Joint Venture Agreements in India: Deadlock, Exit Mechanisms and What Happens When Partners Fall Out

The anatomy of a deadlock, and why 50:50 is both fair and dangerous

Most JV conflicts are structural before they are personal. A 50:50 venture, the most instinctively fair structure, is also the only one that guarantees deadlock is possible: equal boards, equal votes, and a disagreement on anything fundamental stops the company. Unequal ventures, 74:26, 60:40, avoid formal deadlock but recreate it contractually, because the minority’s protection is a reserved matters list, the veto architecture we detailed in our SHA clause guide, and a veto exercised is a deadlock by another name. So the drafting starts with honesty about which decisions can stall the venture: reserve genuinely structural matters, budgets beyond thresholds, charter changes, new capital, exits, and leave operations to management, because a JV where routine decisions need both parents’ signatures has built its own gridlock. Define deadlock precisely: a reserved matter proposed and rejected, or a board tie, persisting across two meetings held not less than a defined interval apart, because a loose definition lets an impatient partner manufacture deadlock as an exit trigger, a move Indian JV disputes have seen more than once.

The escalation ladder: most deadlocks are actually solvable

Good agreements resolve most deadlocks without touching the dramatic clauses. The standard ladder: referral upward, the disputed matter goes from the JV board to the chairpersons or managing directors of the parent companies, people with relationship capital and settlement authority, within a fixed window. Then structured mediation, a neutral facilitator, often within an institutional framework, with a short clock. Then, for suitable subjects, expert determination, pricing disputes to an investment banker, technical disputes to an engineer, binding on the parties. Genuine strategic impasses, though, do not yield to process, and for those the agreement needs an ending.

The buy sell clauses: roulette, shootout, and their gentler cousins

The dramatic machinery, names included, because the names are how the market discusses them. Russian roulette: either party may serve a notice naming a price per share; the recipient must choose, sell all its shares at that price, or buy the notifier’s shares at the same price. The elegance is self policing honesty, name a low price and you will be bought out cheap, name a high one and you must fund it. Texas shootout: both parties submit sealed bids to buy the other out; the higher bid wins and must complete. Dutch auction variants and multi round versions exist for taste. Two Indian law realities temper the drama. First, these mechanisms assume both parties can write a cheque at short notice; between a deep pocketed multinational and an Indian promoter of ordinary means, roulette is not neutral, it is a squeeze dressed as symmetry, and fair drafting adjusts, longer funding windows, financing conditions, or floor valuations by an independent valuer. Second, where a party is non resident, FEMA’s pricing guidelines constrain the transfer price, a non resident buying from a resident, or selling to one, must transact within prescribed valuation bounds, so the clause should expressly subordinate its price mechanics to regulatory pricing, or the beautiful machinery produces an unlawful transfer. The gentler cousins matter more in practice: put and call options, the minority may put its stake to the majority, or the majority may call the minority’s stake, at defined times and formula prices, now clearly enforceable in the SHA context and long since normalised under FEMA’s option framework with its pricing conditions; staged exits, rights to sell down in tranches; and, as the last resort the agreement should say aloud, winding up the venture by agreed dissolution, because sometimes the honest answer is that the marriage is over and the assets should go home.

Exit, survival, and the divorce inventory

Beyond deadlock triggered exits, the agreement should answer the questions every JV divorce actually fights over. The brand and the IP: who owns what the venture built, who licenses what back, and what happens to jointly developed technology, the chapter that outlives everything. Non compete during and after: scoped carefully against Section 27 of the Contract Act’s restraint of trade rule, in a share sale context Indian law tolerates reasonable covenants better than in employment, but breadth kills them. People: which seconded employees return, which stay. Transition: supply, services, and customer handover for a defined runway, because ventures embed into their parents’ operations and unplugging takes a year. And the deed of adherence discipline for any new shareholder, so the machinery binds everyone holding shares on the day it fires. One more Indian essential, the theme of our companion article on articles of association: every clause above that is meant to bind the JV company itself, transfer restrictions, vetoes, buy sell mechanics, must be mirrored into the company’s articles, or enforcement risks dissolving into a contract claim against a partner while the company lawfully acts as if the SHA did not exist. In JV disputes, where the company is the battlefield, this is not a technicality; it is the war.

Arbitration: where JV fights actually go

Indian JV agreements almost universally choose arbitration, institutional rules, a neutral seat, and, in cross border ventures, often a foreign seat with Indian assets enforcement in mind. Draft it knowing the pattern of JV warfare: interim reliefs matter most, blocking a board meeting, restraining a share transfer, preserving accounts, so preserve access to courts for interim measures under Section 9 of the Arbitration Act alongside the arbitration clause, and align the arbitration’s scope with the oppression and mismanagement jurisdiction of the NCLT, which no private clause can fully oust, a parallel track sophisticated parties game out in advance.

Can AI help with JV agreements?

At the modelling and vigilance layers, yes. AI tools can simulate the buy sell mechanics before signature, at what price does roulette favour the richer partner, what does the put option yield across valuation scenarios, and stress test the deadlock definition against manufactured impasse tactics. In a live venture, AI can track reserved matter approvals, flag decisions drifting past their consent requirements, and keep the SHA against AOA diff current as the venture amends documents. In a dispute, it accelerates the archaeology, years of board minutes and correspondence summarised into a chronology, that JV litigation runs on. The boundaries are those of every adversarial domain: enforcement questions, FEMA pricing interactions, NCLT strategy, and the negotiation itself are practitioner territory, and machine confidence about clause enforceability in Indian courts deserves verification every time. Model with AI; decide, draft finally, and fight with qualified humans.

When to Review This

• A joint venture agreement earns its fee in two chapters: deadlock, defined precisely, escalated through a ladder, and resolved by calibrated buy sell machinery that respects FEMA and funding asymmetries; and exit, options, staged sell downs, and a divorce inventory covering IP, people, and transition. Indian enforcement adds two constants: mirror the machinery into the articles, and design the arbitration around interim relief and the NCLT’s parallel jurisdiction. Draft the unhappy chapters as if they will be used, and they probably will not need to be, which is the paradox, and the point.