In Anand Khosla v. Punam Kumari Singh, the Bombay High Court quietly retreats from the Supreme Court’s settled position on the arbitrability of IP disputes. Analysing the decision, Aditya Bhargava argues that it misapplies the rights in rem/personam distinction while considering the arbitrability of IP disputes under Vidya Drolia, and improperly legitimises the splitting of causes of action. Aditya is a fourth year law student at the National Law School of India University, Bangalore. [Long post ahead]
The Unresolved Code: Anand Khosla, Arbitrability, and the Ghost of Rights in Rem
By Aditya Bhargava
The relationship between arbitration and intellectual property law in India continues to be unsettled. Over the last decade, the Supreme Court (“SC”) has steadily expanded the scope of what disputes can be resolved through arbitration. Yet, at the level of the High Courts, this progress is not always uniformly reflected (read here, here, and here). Some decisions fall back on a narrow and formalistic approach that risks undermining the SC’s work. The recent judgment of the Bombay High Court (“BHC”) in Anand Khosla & Anr. v. Punam Kumari Singh is a clear example of this trend. The dispute arose between partners of an LLP. The arbitral tribunal declined to decide the question of ownership of software-related IP, and the BHC upheld the award, holding that the refusal was rightly made.
In this post, I argue that Anand Khosla marks a quiet but significant departure from Vidya Drolia v. Durga Trading Corporation. By validating the refusal of an arbitral tribunal to decide on the ownership of a partnership asset, the BHC has misapplied the distinction between rights in rem and rights in personam, resulting in a departure from the mandate that the SC explicitly established. Further, the decision runs contrary to the principle against splitting causes of action that was established in Booz Allen & Hamilton Inc. v. SBI Home Finance Ltd.
The Facts
The dispute originated from a Limited Liability Partnership Agreement between the petitioners, Anand Khosla and Universal Test Solutions LLP, and the respondent, Punam Kumari Singh. They established the LLP to develop and market a software product named “Test Magic.” A separate limited liability company was incorporated in the UAE to market the products. Khosla invested approximately Rs. 3.05 Crores in the business. Internal conflicts eventually emerged. This resulted in Punam Singh’s expulsion from the LLP in April 2016.
Punam Singh invoked the arbitration clause in the LLP Agreement following her expulsion. The dispute went before a Sole Arbitrator who delivered an award in 2019. The Arbitrator upheld her expulsion and rejected monetary claims against Khosla and the LLP.
A crucial aspect of the arbitral award concerned the treatment of the software itself. The Tribunal declined to decide the question of ownership of IP rights in the “Test Magic” software. To understand the mistake of the Tribunal, it is necessary to examine its reasoning closely. The Tribunal held that deciding the ownership of the software would “necessarily imply” deciding the ownership of the copyright itself. It reasoned that since the work on the software commenced prior to the LLP Agreement, questions regarding the author’s agreement and the nature of ownership were linked to copyright. The Tribunal further noted that adjudicating a case of “simultaneous use” or “copying” of the software would also “enter into the realm of ownership of copyrights,” thereby conflating factual usage with legal status.
The petitioners’ counsel had conceded that while the arbitrator might not have jurisdiction to decide the validity of the copyright, they could decide the ownership of the software itself as an asset. However, the Tribunal failed to act on this distinction. The Court’s judgment reveals a conflation between the “mark” (trademark) and the software code (copyright). In ¶33, Justice Marne observes that Punam Singh claimed ownership in the “mark of the said software” and concluded that rights arising out of the “mark” would be rights in rem.
Anand Khosla and the LLP subsequently challenged the arbitral award under Section 34 of the A&C Act, 1996. In ¶¶36–37, the Court observed that the Arbitral Tribunal had rightly not adjudicated the issue of ownership of IP rights in the software, had not travelled outside the contract, and that the award was unexceptionable.
The Vidya Drolia Misstep: Conflating Validity with Ownership
The core mistake in Anand Khosla lies in the Court’s acceptance of the Tribunal’s refusal to decide the question of ownership. By stating that the issue was “rightly not adjudicated,” the Court implies that the Tribunal correctly held it lacked jurisdiction to decide the matter (¶ 33). This must be distinguished from a finding that a claim failed for lack of evidence. Had the Tribunal examined the material on record and concluded that ownership was not proved, it would still have adjudicated the dispute, even if the claim was ultimately rejected. By endorsing the view that the issue was “rightly not adjudicated,” the Court effectively places the question of software ownership outside the Tribunal’s authority altogether. However, I argue that the question of ownership of “TestMagic” was indeed arbitrable.
In Drolia, the SC laid down a clear four-fold test for determining when a dispute is non-arbitrable (¶76). A dispute is non-arbitrable only if it meets one of these conditions.
(1) It must relate to actions in rem that do not involve subordinate rights in personam arising from rights in rem.
(2) It must affect third-party rights or have erga omnes effect.
(3) It must concern inalienable sovereign functions.
(4) Or it must be expressly declared non-arbitrable by statute.
By validating the tribunal’s reasoning, the BHC appears to have treated the ownership dispute over the software as falling within the first category—an action in rem. This is a doctrinal error. Vidya Drolia itself draws a careful distinction between a right in rem and a subordinate right in personam. In ¶43, Justice Sanjiv Khanna clarifies that a judgment in rem determines the status of a thing itself, rather than merely adjudicating the interests of the parties before the court.
Applied to Anand Khosla, the distinction becomes clear. The validity of copyright in the software ‘Test Magic’ is undoubtedly a right in rem. However, the question of who owns that copyright between two partners in an LLP is a subordinate right in personam. It concerns only the inter se rights and obligations of the parties (see Prof. Bhasheer’s and Inika Charles’ posts).
This regression is particularly stark when contrasted with the BHC’s own prior jurisprudence. In Eros International Media Ltd. v. Telemax Links India Pvt. Ltd., Justice Patel clarified that where a plaintiff alleges a breach of copyright arising from a contract, the action is in personam, and expressly rejected the view that the mere existence of a statutory right renders a dispute non-arbitrable (Read here). Similarly, in EuroKids International Pvt. Ltd. v. Bhaskar Vidyapeeth Shikshan Sanstha, the BHC held that a dispute under a franchise agreement involving IP rights was arbitrable, as it was confined to the contractual relationship between the parties.
In the present case, the dispute before the Tribunal was not about whether Test Magic was a copyrightable work under the Copyright Act, 1957, nor about the subsistence or validity of copyright. The issue was whether, under the LLP Agreement, the valid copyright vested in Partner A or Partner B. That is a straightforward contractual inquiry. By refusing to decide it, the Tribunal failed to exercise the jurisdiction flowing directly from the arbitration agreement.
In Vidya Drolia, particularly ¶38, the Supreme Court cautioned against an overly broad use of the in rem exception, observing that every right in rem has a corresponding right in personam. The relevant inquiry is whether deciding the dispute requires the exercise of a sovereign function, such as altering a public register. Determining which partner owns a software asset requires no such exercise. It only involves examining the partnership deed and the circumstances in which the code was written.
The Booz Allen Paradox: Sanctioning Bifurcation
The second major infirmity in Anand Khosla is its violation of the anti-bifurcation principle laid down in Booz Allen. In Booz Allen, the SC considered the arbitrability of mortgage suits and, while holding that a suit for sale of mortgaged property is an action in rem and therefore non-arbitrable, laid down a critical procedural rule in ¶29:
“If bifurcation of the subject matter of a suit was contemplated, the legislature would have used appropriate language to permit such a course. Since there is no such indication in the language, it follows that bifurcation of the subject matter of an action brought before a judicial authority is not allowed.”
The SC clearly held that a cause of action cannot be split. You cannot arbitrate part of a dispute and litigate the rest. The dispute must be resolved in one forum.
In Anand Khosla, the dispute between the partners was comprehensive, encompassing expulsion, monetary accounts, and ownership of the partnership’s central asset. While the Tribunal did note a lack of pleadings and evidence regarding the specific ownership of the software (¶155 of the Award), the Court did not uphold the award solely on this evidentiary basis. Instead, it expressly endorsed the Tribunal’s refusal to adjudicate ownership on jurisdictional grounds, holding in ¶35 that rights arising from ownership of the mark “Test Magic” were rights in rem and therefore non-arbitrable. Ironically, the High Court relied on Booz Allen (¶35) to reach this conclusion—using a precedent that prohibits bifurcation to justify it. By sustaining an award that resolved expulsion while “rightly” declining to decide ownership on jurisdictional grounds, the BHC has effectively sanctioned the very bifurcation that Booz Allen disallows.
Hypothetically, consider the position of the parties post-judgment. The expulsion is valid and the accounts are settled, but ownership of Test Magic remains uncertain. The petitioners must now approach a civil court to seek a declaration of ownership. That court will necessarily re-examine the same LLP Agreement, the same facts concerning the software’s creation, and the same partnership history already considered by the Arbitrator.
This creates precisely the “possibility of conflicting judgments” that Booz Allen (¶25) identified as a core rationale for non-arbitrability. If the civil or commercial court were to find that the expulsion was motivated by an attempt to misappropriate the software, such a finding on ownership would undermine the factual and normative basis of the Arbitrator’s conclusion on expulsion.
If the HC believed the IP dispute to be non-arbitrable, it ought to have set aside the reference entirely or held that the dispute as a whole, being inextricably intertwined with the IP, was non-arbitrable. It cannot logically hold that the Tribunal was competent to decide the status of the partnership but incompetent to decide the partnership assets.
The Sovereign Function Fallacy
The judgment implicitly relies on the notion that IP ownership is a “sovereign function” under the third prong of the Vidya Drolia test. This view stems from a misunderstanding of how IP rights function in a commercial context.
In Vidya Drolia (¶40), the SC noted that “sovereign functions” are those like taxation or criminal justice, which are inalienable. The grant of a patent is a sovereign act. The registration of a trademark is a sovereign act. However, the transfer of these rights is a commercial act.
Section 34 allows a court to set aside an award if it conflicts with the “public policy of India.” The BHC in Anand Khosla (¶37) found no conflict with public policy. Yet, there is a strong argument that refusing to resolve the core commercial issue between parties is itself against public policy.
The DHC, for example, has consistently affirmed this distinction. In Golden Tobie Pvt. Ltd. v. Golden Tobacco Ltd., the Court held that a dispute regarding the termination of a trademark license agreement was arbitrable because it did not involve the sovereign act of registration, but the commercial act of licensing (Rounak’s Post). Similarly, in Hero Electric Vehicles Pvt. Ltd. v. Lectro E-Mobility Pvt. Ltd., the Court correctly identified that where the dispute is about the terms of usage of a trademark, and not the registration of the mark itself, the dispute is arbitrable (here and here).
Anand Khosla ignores this distinction. The ownership of software in an LLP is governed by Section 17 of the Copyright Act (first owner) and the LLP Act. Neither statute confers exclusive jurisdiction on a civil court to decide ownership disputes between partners. There is no “special statute” here that excludes arbitration, unlike the DRT Act in Booz Allen or the Rent Control Acts discussed in Vidya Drolia.
Implications and Conclusion
The Anand Khosla decision is a step backward that encourages parties to manufacture reasons to avoid arbitration. By allowing respondents to argue that tribunals cannot decide IP ownership, the ruling forces disputes into slow civil courts. This harms the startup ecosystem, where Founders Agreements rely heavily on IP. If an arbitrator cannot determine who owns the technology when founders split, the arbitration clause becomes useless, defeating the very purpose of choosing alternative dispute resolution.
The High Court has created a paradox where one can arbitrate a contract breach but not the ownership of the asset within it. Instead of clarifying the law, the Court supported the tribunal’s refusal to act. This contradicts the SC’s vision in Vidya Drolia of arbitration as a “one-stop” solution. The ruling suggests that whenever intangible assets are involved, arbitration is merely a prelude to inevitable civil litigation rather than a final resolution.
The best scenario would have been for the High Court to set aside the specific finding on non-arbitrability and remand the matter or clarify the legal position while upholding the result on evidentiary grounds.
* I would like to thank Daanish Naithani and Kartik Sharma for their valuable comments and suggestions.
