Arbitration in India was conceived as an escape route from the procedural quagmire of ordinary civil courts. Yet, three decades after the enactment of the Arbitration and Conciliation Act, 1996 (‘the Act’), it remains caught in a persistent paradox: a dispute-resolution mechanism designed to operate with substantial independence from the State is still tethered to judicial supervision. Nowhere is this tension more apparent than at the threshold, in determining whether an arbitration agreement exists, who is bound by it, and how the arbitral tribunal is constituted. For most of this period, Indian arbitration jurisprudence has been engaged in an institutional tug-of-war. On one side lies party autonomy: the foundational freedom of commercial actors to design their own dispute-resolution mechanisms. On the other stands judicial and constitutional supervision, justified by concerns of fairness, equality, independence, impartiality, and statutory legality.
Recent Supreme Court decisions are often presented as progressive corrections of earlier excesses. Yet they also expose a deeper structural problem: Indian arbitration law has repeatedly moved through cycles of judicial intervention, statutory correction and larger-bench reconsideration. The resulting uncertainty is itself a commercial cost. The tension is not simply between courts and parties. It is between three forms of autonomy and control: the parties’ autonomy to agree to arbitration; the tribunal’s autonomy to determine its own jurisdiction and adjudicate the dispute; and the judiciary’s responsibility to police the outer boundaries of consent, legality and procedural fairness. The controversy does not stem from judicial intervention itself but from the Court’s inability to fix a predictable boundary where such intervention should end.
The statutory entry point into arbitration is Section 7 of the Act. An arbitration agreement is not merely another boilerplate clause; it represents the parties’ decision to substitute private adjudication for ordinary civil litigation. That makes certainty particularly important at the point of formation. Yet Section 7 does not create an exclusively signature-based regime.
Section 7(3) requires an arbitration agreement to be in writing. Section 7(4), however, adopts a deliberately functional approach to what constitutes writing. The agreement may be contained in a document signed by the parties, in an exchange of letters, telegrams or other means of telecommunication, including electronic communications, or in an exchange of statements of claim and defence in which the existence of the agreement is alleged by one party and not denied by the other. While an oral agreement may survive in general contract law, it cannot engage the Act’s statutory machinery without a written record satisfying Section 7.
The second requirement is substantive rather than merely documentary: the parties must demonstrate a present intention to submit disputes to arbitration. In Jagdish Chander v. Ramesh Chander (2007), the Supreme Court distinguished between language that creates a present and binding obligation to arbitrate and language that merely contemplates arbitration as a possible future arrangement. The use of words such as “may” is not automatically fatal; the question is whether, read in context, the clause evidences an unequivocal commitment to arbitrate. The principle is important because arbitration rests on consent. A court should not transform a negotiation mechanism or an agreement to agree into a binding arbitration clause merely because arbitration appears somewhere in the contractual architecture.
Section 7(5) permits an arbitration agreement to arise through incorporation by reference. The case M.R. Engineers and Contractors Pvt. Ltd. v. Som Datt Builders Ltd. (2009) established that a mere general reference to another contract will ordinarily not suffice to incorporate its arbitration clause. But the proposition should not be stated as an absolute requirement of a specific reference in every case. Subsequent Supreme Court jurisprudence has recognised circumstances in which standard-form terms, particularly those routinely incorporated into commercial contracts, may effectively import an arbitration clause. The inquiry turns on objective intention and effective incorporation, not whether the agreement mechanically repeated the words “arbitration clause”.
The doctrine of separability provides another important qualification to formal contractual analysis. Section 16(1) treats an arbitration clause as an agreement independent of the other terms. Consequently, a finding that the underlying contract is void does not, by itself, invalidate the arbitration agreement. This is critical to the functioning of arbitration. Otherwise, virtually every allegation that the underlying contract was void, rescinded, repudiated or otherwise discharged could be deployed to prevent the tribunal from deciding whether that very allegation is correct.
Separability, however, should not be overstated. It does not manufacture an arbitration agreement where none was ever formed, nor does it cure a defect directed specifically at the arbitration agreement itself.
The more consequential development concerns not what constitutes an arbitration agreement, but who constitutes a party to it. Section 7 requires an arbitration agreement to be in writing, but it does not require every person who may ultimately be bound by the agreement to have physically signed the document. The principle that a non-signatory may be bound received major constitutional treatment in Cox and Kings Ltd. v. SAP India Pvt. Ltd. (2023), where the five-judge Constitution Bench considered the Group of Companies doctrine. The Court rejected an approach founded merely on the idea that affiliated companies constitute a “single economic entity”. Instead, the focus must remain on consent and the parties’ intention, which may, in appropriate cases, be discerned from conduct, the relationship between the parties, the composite nature of the transaction and other surrounding circumstances.
Recently, that reasoning has been applied in KKH Finvest Pvt. Ltd. v. Ashiesh Shukla (2026). Notably, KKH Finvest was decided by a two-judge Bench of the Supreme Court, which considered whether a non-signatory was nevertheless a “veritable party” to the arbitration agreement contained in a broader Memorandum of Settlement. The Court concluded that the surrounding transactional circumstances, including his role and involvement in implementing the settlement, justified treating him as a party to the arbitration arrangement.
Complex transactions are rarely contained within a single instrument executed by a neatly defined set of counterparties. Acquisitions, restructurings, settlements and infrastructure projects frequently operate through a network of interdependent agreements. However, commercial pragmatism comes with a cost. The danger is not that every non-signatory will automatically be bound. The danger is that the inquiry into consent becomes increasingly evidentiary. A corporate affiliate’s participation in negotiations, performance or implementation may be relevant evidence of consent, but such participation is not synonymous with consent to arbitrate.
For multinational investors and corporate groups that deliberately maintain separate legal personalities, the expansion of the non-signatory doctrine creates a difficult boundary question: when does commercial involvement transform into legal consent to arbitrate?
That question becomes acute at the Section 11 stage. In Cox & Kings, the Constitution Bench recognised that whether a signatory seeks to rope in an unwilling non-signatory, or a non-signatory itself steps forward to invoke the clause, the referral court’s role is strictly confined to a prima facie determination of whether the entity qualifies as a “veritable party.” Because resolving that question requires evaluating complex commercial matrices and evidentiary conduct, the Constitution Bench cautioned referral courts against conducting mini-trials. True fidelity to the doctrine of Kompetenz-Kompetenz requires deferring that inquiry to the arbitral tribunal under Section 16, which must accord the non-signatory a full hearing under the principles of natural justice before asserting jurisdiction over it.
The doctrine stretches the traditional boundaries of consent while leaving its formal requirement intact by leaving it to the arbitral tribunal, rather than the referral court, to do the heavy evidentiary lifting.
Once an arbitration agreement has been established, the tribunal must be constituted. Here too, the statutory architecture begins with party autonomy and uses judicial appointment as a fallback where the agreed machinery fails.
Section 10 permits parties to determine the number of arbitrators, subject to the prohibition on an even number. In the absence of agreement, the default is a sole arbitrator.
Section 11 then provides mechanisms for judicial intervention where the agreed appointment procedure breaks down. Sections 11(4) and 11(5) prescribe specific 30-day periods for appointments in the circumstances covered by those provisions; Section 11(6) addresses failure of an agreed appointment procedure more generally. The architecture is therefore not one in which courts constitute every tribunal. In fact, judicial appointment is intended to operate as a fallback where the parties’ chosen machinery has failed.
The 2015 Amendment to the Act introduced Section 11(6A), directing the Supreme Court or High Court, while considering applications under Section 11, to confine its examination to the existence of an arbitration agreement. The provision was intended to prevent Section 11 from becoming a preliminary trial of the dispute. The statutory position subsequently became complicated because the 2019 Amendment Act purported to omit Section 11(6A), but that omission was never brought into force. The Supreme Court has accordingly continued to apply Section 11(6A).
In SBI General Insurance Co. Ltd. v. Krish Spinning (2024), the Court reaffirmed that the Section 11 inquiry is extremely narrow and directed strictly to a prima facie determination of the existence of the arbitration agreement. Contested questions of fact and mixed questions of law and fact ordinarily belong before the arbitral tribunal under the doctrine of Kompetenz-Kompetenz.
Earlier decisions had consistently expanded the scope of judicial intervention under Section 11. In NTPC Ltd. v. SPML Infra Ltd. (2023), the Court devised the "eye of the needle" test to scrutinise accord and satisfaction, while in Arif Azim Co. Ltd. v. Aptech Ltd.(2024), it authorised referral benches to weed out "deadwood" claims barred by limitation. Krish Spinning substantially dismantled this creeping jurisdiction. Overruling the application of the "eye of the needle" standard to accord and satisfaction, and expressly clarifying Arif Azim to bar referral courts from conducting evidentiary inquiries into time-barred claims, the Court confined the judicial inquiry on limitation strictly to whether the Section 11 petition itself was filed within three years of default. The significance of Krish Spinning lies in restoring an institutional division of adjudication: courts verify the prima facie existence of an arbitration agreement; tribunals decide everything that lies beyond it.
Under the rules of institutional arbitration centres such as the Mumbai Centre for International Arbitration, the Singapore International Arbitration Centre, and the International Chamber of Commerce, the arbitral institution's registrar or the appointing authority handles default appointments, shielding parties from court backlogs.
Notably, special statutes bypass party autonomy altogether; for example, under Section 18(3) of the Micro, Small and Medium Enterprises Development Act, 2006, statutory reference to the Facilitation Council completely extinguishes private arbitration clauses, demonstrating how easily legislative paternalism overrides commercial agreements.
The development of Indian arbitration law can be understood as a series of interventions designed to cure one institutional problem, sometimes producing another. Three episodes illustrate the pattern particularly well, viz., unilateral appointments, the 'veritable party' doctrine, and the stamping saga.
In Central Organisation for Railway Electrification v. ECI-SPIC-SMO-MCML (JV) (2024), a five-judge Constitution Bench confronted appointment mechanisms under which public-sector entities retained substantial control over the pool from which arbitrators could be selected.
The Court overruled the earlier 2019 CORE decision and held that the relevant unilateral appointment mechanisms were incompatible with the requirements of independence and impartiality. It relied, among other things, upon the equality principle embodied in Section 18 of the Act. The result was significant. Party autonomy could not be used to justify an appointment mechanism that effectively permitted one side to control the constitution of the tribunal.
However, the temporal effect of the judgment matters. The Court prospectively overruled the earlier position rather than simply invalidating every tribunal previously constituted under it. That distinction is crucial to the broader argument. The CORE saga demonstrates not merely that judicial intervention can be necessary, but that doctrinal instability has consequences of its own. A legal system that first permits a procedure, then validates it, and subsequently declares it impermissible, inevitably creates uncertainty for parties who structured their contracts in reliance upon the earlier rule.
There is, moreover, a continuing doctrinal debate about the route by which the Court reached its conclusion. Post-retirement, Justice Rohinton F. Nariman has questioned whether Section 18 was the appropriate doctrinal foundation for addressing appointment-stage equality, even while supporting the broader concern with unilateral appointment mechanisms. The deeper lesson is that a correct result does not necessarily eliminate questions about the institutional path taken to reach it.
The Court’s conclusion (per Cox & Kings, and KKH Finvest) that a non-signatory may qualify as a “veritable party” introduces a degree of uncertainty into what was traditionally regarded as a threshold question. An arbitration agreement is supposed to be consensual. If consent can be inferred from performance and transactional involvement, the dividing line between participating in a transaction and consenting to arbitration becomes critical.
The challenge for future cases is therefore not whether non-signatories can ever be bound. That proposition is now well established. The challenge is ensuring that the doctrine remains anchored in demonstrable intention rather than becoming a mechanism for imposing arbitration merely because a corporate entity was commercially involved.
The stamping saga provides perhaps the clearest example of judicial correction generating its own institutional disruption. The Supreme Court’s seven-judge Constitution Bench in In Re: Interplay between Arbitration Agreements under the Arbitration and Conciliation Act, 1996 and the Indian Stamp Act, 1899 (2023) overruled the position adopted in SMS Tea Estates Pvt. Ltd. vs Chandmari Tea Co. Pvt Ltd. (2011) and NN Global Mercantile Pvt. Ltd. vs Indo Unique Flame Ltd. (2023). The Court held that defects relating to non-stamping or insufficient stamping of the underlying instrument were curable and did not render the arbitration agreement itself void. The judgment restored an important pro-arbitration principle: fiscal defects in the underlying instrument should not ordinarily become a jurisdictional barrier to the commencement of arbitration.
But the episode remains instructive. The movement from SMS Tea Estates to NN Global and finally to the seven-judge Interplay decision illustrates how questions at the intersection of arbitration and other statutory regimes can rapidly become existential questions for the arbitral process. It would be an overstatement to say that stamping brought Section 11 proceedings to an “absolute halt” across India. But the jurisprudential uncertainty undeniably generated delay, additional litigation, and commercial disruption before the larger Bench restored the arbitration-friendly position. The pattern is difficult to ignore: a threshold doctrine is first expanded, then produces systemic friction, and is eventually recalibrated by a larger Bench.
Conclusion
True arbitral autonomy does not mean the absence of judicial supervision. An arbitration regime cannot function credibly if parties can unilaterally manipulate tribunal composition, if non-existent arbitration agreements can be manufactured through expansive inference, or if statutory requirements can simply be ignored. The real objective is more precise: judicial intervention should be principled, proportionate and confined to the institutional function courts are uniquely equipped to perform.
Each of these decisions exposed a genuine institutional vulnerability. CORE protected parity against departmental bias; the Interplay reference rescued commercial intent from fiscal technicalities; Cox and Kings and KKH Finvest adjusted the law to the reality of composite M&A deals; and Krish Spinning finally pushed threshold courts out of evidentiary mini-trials.
India’s ambition to become a credible global arbitration hub cannot rest merely on having arbitration-friendly outcomes. It requires predictability in the rules by which those outcomes are reached.
The recurring cycle of smaller-bench innovation, systemic uncertainty, and larger-bench correction carries a cost that cannot be measured solely in reported judgments. Commercial parties draft contracts against a legal background. When that background shifts repeatedly, contractual certainty becomes provisional.
The solution lies not in banishing the court, but in enforcing a strict gatekeeping boundary. At the Section 11 threshold, the court’s inquiry should remain narrow and expeditious, principally directed towards the existence and formal validity of the arbitration agreement. Where difficult questions of fact, merits, limitation or substantive jurisdiction arise, they should ordinarily be left to the arbitral tribunal under Section 16. Where the issue concerns a non-signatory, the court may necessarily have to undertake a prima facie inquiry into whether that person is a party to the arbitration agreement, but that inquiry should remain anchored in consent rather than corporate association or commercial involvement alone. Arbitration thrives neither when courts abandon it nor when courts govern it. It thrives when courts know precisely where their role begins, and, equally important, where it ends.