Author: Devanshu Sajlan*
| Jurisdiction: | Topics:
|
The Supreme Court of India’s recent judgment in Gayatri Balasamy v. ISG Novasoft Technologies Ltd., 2025 INSC 605, marks a significant development in Indian arbitration law, particularly on the scope of judicial intervention under Section 34 of the Arbitration and Conciliation Act, 1996. While the judgment affirms that courts may, in exceptional cases, “modify” an arbitral award, it grounds this power in the doctrine of severability. Yet a closer reading reveals a subtle but important tension in the judgment’s analytical architecture—between its initial embrace of a narrow legal standard and its eventual pivot to a more open-ended structural test.
The majority opinion begins with a commitment to a narrow, legally disciplined standard for judicial modification—one grounded in comparative authority and the principle of non-appellate review. Yet, as it develops its operative test, the Court quietly pivots to a more open-ended structural inquiry based on the separability of valid and invalid portions of the award. This conceptual shift, from a framework based on legal errors to one rooted in structural interdependence, introduces ambiguity that may inadvertently loosen the very limits the Court initially sets out to reinforce.
This blog post argues that the severability doctrine in Gayatri Balasamy needs further legislative or doctrinal tightening to prevent merit-based intrusions through the back door.
- Context: From Setting Aside to Modification
Under Section 34 of the 1996 Act, courts are empowered to set aside arbitral awards on limited, enumerated grounds. Traditionally, Indian courts have been reluctant to exercise any power of “modification” akin to the broad corrective powers seen in U.S. or U.K. arbitration law. In Project Director, NHAI v. M. Hakeem ([2021] 5 S.C.R. 368), the Supreme Court of India drew a bright line: courts may only set aside or remand, not modify, arbitral awards under Section 34 (¶ 40).
Gayatri Balasamy, however, tempers Hakeem by holding that in limited cases, particularly those involving severable excesses of jurisdiction under Section 34(2)(a)(iv), courts may indeed modify or excise the invalid portion while upholding the remainder. In so doing, the Court introduces a narrow remedial flexibility while reaffirming the principle of arbitral finality.
- Mustill and Boyd: A Legally Disciplined Starting Point
The majority begins by invoking Lord Mustill and Stewart Boyd’s classic text on commercial arbitration. The Court quotes their observation that variation of an award by a court is justified only when “the modification follows inevitably from the tribunal’s determination of a legal question.” (¶ 46).
This citation is significant for two reasons.
First, it underscores that judicial modification is permissible not as an appellate correction but as a logical extension of a legal conclusion already reached by the tribunal. If a tribunal makes a legal determination and draws a conclusion from it, but then proceeds to grant relief that exceeds its jurisdiction or misapplies its own legal reasoning, courts may step in, not to rewrite the merits, but to realign the award with the law the tribunal itself applied.
Second, this framework draws a clear boundary: courts may engage with the award only at the level of legal error, not factual reconsideration. By endorsing this comparative principle, the Court aligns itself with a widely accepted international standard—judicial minimalism premised on respect for arbitral autonomy.
III. The Operative Test: Structural Severability
However, this clarity begins to blur when the Court articulates the actual test for when judicial modification may be exercised.
The Court states that an award may be modified or partially set aside if the valid and invalid components are “legally and practically separable.” (¶35, 45). This severability standard is not new and has been used before by different High Courts to modify an award (for example, see R.S. Jiwani (M/s), Mumbai v Ircon International Ltd [2009] 2 MhLJ 758).
The Supreme Court also appends an important clarification: “validity” in this context does not mean correctness on the merits. Rather, it refers to validity under the proviso to Section 34(2)(a)(iv) (See Majority Opinion, footnote 35).
This is a useful doctrinal move. It ensures that “severability” does not become a pretext for judicial second-guessing of arbitral reasoning. The Court explicitly affirms that “validity” must not be read as a license to review the merits. So far, so good.
But the ambiguity creeps in elsewhere.
- From Legal Error to Structural Interdependence: A Conceptual Shift
Unlike Mustill and Boyd’s standard, which ties modification strictly to legal reasoning and its inescapable consequences, the severability test hinges on a different concept: interdependence (¶35). Courts must ask whether the valid and invalid parts of the award are “intrinsically intertwined”—a term that, while useful in theory, lacks a clear boundary in practice.
What does “intrinsic interdependence” mean? The court gives an alternative phrase – “legally and practically inseparable”. (¶35, 45). However, the majority does not clarify whether courts must determine interdependence solely by looking at the face of the award or whether they may peek into the tribunal’s reasoning, weigh its inferences, or assess the evidentiary record to determine if severance is feasible. This opens the door to interpretive discretion that sits uneasily with the Court’s stated commitment to non-appellate review.
Here lies the central tension. The Court begins with a strict commitment to legal error-based correction, tethered to the Mustill-Boyd principle that judicial action must flow “inevitably” from a legal conclusion. But it ends by adopting a severability standard that lacks the same discipline, allowing courts to assess whether parts of the award are practically separable.
- The Dissent’s Warning: Backdoor Merits Review
Justice Viswanathan’s dissent draws a clear doctrinal boundary between the permissible use of severability for setting aside and its impermissible extension to modification. He expressly rejects the majority’s view that severability can justify judicial rewriting of an award under Section 34, maintaining instead that the two are conceptually and legally distinct: “modification” involves changing, varying, or qualifying an award, while “severance” merely entails separating or disjoining parts of the award (¶8).
He affirms that severability is available to courts only for partially setting aside an award (¶153). This view builds on the test laid down by the Delhi High Court in NHAI v. Trichy Thanjavur Expressway Ltd. (2023 SCC OnLine Del 5183), which he explicitly endorses (¶152-153). That test requires the offending portion to be based on distinct facts and obligations, such that its removal does not have a “cascading impact” on the rest of the award (¶152–153). Where claims are intertwined, partial annulment would be impermissible (¶152).
Conversely, Justice Viswanathan rejects modification as falling outside the remit of Section 34. He notes that “unless expressly authorized by law,” a court “cannot modify or vary the award” because doing so “will be tantamount to exercising the power of merits review” (¶89). This position is further elaborated where he emphasizes that the “power to modify” involves the court “entering the arena of adjudicating the dispute on merits,” which is antithetical to the parties’ choice to resolve their disputes through a non-judicial mechanism (¶93).
Accordingly, the dissenting judgment stands for the proposition that severability must remain a subtractive tool—used to excise defective portions that are self-contained—but never a means to reconstruct or revise the award through judicial substitution.
- Illustrative Example: The Severability Dilemma
Consider a tribunal that awards damages under three heads: (1) delay in supply, (2) defect in quality, and (3) wrongful termination. Suppose the court finds that the wrongful termination claim was beyond the scope of the arbitration clause, but the other two were not. Can the court simply delete the damages under (3) and let the rest stand?
The Gayatri Balasamy test says the court must assess whether the heads of damages are legally and practically separable. But this inquiry is not confined to a legal threshold. It requires the court to examine whether the findings on delay and defect are “intrinsically intertwined” with those on wrongful termination. That, in turn, may involve reviewing how the tribunal reasoned through causation, quantum, and the interdependence of claims.
In effect, severability under this test can pull the court into the factual and evidentiary matrix of the award—not just the jurisdictional contours. The court may need to assess whether damages were allocated distinctly or awarded as a package, and whether one head influenced the assessment of another. This is no longer a simple question of whether the tribunal exceeded its mandate—it becomes an inquiry into how the tribunal exercised that mandate.
This blurs the line between legal review and merits review. While the test aspires to preserve judicial restraint, its practical application risks courts revisiting the substance of the award under the guise of severability.
VII. Severability or Surgery? Why Courts Can Cut but Not Stitch Arbitral Awards
The issue becomes even sharper when we distinguish between setting aside and modifying an arbitral award. Both the majority and the dissent in Gayatri Balasamy agree that severability can be used to partially set aside an award—such as by striking down the damages for wrongful termination in the example above. But the dissent draws a clear line: severability cannot be used to modify the award, especially when doing so requires the court to intrude into the tribunal’s reasoning.
Suppose instead of setting aside the damages under (3), a party asks the court to reduce them—for instance, arguing that even if wrongful termination exceeded the tribunal’s jurisdiction, some compensation was still deserved. To entertain that, the court would have to examine how the tribunal calculated damages under (3), whether it contemplated a fallback amount, and whether reducing the figure would disrupt the balance of the award—perhaps the amount under (3) was given in lieu of expanding heads (1) or (2), or factored into an overall compromise.
This moves the court from simply cutting an invalid portion to reconstructing the tribunal’s intent, effectively rewriting the award. That crosses into a merits-based review, something Indian courts have long cautioned against. The dissent’s position reflects this concern: while severability is a valid tool for excising what’s invalid, it cannot be stretched into a license for judicial surgery.
VIII. Toward a Coherent Doctrine: Legislative or Doctrinal Fixes
To resolve this ambiguity, a few options present themselves:
- Statutory Clarification: Parliament could amend Section 34 to explicitly define the limits of factual inquiry in severability determinations.
- Doctrinal Refinement: Future benches of the Supreme Court could narrow the scope of severability for modification by holding that only those defects that are (a) facially evident and (b) severable without interpretive or factual analysis may qualify for judicial modification.
Absent such clarifications, the risk remains that severability will become a new frontier for judicial overreach, precisely what Gayatri Balasamy set out to avoid.
Conclusion: A Judgment Worth Celebrating—With Caution
Gayatri Balasamy is a landmark for what it affirms: that arbitral finality does not foreclose all judicial correction, and that Section 34 can accommodate a limited power of modification where the award exceeds the scope of reference. It draws a careful boundary between setting aside and surgical correction, and in doing so, offers a middle path between Hakeem’s rigidity and the more permissive precedents of the past.
Yet, it also embodies an unresolved tension—between the Court’s initial reliance on legal logic and its eventual turn to structural severability. This slippage, while subtle, is significant. It suggests that the test for judicial modification, unless more sharply defined, may blur the line between legal error correction and merits review.
That line is not merely academic. It marks the boundary between a court that respects arbitral autonomy and one that risks undermining it. Gayatri Balasamy deserves to be read with that caution in mind.
* Devanshu Sajlan is a Judge in the Delhi Judicial Services. A gold medalist from the National Law University, Delhi, he began his career at AZB & Partners, one of India’s leading law firms, and later worked with Mr. Neeraj Kishan Kaul, Senior Advocate, before joining the judiciary. He was awarded the prestigious Fulbright-Nehru Master’s Fellowship to pursue an LL.M. at Columbia Law School, where he graduated with High Honors (awarded to the top ten percent of the class). His research interests include international arbitration, white collar crimes, and the rights of marginalised communities.