[By Utkarsh Trivedi]
The author is pursuing LL.M in IADR at the National University of Singapore.
Introduction
On 24 August 2026, the Singapore International Commercial Court (“SICC”) dismissed Tata Power’s application to set aside a SIAC award of nearly US$500 million in favour of Kleros Capital Partners (“Kleros”), a BVI vehicle that claimed to have lost the chance to develop a coal deposit in Kamchatka, Russia, as a result of Tata Power’s conduct, as enumerated in The Tata Power Company Limited v Kleros Capital Partners Limited, [2026] SGHC(I) 15. With costs, interest and an earlier costs award, Tata Power’s liability now approaches US$640 million. The coram of Justice S Mohan and International Judges Anthony Besanko and Anthony Meagher rejected every ground Tata Power advanced, namely breach of natural justice in respect of causation, remoteness and mitigation, and apparent bias on the part of the tribunal majority. This piece argues that the judgment matters less for what it restates about Singapore’s narrow standard of curial review than for what it reveals by comparison with India: the substantive tests applied in the two jurisdictions are closely aligned, and the divergence in finality lies instead in the number of procedural opportunities each system affords a losing party to reopen an award.
The dispute and the challenge in brief
In 2013, Kleros approached Tata Power as a prospective co-investor in an undeveloped coal deposit on the Kamchatka peninsula, and the parties entered into two non-disclosure agreements (the “NDAs”) that bound Tata Power to confidentiality and prohibited it from circumventing Kleros’s economic interests. After the relationship broke down over who would lead the bid for the mining licence, Tata Power acquired the licence at public auction in 2017 and later abandoned the project as unviable. Kleros commenced arbitration proceedings that were subsequently consolidated before a single SIAC tribunal.
In 2023, the tribunal (Lawrence Boo, Stuart Isaacs KC and former Indian Supreme Court judge AK Ganguli) found that Tata Power had misused confidential information, made false statements and concealed its intention to supplant Kleros, findings Tata Power did not challenge. In the 2024 quantum award, the majority valued the project at over US$1 billion, assessed Kleros’s lost chance at 60% and awarded US$490 million; the dissenting arbitrator would have awarded roughly US$13.5 million, reflecting only the negotiating leverage Kleros had lost.
Before the SICC, Tata Power contended that the majority had failed to deal with causation, remoteness and mitigation in the manner the parties had agreed, in breach of natural justice. The SICC rejected each ground. It held that the causation complaint had shifted so substantially between the originating application and later submissions as to constitute a new, time-barred ground that was in any event without merit; that the tribunal was entitled to treat causation, remoteness and foreseeability as overlapping inquiries; and that the majority had twice rejected Tata Power’s mitigation case with reasons. The SICC also dismissed the bias allegations, which rested on undisclosed appointments of Mr Boo and Mr Isaacs in matters involving Kleros’s funder, Omni Bridgeway, and its counsel, Squire Patton Boggs, and on Mr Boo’s links with Benjamin Hughes, then chair of Omni Bridgeway’s investment committee. In closing, the coram described Tata Power as an unduly sensitive or suspicious litigant that had abandoned several of its original complaints and timed its challenge to land shortly before the quantum award.
Analysis: process review and the limits of relabelling
The judgment rests on the settled distinction between review of the merits, which is excluded, and review of the arbitral process, which is all that a set-aside application under Section 24(b) of the International Arbitration Act 1994 (the “IAA”) and Article 34(2) of the UNCITRAL Model Law permits. The governing framework remains Soh Beng Tee & Co Pte Ltd v Fairmount Development Pte Ltd [2007] 3 SLR(R) 86 at [29], which requires an applicant to identify the rule of natural justice breached, show how it was breached, and establish that the breach was connected to the making of the award and prejudiced its rights. Where the complaint is that the tribunal failed to consider an issue, the High Court of Singapore in Lee Boon Khim v Entech Global Solutions Pte Ltd [2026] SGHC 141 at [22] held that the inference must be clear and virtually inescapable before a court will conclude that the tribunal failed to apply its mind to an essential issue, as opposed to every argument raised. Consistent with that approach, the SICC treated the award’s reasoning as sufficient if discernible when read as a whole, with appropriate latitude for imprecise drafting.
On bias, the SICC first applied the time limit for arbitrator challenges in Rule 15.1 of the SIAC Rules 2016, which requires a notice of challenge within 14 days after the circumstances giving rise to the challenge became known, or should reasonably have become known, to the challenging party, and treated that limit as presumptively fatal before reaching the substance. On the merits, the court drew on the IBA Guidelines on Conflicts of Interest in International Arbitration and required more than professional overlap or repeat institutional affiliation. Non-repeat appointments in unrelated matters did not give rise to a reasonable suspicion of bias, nor did indirect social ties to a funder-affiliated individual absent evidence of communication about the case.
The case also confirms that Singapore courts will look past the label a party places on its complaint. Framing a substantive disagreement with a tribunal’s assessment of loss of chance, causation or mitigation as a procedural failure does not convert it into one. The analytical significance lies in what follows from this. By absorbing the remoteness complaint into a finding that the tribunal had engaged with overlapping issues, the SICC placed the question whether the tribunal applied the correct test beyond the reach of Section 24(b) of the IAA. An error in applying a legal test, however consequential to quantum, remains an error within jurisdiction and is not reviewable. That is a deliberate allocation of risk: parties who choose Singapore as the seat accept that a tribunal may err on the law in exchange for an award that cannot be relitigated in the guise of a process complaint. It is against this allocation that the Indian position should be measured.
Same test, different doors: the contrast with India
First, the Singapore process offers a single further tier. Tata Power’s only remaining avenue is an appeal to the Court of Appeal, after which there is no further recourse: Singapore has no analogue to the special leave petition under Article 136 of the Constitution of India or to the curative petition recognised in Rupa Ashok Hurra v Ashok Hurra (2002) 4 SCC 388. If the Court of Appeal declines to disturb the SICC’s characterisation of what counts as engagement with an issue, the liability of approximately US$640 million will stand.
Second, and more significantly, the substantive tests are closely aligned. Singapore’s standard under Section 24(b) of the IAA sits close to the natural justice and public policy review articulated by the Supreme Court of India in Associate Builders v Delhi Development Authority (2015) 3 SCC 49 and, following the 2015 amendments, in Ssangyong Engineering & Construction Co Ltd v National Highways Authority of India (2019) 15 SCC 131. The divergence lies in the additional ground available alongside it. For awards in domestic arbitrations, Section 34(2A) of the Arbitration and Conciliation Act 1996 (the “1996 Act”) permits a court to set aside an award vitiated by patent illegality appearing on its face. Although that provision excludes the mere erroneous application of law and any reappreciation of evidence, it still invites scrutiny of whether the tribunal’s treatment of an issue was perverse, rather than merely whether the tribunal addressed it. Singapore has no equivalent. A Tata Power-style complaint that the tribunal misapplied the remoteness test, rather than ignored it, would find no foothold under Section 24(b), but could plausibly be framed as patent illegality before an Indian court in a domestic arbitration.
The consequences of that additional ground are illustrated by Delhi Metro Rail Corporation Ltd v Delhi Airport Metro Express Pvt Ltd 2024 INSC 292 (“DMRC”). A 2017 award in favour of the concessionaire was upheld by a single judge of the Delhi High Court under Section 34 of the 1996 Act, set aside by a Division Bench in an appeal under Section 37, restored by the Supreme Court in 2021 on a special leave petition, and finally set aside in 2024 when the Supreme Court allowed a curative petition on the ground of patent illegality. The same award thus passed through four rounds of review, with substantially the same standard nominally applied throughout. DMRC is relevant to this comparison not because its facts resemble those of Tata Power, but because it shows that a narrow substantive test does not of itself secure finality where the procedural architecture allows a merits complaint to be re-presented at successive tiers under a broader label.
Conclusion
Tata Power and DMRC articulate the standard of curial review in almost identical terms: both courts disclaim any appellate function over the award and confine review to a narrow, process-based inquiry. Yet DMRC required four rounds of review, concluding more than a decade after the dispute arose, before the Supreme Court’s curative jurisdiction brought it to an end. This piece contends that the Singaporean approach should be preferred, not because its doctrinal test is more exacting, but because its procedural architecture gives practical effect to that test. India’s difficulty is structural rather than doctrinal: a losing party may proceed through a Section 34 application, a Section 37 appeal, a special leave petition and, exceptionally, a curative petition, and each tier offers a further opportunity to recast a merits complaint as patent illegality. If India seeks the degree of finality the SICC secured in Tata Power, amending the language of Section 34 of the 1996 Act will not suffice; what is required is a reduction in the opportunities to relitigate the same award, whether by curtailing successive appeals in arbitration matters or by confining the curative jurisdiction to the exceptional cases for which it was designed. The trade-off is real, since fewer tiers mean that a tribunal’s error of law will more often stand uncorrected. That, however, is the bargain Singapore has made explicit, and one that parties accept when they select it as the seat.