In early 2020, a federal judge in San Francisco issued an unusual order: He compelled DoorDash, the food-delivery company, to submit to the arbitration proceedings it had itself contractually required. Over 5,000 delivery drivers had filed individual arbitration demands – in compliance with DoorDash’s standard terms – and paid over $1 million in filing fees to initiate them. DoorDash, facing administrative fees of nearly $12 million, refused to pay its share, preventing the proceedings from going forward. The American Arbitration Association closed the files, and the drivers went to court to compel DoorDash to proceed with the arbitrations. The judge ruled that, having drafted the arbitration clauses, DoorDash would have to live with them.

The DoorDash episode was not an anomaly but a predictable result of a structural dynamic that we write about in a forthcoming article: When legal systems suppress collective litigation without addressing the conditions that generate collective claims, the pressure for claimants to band together does not disappear — it migrates. The question is not whether that pressure will return, but in what form, and whether institutions will be ready.

The legal framework that produced mass arbitration in the United States was built through a sequence of Supreme Court decisions – Stolt-Nielsen, AT&T Mobility v. Concepcion, and Epic Systems – that narrowed the availability of class actions while strengthening the enforceability of arbitration agreements. The underlying logic was straightforward: If companies could require bilateral arbitration and bar class proceedings, collective litigation would become practically unavailable. Individual claims too small to pursue alone would not be brought. Notably, none of this required any change in arbitration doctrine. The rules stayed fixed; what changed was the scale and coordination of claims brought within them.

Once collective judicial mechanisms were closed off, however, plaintiffs’ lawyers began coordinating thousands of individual arbitration demands against single corporate respondents. Bilateral arbitration, designed to fragment claims, became the means of reaggregating them at scale. Companies such as Uber, Amazon, and DoorDash, having embedded class-action waivers in many consumer and employment agreements, found themselves exposed to front-loaded administrative fees, logistical strain, and settlement pressure that bilateral arbitration systems were not designed to absorb.

The corporate response was telling. Rather than accept the new dynamic, many companies drafted mass arbitration waivers into their agreements or introduced multi-step pre-arbitration filtering mechanisms designed to slow the filing of coordinated claims. More strikingly, some companies moved in the opposite direction – restoring the availability of class action litigation in court, thereby circumventing the dynamics and cost of mass arbitration. Amazon was among those that reportedly amended their terms of service to reintroduce the possibility of class proceedings, the very mechanism their arbitration clauses had been designed to foreclose. The logic had inverted: Having relied on arbitration to limit aggregate exposure, some companies found class litigation preferable to the front-loaded costs and logistical pressures of mass arbitration. Corporate dispute-resolution strategy had come full circle, exposing how unstable any single procedural configuration can be when the underlying economic incentives remain unaddressed.

Crisis-Driven Adaptation and Its Limits

The institutional consequences were substantial. By 2022, mass-arbitration filings reportedly accounted for more than 80 percent of the docket administered by the two largest U.S. arbitration providers, AAA and JAMS. Institutions built for bilateral disputes were forced to improvise under acute strain – revising fee schedules, introducing bellwether processes to resolve representative cases before adjudicating the remaining claims, and redesigning arbitrator-appointment procedures. Those redesigns carried their own legal risks. In Heckman v. Live Nation Entertainment, the Ninth Circuit Court of Appeals affirmed a refusal to compel arbitration under a mass arbitration framework that Live Nation had imposed through Ticketmaster, holding it unconscionable under California law. The court found that the combination of bellwether outcomes that effectively governed other claims – under which decisions in three representative cases became precedent for all remaining claimants, who had no notice of or participation in those proceedings — the absence of discovery, and asymmetric appeal rights created a process that systematically favored the defendant. The case is a reminder that institutional adaptations to mass arbitration may themselves become targets of legal challenge – and that state-law oversight persists even under the strong federal policy favoring arbitration.

The deeper lesson is not procedural but institutional. Crisis-driven adaptation is necessarily reactive, occurring only after the underlying pressures have already materialized, and under conditions that leave little room for deliberate design. The AAA and JAMS did not set out to become ad hoc regulators of mass litigation. They were pushed into that role by the sudden scale of coordinated filings. Their responses may have been sensible as an administrative matter, but they could not resolve the harder questions that mass arbitration exposed: Who may act on behalf of dispersed claimants, under what conditions absent parties may be bound, and how far institutional rulemaking can go without clearer legal authority. Those are questions of legitimacy, not logistics.

Brazil: A Pre-Crisis System

Brazil offers an instructive contrast – not because it has found a better answer, but because it has not yet been forced to find one. But the comparison should not be overstated. Brazil is not currently exposed to a generalized risk of large-scale coordinated arbitration across consumer and employment markets, and the most plausible source of scale-related stress lies in securities and corporate disputes subject to mandatory arbitration.

Arbitration is deeply embedded in Brazil’s capital markets. Companies listed on the Novo Mercado and related governance segments of Brazil’s stock exchange are required to include arbitration clauses in their bylaws, binding listed companies, controlling shareholders, officers, and various investors to arbitrate corporate and securities disputes. Some of the structural conditions that made mass arbitration possible in the United States are already present in Brazil – mandatory arbitration, standardized clauses, and a large, dispersed claimant base – even though Brazil differs in certain institutional respects.

A single market event – a disclosure failure, a governance breakdown, a corporate fraud – could generate thousands of substantially similar claims arising from substantially the same facts, creating strong incentives for coordinated filings. Collective arbitration in Brazil has so far remained rare, but that should not be mistaken for evidence of adequate institutional design. It more likely reflects the absence of triggering conditions. Brazil still lacks a dedicated statutory framework for collective arbitration, and its rules governing collective relief were developed before arbitration became central to capital-markets disputes, leaving a significant legislative gap where aggregation pressures are most likely to emerge.

The Brazilian context also presents a structural feature absent in the U.S.: “extraordinary” representatives – such as public prosecutors and qualified associations empowered to act in defense of collective interests – are not parties to individual arbitration agreements and therefore are generally not bound by them. The impact of this factor is most significant in securities and capital markets disputes, where mandatory arbitration clauses in corporate bylaws already bind a large and dispersed base of investors to resolve disputes in arbitration — precisely where a single market event, such as an alleged disclosure failure or governance breakdown, is most likely to generate arbitration filings and parallel collective judicial proceedings by representatives who were never party to those clauses. The result is not merely a risk of mass arbitration, but a potential for simultaneous arbitration and judicial proceedings over the same underlying events — a coordination challenge that cuts across institutional and legislative domains and that no institutional rule alone can resolve.

The Two-Track Design Challenge

That challenge matters because institutional design and legislative authority address different problems. Arbitration institutions can create procedures for coordinated filings, adjust fee allocation, and adopt sequencing mechanisms to manage scale. What they cannot do is resolve questions of representational authority: who may legitimately aggregate claims on behalf of absent parties, and under what conditions an arbitration award may bind an investor who did not participate in the proceeding. In Brazil, that authority has traditionally been defined by statute and subject to judicial oversight rather than being left to parties’ agreement. That logic does not translate easily into arbitration. The deeper point is that arbitration is not socially or legally authorized in Brazil to perform the same collective functions it was pressed into performing in the United States, and any workable framework must start from that constraint.

Anticipatory design in Brazil must therefore proceed on two tracks. On the institutional track, Brazil’s arbitration bodies are well positioned to develop procedures for handling aggregated claims before crisis conditions force improvisation – and the work has already begun. On the legislative track, clarification is needed on the questions that institutional practice alone cannot settle: who may represent dispersed investors, under what conditions absent parties may be bound by arbitration outcomes, and how collective arbitration relates to Brazil’s existing framework for collective relief. Coordination with the Brazilian Securities and Exchange Commission (CVM) will also be essential. A federal bill points in this direction but leaves unresolved the harder questions any workable framework must address – above all, what standards govern the adequacy of representation and when arbitration awards may bind shareholders who did not participate in the proceeding. A legislative initiative is a necessary step but not, by itself, a solution.

The broader lesson extends well beyond either country. Systems that suppress aggregation rarely eliminate it. They displace it into new procedural forms, at unpredictable scale, under conditions ill-suited to careful institutional response. Brazil is, for now, in a pre-crisis position that resembles the one the United States occupied before mass arbitration arrived, albeit within a different institutional setting. Brazil faces emerging aggregation pressures within a framework designed for bilateral disputes, while retaining time to make deliberate choices. That time will eventually expire. The central design question is not whether collective claims will arise in Brazilian arbitration, but whether the institutions that must handle them will have been built for the task.

Bruno Salama is a lecturer at UC Berkeley School of Law’s Legal Studies Program and a professor of law at FGV Law School in São Paulo, and an arbitrator. Flávio Yarshell is a professor at the University of São Paulo Law School, an attorney, and an arbitrator. This post is based on their recent article, “Design Early or Inherit the Crisis: Mass Arbitration in the United States and Collective Arbitration in Brazil,” forthcoming in the American Review of International Arbitration and available here.