On Petition for Review from the Court of Appeals for the Second District of Texas Argued September 18, 2019 Justice Guzman delivered the opinion of the Court. The Federal Arbitration Act embodies a “liberal federal policy favoring arbitration agreements,”[1] but because arbitration is “a matter of consent, not coercion,”[2] parties cannot be compelled to arbitrate any dispute absent an agreement to do so.[3] In this arbitration case, the ultimate issue is whether the parties agreed to arbitrate class-action claims, but the threshold issue is whether a court or arbitrator is empowered to make that determination. The trial court declined to compel arbitration of class claims under the parties’ arbitration agreement, and the court of appeals affirmed.[4] We hold that (1) arbitrability of class claims is a “gateway” issue for the court unless the arbitration agreement “clearly and unmistakably” expresses a contrary intent; (2) “[a] contract that is silent on a matter cannot speak to that matter with unmistakable clarity”;[5] and (3) an agreement to arbitrate class claims cannot be inferred from silence or ambiguity—an express contractual basis is required.[6] The lower courts correctly applied these principles in declining to compel class arbitration. We affirm. I. Background This arbitration dispute between homeowners and their home-warranty company began as an individual action for construction-defect damages and evolved into a putative class action complaining about “deliberately overbroad” releases the warranty company allegedly “demanded” before making covered repairs. Only the class claims are at issue in this appeal. The homeowners, Nathan and Misti Robinson, purchased a newly constructed residential home that was enrolled in a limited warranty program operated by Home Owners Management Enterprises, Inc. d/b/a HOME of Texas, and Warranty Underwriters Insurance (collectively, HOME). When construction-related defects were discovered, the Robinsons sued HOME and other defendants alleging the defects were not promptly or properly resolved. Over the Robinsons’ vigorous opposition, the trial court abated the case and compelled arbitration in accordance with the terms of the limited warranty and its addendum. The limited warranty requires “mandatory binding arbitration of Unresolved Warranty Issues” and provides that “[t]his binding arbitration is governed by the procedures of the Federal Arbitration Act [FAA].”[7] The addendum requires the same: “All Unresolved Warranty Issues will be submitted to binding arbitration . . . [and] [t]his binding arbitration is governed by the procedures of the Federal Arbitration Act.” Both contracts broadly define “Unresolved Warranty Issue” using identical language: All requests for warranty performance, demands, disputes, controversies and differences that may arise between the parties to this [Limited Warranty or Addendum] that cannot be resolved among the parties. An Unresolved Warranty Issue may be a disagreement regarding: the coverages in this [Limited Warranty or Addendum]; an action performed or to be performed by any party pursuant to this [Limited Warranty or Addendum]; [or] the cost to repair or replace any item covered by this Limited Warranty [or Addendum]. The addendum further defines “Unresolved Warranty Issue” as including “any other complaint or controversy regarding this TRCC Addendum between the parties to the Addendum.” Notably, neither the limited warranty nor the addendum mentions delegation of arbitrability questions. Nor does either contract reference class arbitration. And though the American Arbitration Association (AAA) has promulgated rules pertaining to both matters, neither the limited warranty nor the addendum references or incorporates the AAA rules or any others.[8] Yet, with less than a month before the scheduled arbitration, the Robinsons filed an amended statement of claims seeking to add class-action claims against HOME to the arbitration proceeding. The new—and entirely independent—claims alleged that HOME routinely demanded overbroad releases as a precondition to fulfilling its warranty obligations. HOME promptly filed written objections to the amended statement and moved to strike the class claims from the arbitration proceeding. HOME objected that the putative class claims were “beyond the scope” of the order referring the case to arbitration, were untimely under that order, and were untimely under the arbitrator’s scheduling order. HOME’s motion also addressed the merits of class certification, arguing a class could not properly be certified under the rules of civil procedure. The following week, mere days before the arbitration began, the arbitrator denied HOME’s objections and motion to strike “in its entirety,” but bifurcated the class claims from the Robinsons’ construction-defect claims. After arbitration on the Robinsons’ individual claims had concluded, but before the arbitrator had issued a decision, HOME asked the trial court to clarify the “scope of the issues” referred to the arbitrator and, in the alternative, to strike the Robinsons’ class claims. While HOME’s motion was pending in the trial court, the arbitrator ruled against HOME on the warranty claims and awarded the Robinsons substantial damages, costs, and fees.[9] Further, and in accordance with the arbitration agreement’s terms, the arbitrator awarded HOME the costs and fees it had incurred compelling arbitration over the Robinsons’ resistance.[10] With the arbitrator’s award in hand, the Robinsons returned to the trial court to file a ” Statement of Claims, Individually and as the Representatives of All Persons Similarly Situated.” Once again, the Robinsons’ putative class action alleged HOME refused to pay for home repairs unless the homeowners executed overbroad releases. But this time, the Robinsons did not resist arbitration; they demanded it, asserting HOME was required to arbitrate the class claims under the broad arbitration provisions in the limited warranty and addendum. HOME responded with a motion to dismiss, disputing that the arbitration agreement authorized class arbitration and arguing that only the court, not the arbitrator, could make that determination. The trial court ruled in HOME’s favor, concluding that: The question of whether the parties agreed to class arbitration is a question of arbitrability for [the court]. The Parties did not “clearly and unmistakably” provide that the arbitrator is to decide issues of arbitrability; thus, [the court] shall determine the issue of class arbitrability. The Court determines and finds that the Warranty Agreement between the Parties does not permit class arbitration.[11] On interlocutory appeal,[12] the court of appeals affirmed, holding the trial court applied the correct legal standards and did not abuse its discretion in refusing to compel arbitration.[13] Affirming the trial court’s conclusion that availability of class arbitration is a gateway issue for the court, the appeals court declined to follow our In re Wood decision, which held that an arbitrator should rule on class certification issues when an arbitration agreement is governed by the FAA and the parties have agreed to submit all disputes arising out of the agreement to the arbitrator.[14] After surveying federal authority issued after Wood, the court concluded that Wood “was based entirely on presumably binding [Supreme Court] authority that has since proved to be a chimera” and thus was not binding authority because it “did not really ‘squarely decide’ the issue.”[15] The court observed that post-Wood federal authority effectively abrogated the legal premise on which Wood was based. Deciding the “who decides” issue anew, the court explained that (1) the United States Supreme Court has clarified that the matter Wood took to be settled law is actually an open question; (2) every federal circuit court to consider the matter in light of the Supreme Court’s clarification has held that class arbitration presents a gateway issue presumptively for determination by the court; and (3) a “bilateral arbitration agreement [that] says nothing about delegating the question of class-arbitration availability to an arbitrator” is not a “clear and unmistakable delegation” to the arbitrator.[16] “Having concluded the trial court properly assumed the role of ‘decider,’” the court then examined the arbitration agreement’s text to determine whether the agreement evinced the parties’ plain intent that class-based claims would be arbitrated.[17] Adhering to the Supreme Court’s mandate that a party “‘may not be compelled under the FAA to submit to class arbitration unless there is a contractual basis for concluding that the party agreed to do so,’” the appeals court found no agreement to arbitrate class claims because the agreement contained “not one word about class arbitration.”[18] The Robinsons’ petition for review presents three issues: (1) who decides whether parties have agreed to class arbitration; (2) whether the Robinsons and HOME agreed to class arbitration; and (3) whether HOME otherwise consented or acquiesced to arbitration of the class claims. Though an order denying an arbitration demand is reviewed for abuse of discretion,[19] the issues on appeal present only questions of law, which are subject to de novo review.[20] II. Discussion Arbitration is “simply a matter of contract between the parties; it is a way to resolve those disputes—but only those disputes—that the parties have agreed to submit to arbitration.”[21] Arbitration is thus governed by two fundamental principles: arbitration agreements are contracts that must be enforced according to their terms, and a party cannot be compelled to arbitrate any dispute absent an agreement to do so.[22] Although arbitration is favored under both state and federal law, “arbitrators wield only the authority they are [contractually] given.”[23] So, to ensure parties are not forced to arbitrate matters without their agreement, a substantive question of arbitrability—i.e., whether the parties have actually agreed to submit a particular dispute to arbitration—”‘is an issue for judicial determination unless the parties clearly and unmistakably provide otherwise.’”[24] The phrase “question of arbitrability” refers to the “narrow circumstance where contracting parties would likely have expected a court to have decided the gateway matter.”[25] Such circumstances are limited to (1) whether the parties have a valid arbitration agreement at all and (2) whether an arbitration clause in a concededly binding contract applies to a particular type of controversy.[26]“[R]eference of [a] gateway dispute to the court avoids the risk of forcing parties to arbitrate a matter they may well not have agreed to arbitrate.”[27] Gateway arbitrability issues are distinct from procedural or subsidiary questions that grow out of an arbitrable dispute and are presumptively for an arbitrator to decide.[28] Examples include fulfillment of prerequisites to arbitration; limitations, notice, laches, estoppel, and the like; and waiver of limitations periods, claims, or defenses.[29] Subsidiary issues present “questions for the arbitrator not only because the ‘parties would likely expect that an arbitrator would decide [them],’ but also because the questions do not present any legal challenge to the arbitrator’s underlying power.”[30] Whether an arbitrator or the court has the primary authority to determine a disputed issue is consequential, not only from a contractual-expectations standpoint, but also because appellate review of an arbitrator’s decisions is significantly more deferential than review of a court’s decisions. If the parties have agreed to submit an issue to an arbitrator, a court can set aside the arbitrator’s decision only in finite circumstances.[31] But if the parties have not agreed to submit a particular dispute to arbitration, the court must decide the answer independently.[32] And unlike arbitral awards, reviewing courts afford no deference to the trial court’s legal determinations and must only defer to factual determinations that are supported by competent evidence.[33] Here, the threshold—and potentially dispositive—issue is who decides whether the parties agreed to arbitrate class claims. The “who decides” question involves two inquiries: whether the availability of class arbitration is a question of arbitrability presumptively for the court or a question to be arbitrated and, thus, presumptively for the arbitrator; and whether the arbitration agreement clearly and unmistakably evinces a contrary intent. Only if we determine that arbitrability of class-action claims is a question for judicial determination are we then empowered to consider whether the Robinsons and HOME agreed to arbitrate those claims.[34]The Robinsons suggest we could simply skip the first step and conclude—based solely on the breadth of the arbitration clause—that the parties agreed to delegate arbitrability questions to the arbitrator. But on that point, we cannot agree. The answer to the first question—who the presumptive decider is—affects whether the contractual language rebuts that presumption with unmistakable clarity. Accordingly, we must first determine whether arbitrability of class claims is a gateway or subsidiary question. A. Who Decides We answered the “who decides” question fifteen years ago in In re Wood, applying Supreme Court authority as settled law.[35] In Wood, we found the Supreme Court’s opinion in Green Tree Financial Co. v. Bazzle[36] to be “directly on point” and relied exclusively on that case as authority for the proposition that the arbitrator has the power to rule on class certification issues when the contract commits all disputes arising out of the agreement to the arbitrator. We cited Bazzle as holding that whether a contract “‘forbids the use of class arbitration procedures[] is a dispute ‘relating to’” the parties’ contract and that “as a question of contract interpretation, the issue of class arbitrability had [therefore] been committed to the arbitrator.”[37] In Bazzle, a plurality of the Supreme Court had concluded that the availability of class arbitration is a contract-interpretation question to be arbitrated because it implicates “what kind of arbitration proceeding the parties agreed to,” rather than an arbitrability question that implicates “the validity of the arbitration clause [or] its applicability to the underlying dispute between the parties.”[38]Based on this conclusion and a broad arbitration clause that did not expressly say otherwise, the plurality concluded the parties had committed “this matter of contract interpretation” to the arbitrator.[39] We relied on Bazzle as controlling on the salient legal point based on the Fifth Circuit’s decision in Pedcor Mgmt Co. Welfare Benefit Plan v. Nations Personnel of Texas, which characterized Justice Stevens’s concurring opinion in Bazzle as supporting the plurality’s “governing rationale” “that arbitrators should be the first ones to interpret the parties’ agreement.”[40] A Supreme Court plurality opinion “holds the stature of the Court’s holding” as to any “‘position taken by those Members who concurred in the judgment on the narrowest grounds ……’”[41] In Wood, we followed Pedcor by treating Bazzle as settling the “who decides” question, stating the Supreme Court had “held that, where parties agreed to submit all disputes to an arbitrator under the Federal Arbitration Act, issues of class arbitration are for the arbitrator to decide.”[42] Over time, however, our reliance on Bazzle and Pedcor has proven to be misplaced.[43] Though Wood squarely decided the “who decides” issue, this is one of those rare circumstances requiring us to reconsider our prior decision.[44] Given the persuasive authority casting doubt on Wood’s core holding, the court of appeals anticipated as much. But even though we reach the same conclusion about Wood’s continued vitality for essentially the same reasons articulated in the court of appeals’ thoughtful and well-written opinion, only this Court can abrogate established precedent.[45] The court of appeals understandably viewed Wood as an anachronism but was obliged to follow it as precedent until we overruled that decision.[46] We do so today. 1. Class Arbitration Presents A Gateway Question of Arbitrability Since Wood issued in 2004, the jurisprudential landscape has evolved to provide a clearer, and distinctly different, perspective. In the last decade, the Supreme Court has issued two opinions emphasizing that whether class arbitration is a gateway or subsidiary question remains an open question that was not answered by Bazzle. In Stolt-Nielsen S.A. v. AnimalFeeds International Corp., the “who decides” issue was not presented because the parties had referred the class-arbitration issue to the arbitration panel via a supplemental agreement.[47] The Court nonetheless discussed Bazzle in some detail, observing that the arbitration panel’s view that Bazzle “controlled” the class-arbitration question was, in the Court’s words, “incorrect.”[48] The Court also noted that the separate writings in Bazzle had evidently “baffled the parties” and caused them to “believe[] that the judgment in Bazzle requires an arbitrator, not a court, to decide whether a contract permits class arbitration.”[49] Examining each of the writings in Bazzle, the Court observed that “no single rationale commanded a majority” and that “only the plurality” had decided that an arbitrator must decide whether a contract permits class arbitration.[50] In Oxford Health Plans LLC v. Sutter, the Supreme Court again addressed Bazzle on the arbitrability question.[51] As in Stolt-Nielsen, the “who decides” issue had been predetermined by the parties, taking the matter out of the Court’s hands and requiring deference to the arbitrator’s construction of the parties’ contract “however good, bad, or ugly.”[52] The only question before the Court was “whether the arbitrator (even arguably) interpreted the parties’ contract, not whether he got its meaning right or wrong.”[53] Observing that a different standard of review might have applied if Oxford had argued “that the availability of class arbitration is a so-called ‘question of arbitrability,’” the Court pointed out that Stolt-Nielsen had (1) “made clear that [the] Court has not yet decided whether the availability of class arbitration is a question of arbitrability” but (2) had “flagged” that the issue “might be a question of arbitrability.”[54] Following the Supreme Court’s rather pointed clarification, several federal circuit courts have addressed the open question unburdened by any misconception about Bazzle’s authoritative force. To date, every one of those courts—the Third, Fourth, Fifth, Sixth, Seventh, Eighth, Ninth, and Eleventh Circuits—has concluded that class arbitrability is for the courts to determine as a gateway matter absent clear and unmistakable language delegating arbitrability matters to the arbitrator.[55] Two supporting rationales have been advanced: (1) the availability of class arbitration invokes contract-formation issues because it implicates whether a presently binding and enforceable agreement to arbitrate exists as to each class member and (2) class action arbitration is so fundamentally different from bilateral arbitration that it implicates the type of controversy the parties agreed to submit to arbitration. Both rationales turn on the parties’ expectations, thus protecting unwilling parties from compelled arbitration of matters they reasonably expected a judge, not an arbitrator, would decide.[56] But most courts to consider the issue have focused on the second rationale. With the benefit of a more full-bodied jurisprudential debate, we are persuaded that—for either or both of the proffered rationales—determining whether the parties have agreed to arbitrate disputes as a class is a threshold question of arbitrability.[57] Class arbitration radically alters the dispute-resolution bargain, changing the very nature of the arbitral proceeding and undermining its principal benefits.[58] As the Supreme Court has often noted, class actions are “an exception to the usual rule that litigation is conducted by and on behalf of the individual[ly] named parties only,”[59] and when bilateral arbitration shifts to class arbitration, the proceedings are fundamentally transformed.[60] In class arbitration, an arbitrator’s award “no longer resolves a single dispute between the parties to a single agreement, but instead resolves many disputes between hundreds or perhaps even thousands of parties.”[61] As a general proposition, “[g]ateway questions are fundamental to the manner in which the parties will resolve their dispute—whereas subsidiary questions, by comparison, concern details. And whether the parties arbitrate one claim or 1,000 in a single proceeding is no mere detail.”[62]Due to enhanced complexity and multiplicity of claims, class arbitration is a high-stakes endeavor that “sacrifices the principal advantage of arbitration—its informality,” making “the process slower, more costly, and more likely to generate procedural morass than final judgment.”[63] Due process rights of absent class members, loss of speed and efficiency, increased costs, and confidentiality concerns are among the unique difficulties class arbitration presents: Classwide arbitration includes absent parties, necessitating additional and different procedures and involving higher stakes. Confidentiality becomes more difficult. And while it is theoretically possible to select an arbitrator with some expertise relevant to the class-certification question, arbitrators are not generally knowledgeable in the often-dominant procedural aspects of certification, such as the protection of absent parties.[64] With so much on the line, “[c]ourts cannot assume that parties would want these kinds of questions to be arbitrated.”[65] It is simply too “hard to believe that defendants would bet the company with no effective means of review.”[66] Considering the “obvious,”[67] “structural,”[68] and “fundamental”[69] differences between bilateral and class arbitration, which “change the nature of arbitration altogether,”[70] we hold that the question of class arbitration is more akin to what type of controversy shall be arbitrated—a question for the courts—not a procedural question presumptively for the arbitrator.[71] The distinctions between bilateral and class arbitration implicate the principal characteristic of gateway issues—namely, the expectation that a judge would ordinarily decide arbitrability of such matters. Concluding that the threshold issue here is a gateway matter also aligns, at least by analogy, with our view that whether a nonsignatory is bound to an arbitration agreement is a gateway matter for judicial determination.[72] Because individual class members may have individual defenses to arbitration—here for example, the limited warranty allows homeowners with VA or FHA financing to elect a judicial remedy—the availability of class arbitration relates to “whether the parties are bound by a given arbitration clause” and implicates “whose claims an arbitrator may decide.”[73] Thus under either of the prevailing rationales, arbitrability of class claims is presumptively for the court, but ultimately depends on what the parties’ contract says about the matter. 2. Delegation to the Arbitrator Cannot Be Inferred from Silence The presumption that courts will decide gateway issues and arbitrators will decide subsidiary issues are interpretive constructs based on assumptions about the parties’ expectations.[74] But arbitration is a matter of contract, so parties are generally free to alter these presumptions by agreement.[75] And “[w]hen the parties’ contract delegates the arbitrability question to an arbitrator, a court may not override the contract [and] possesses no power to decide the arbitrability issue.”[76] Even so, “[a] party often might not focus [on] the significance of having arbitrators decide the scope of their own powers,” so “[c]ourts should not assume that the parties agreed to arbitrate arbitrability unless there is ‘clear and unmistakable’ evidence that they did so.”[77] In this case, neither the limited warranty nor the addendum contains language clearly and unmistakably delegating arbitrability determinations to the arbitrator. Indeed, both are entirely silent on the topic.[78] Magic words are not necessarily required to commit class arbitrability decisions to the arbitrator, but as we have said before, “a contract that is silent on a matter cannot speak to that matter with unmistakable clarity.”[79] The Supreme Court likewise prohibits interpreting silence as assent, explaining that “to interpret silence or ambiguity on the ‘who should decide arbitrability’ point as giving the arbitrators the power . . . might too often force unwilling parties to arbitrate a matter they reasonably would have thought a judge, not an arbitrator, would decide.”[80] Relying on the “broad and sweeping” arbitration clauses in the limited warranty and addendum, the Robinsons dispute that the contracts are actually silent. They cite Wood and Bazzle as holding that broadly worded arbitration clauses clearly and unmistakably commit all disputes about class arbitrability to the arbitrator. Accordingly, they contend that the garden-variety arbitration provisions in their contracts with HOME necessarily empower arbitrators to resolve arbitrability questions. We disagree. Both Wood and Bazzle rested on the premise that class arbitrability is presumptively a question for the arbitrator—the opposite of the presumption applicable here. Applying the opposite presumption would have required the arbitration contracts in Wood and Bazzle to be more or less express in prohibiting the arbitrator from determining arbitrability issues.[81] Here, the presumption favors judicial determination, so contractual silence about “who decides” arbitrability questions does not delegate those gateway questions to the arbitrator with unmistakable clarity. With not a word about arbitrating arbitrability issues, any question about whether parties agreed to arbitrate class claims was for the court to answer. C. No Agreement to Arbitrate Class Claims 1. No Contractual Basis Construction of an arbitration agreement is ordinarily a question of state law, but “state law is preempted to the extent it ‘stands as an obstacle to the accomplishment and execution of the full purposes and objectives’ of the FAA.”[82] The FAA’s most fundamental precept—the “first principle” underscoring all arbitration decisions—is that arbitration “is a matter of consent, not coercion.”[83] The Supreme Court has thus emphasized that “a party may not be compelled under the FAA to submit to class arbitration unless there is a contractual basis for concluding that the party agreed to do so.”[84]What that means has been the central focus of the Court’s recent writings on class arbitration. As articulated in Stolt-Nielsen S.A. v. AnimalFeeds Int’l Corp. and, most recently, Lamps Plus, Inc. v. Varela, neither a state nor a federal court can compel arbitration on a classwide basis under the FAA when an agreement is silent about the availability of class arbitration or is not silent but ambiguous about the matter.[85] Nor can an agreement to arbitrate class claims be inferred “solely from the fact of the parties’ agreement to arbitrate.”[86] Rather than relying on mere inference, an affirmative contractual basis is required.[87] This is so because the “‘crucial differences’ between individual and class arbitration” give “‘reason to doubt the parties’ mutual consent to resolve disputes through classwide arbitration.”[88] “Like silence, ambiguity does not provide a sufficient [contractual] basis to conclude that parties to an arbitration agreement agreed to ‘sacrifice[] the principal advantage[s]‘” and “undermine the central benefits of arbitration itself.”[89] The Supreme Court’s holdings in this regard accord with a fundamental tenet of Texas contract law: we may neither rewrite the parties contract nor add to its language.[90] And because “class-action arbitration changes the nature of arbitration to such a degree that it cannot be presumed the parties consented to it by simply agreeing to submit their disputes to an arbitrator,”[91] we agree that class arbitration must be explicitly referenced and not merely inferred from the parties’ agreement to arbitrate. Here, the arbitration provisions in the limited warranty and addendum do not reference class claims at all; accordingly, there is no affirmative contractual basis for concluding the parties agreed to classwide arbitration. 2. No Consent or Acquiescence By Conduct With no shelter under the contract language, the Robinsons largely focus on the argument that HOME, through its “inconsistent conduct,” acquiesced or consented to the arbitrator’s authority to determine whether class claims were arbitrable and further affirmed that the limited warranty and addendum encompassed an agreement to arbitrate class claims. The Robinsons cite HOME’s conduct in (1) moving the trial court to compel arbitration of the individual claims based on the broadly worded arbitration clause, (2) moving the arbitrator to strike or dismiss the class-action arbitration, and (3) waiting to ask the trial court to deny class arbitration until after the arbitrator had ruled on HOME’s objections and motion to strike those claims. According to the Robinsons, these are not waiver and estoppel theories but are instead arguments about HOME’s manifestations of contractual intent. To the extent HOME’s conduct is even relevant to what the language in the limited warranty and addendum means,[92] we are not persuaded that it alters the outcome here. None of the conduct the Robinsons cite evidences HOME’s clear intent to arbitrate class claims or to be bound by the arbitrator’s decision on arbitrability. Moving to compel arbitration of individual claims is not inconsistent with construing the contracts as being silent about—and therefore not authorizing—classwide arbitration. This is especially so given that HOME moved to compel arbitration of the Robinsons’ individual claims long before they attempted to introduce entirely independent class claims into the arbitral proceedings. Nor did HOME acquiesce to the arbitrator’s power to determine the gateway arbitrability question. When confronted with the Robinsons’ eleventh-hour addition of putative class claims, HOME’s immediate response was to object that those claims were beyond the scope of the order compelling arbitration. Rather than indicating a clear willingness to arbitrate the class-action issues, HOME’s objection establishes the contrary—that it was insisting on the trial court’s determination of whether the claims were arbitrable. Throughout these proceedings, HOME has consistently maintained that the court, not the arbitrator, was empowered to determine whether the parties had agreed to arbitrate class claims. “Merely arguing the arbitrability issue to an arbitrator does not indicate a clear willingness to arbitrate that issue.”[93] HOME’s alternative request to strike the class claims was entirely responsive to the Robinsons’ last-minute efforts to expand the arbitration to include class claims. These are the reasonable actions of a prudent litigant, not a clear manifestation of intent to arbitrate class claims or to be bound by the arbitrator’s arbitrability decisions.[94] III. Conclusion Considering the fundamental differences between bilateral and class arbitration, and the bedrock principle that a party cannot be forced to arbitrate any dispute absent a binding agreement to do so, we hold that a court must determine, as a gateway matter, whether an arbitration agreement permits class arbitration unless the parties have clearly and unmistakably agreed otherwise. Class arbitration implicates what must be arbitrated and who must arbitrate, matters that are presumptively for the court’s determination. Because the arbitration agreements at issue here are silent as to arbitrability and do not mention class claims at all, the lower courts correctly determined HOME was not bound to arbitrate the Robinsons’ putative class claims. We affirm the court of appeals’ judgment. Eva M. Guzman Justice OPINION DELIVERED: November 22, 2019
