken a side, and London and Kuala Lumpur took the other

The Singapore position is now a deliberate divergence from the Privy Council and from Malaysia, which matters if your group litigates the same debt in more than one place. Singapore Commodities records at [65] to [68] that the Privy Council in Sian Participation Corpn v Halimeda International Ltd [2025] AC 1321 and the Federal Court of Malaysia in V Medical Services M Sdn Bhd v Swissray Asia Healthcare Co Ltd [2025] 2 MLJ 744 have both rejected this approach, preferring to ask whether the debt is genuinely disputed on substantial grounds.

Founder Group (CA) is Founder Group (Hong Kong) Ltd v Singapore JHC Co Pte Ltd [2023] SGCA 40, the decision that came between AnAn and this one. Two practical readings of that sentence. The first is that the Singapore position is settled and will be applied. The second is that the Court of Appeal has said, in terms, that it is open to a properly mounted challenge, which is an unusual thing for a final court to volunteer and tells you where the argument will go next.

If you are running a cross-border group with an arbitration clause in the intercompany or supply documents, that table is a planning tool. The same disputed debt gives the creditor a real insolvency threat in one place and almost none in another.

How a fund that signed nothing was held prima facie bound

The fund in DYX lost the threshold argument on the definitions in a deed it never signed itself, not on any doctrine about groups of companies. This is the part of the decision with the widest reach, because it is a drafting problem rather than an insolvency one.

The deed defined "subsidiary" not by shareholding but by control, as a person over whom another "has or obtains direct or indirect control of the power to direct or cause the direction of the management and decision-making of that person without reference to any other person". The releasing parties were defined to include their direct and indirect subsidiaries. The fund's own operating agreement vested management and control of the fund exclusively in its manager, and the manager was a signatory: [99] to [110]. Separately, that operating agreement authorised the manager to enter into any contract, which was broad enough on its face to reach the deed: [111] to [113].

A definition keyed to control, not to shareholding, reached a separately managed fund that had signed nothing.

Nobody needed to argue a single economic entity or to pierce anything. The definitions did the work. If your group signs a settlement deed, a release or a separation agreement containing a wide definition of affiliates or subsidiaries keyed to control, every managed vehicle in the structure may be inside it, along with the arbitration clause, and the people with the economic interest may never be told. The same discipline applies to the positive case: a well-drafted shareholders' agreement does only what its definitions let it do.

The deed in DYX had a second and more basic fault. On the borrowers’ reading it swept up the claims of one group against the other, said in evidence to exceed 100 million dollars, of which the fund’s own loans were only a small fraction, and it never mentioned those loans at all. The court described the parties as having failed to set out "the most basic settlement thesis of all", being whether all claims were released or some survived, at [28]. Drafts were exchanged in which one side's lawyer tried to confirm in writing that fund-level claims were excluded, and was rebuffed; the deed was signed anyway: [35] to [37]. The litigation that followed ran across two jurisdictions and more than a year.

What this changes in your documents and your first move

Most of the value here is captured before any dispute, in three lines of drafting and one instruction on the day proceedings are served. The law described above is largely favourable to a party with an international arbitration clause. What loses the protection is procedural delay, loose definitions and an unconsidered admission.

Two neighbouring questions are answered elsewhere on this site. Where the other side is drifting towards the courts rather than the tribunal, the question is how to hold them to the arbitration agreement. Where an award already exists and the fight is about reopening it, that is a question of finality rather than of power, and the source of a tribunal's authority to make interim orders in the first place turns on the law of the seat rather than the place of enforcement.

Where this sits in our practice: we act on international arbitration for companies trading between Europe, Australia and Asia, and the Singapore side of that work runs through our Singapore desk.

A closing note on how much weight to put on the September 2026 decision. It is a first-instance judgment of the Singapore International Commercial Court, it carries the standard notice that it remains subject to editorial correction, and the time for any appeal was still running when this article was written. It is good evidence of how the question will be approached. It is not the last word on the judicial management point, and it does not pretend to be.

This article is provided for general information purposes only and does not constitute legal advice. Specialised legal counsel should be sought for specific fact patterns.