Fintech Giant Nubank's Ambitious US Expansion Hits Payments Liability Roadblock

Two significant structural questions within the fintech industry are currently converging, presenting both opportunities and challenges: the feasibility of a Latin American neobank's expansion into the highly saturated US retail banking market, and the critical need to resolve the accountability vacuum surrounding agentic AI payments before it escalates into a major regulatory crisis.

Nubank, a prominent neobank serving approximately 140 million customers globally, has actively pursued its ambitions in the United States on two distinct fronts. Initially, it established a retail banking foothold through a strategic partnership with Lead Bank. More substantially, Nubank secured a conditional approval from the Office of the Comptroller of the Currency (OCC) for a national bank charter, setting the stage for a potential full license by 2027. Obtaining a direct charter is paramount, as it would grant Nubank direct control over its US operations, significantly impacting unit economics, product flexibility, and its overall regulatory standing, rather than relying on a sponsor bank relationship. Concurrently, rumors circulated about a potential acquisition of Monzo, framed as a gateway to European expansion; however, Nubank swiftly denied these reports. The denial is commercially logical, as an acquisition into a new market entails inheriting another company’s regulatory obligations, growth strategy complexities, and management challenges, precisely when Nubank is already absorbing the considerable costs and risks associated with its new US market entry. Furthermore, Monzo's own previous struggles to gain traction in the US make it an unlikely accelerant for Nubank's international push.

The more profound question facing Nubank is whether its successful model, which flourished in Brazil and across Latin America by offering straightforward, low-cost digital accounts to underserved populations in markets with expensive and inaccessible incumbent banking, can translate effectively to the US. The American market presents a stark contrast; it is heavily saturated with established retail banks, well-capitalized digital challengers, and payment providers that already compete aggressively on both product offerings and pricing. Customer acquisition in such an environment is notoriously expensive, switching costs for consumers are tangible, and cultivating a checking account into a primary banking relationship typically spans years. Even Nubank’s impressive 140 million-strong customer base elsewhere does not inherently accelerate this arduous process in the US.

Running parallel to Nubank's geographic expansion is a broader industry-wide discourse concerning the emergence of AI agents. These sophisticated systems are capable of shopping, comparing financial products, managing subscriptions, and initiating payments autonomously on behalf of a customer. While the commercial opportunities presented by such agentic payments are substantial, a robust regulatory and liability framework to support them is conspicuously absent. The core issue stems from a critical diffusion of responsibility. An AI payment agent operates across a complex chain involving the customer, their bank, a payment network, a technology provider, and potentially a merchant. When an agent initiates an unauthorized transaction, becomes compromised, or simply makes a poor decision, existing payment protection frameworks were not designed to effectively assign liability across such a multi-party chain. A fundamental question, largely unresolved in most jurisdictions, is whether a customer’s general permission to use an agent constitutes blanket authorization for every individual transaction the agent might subsequently initiate.

Banks developing proprietary AI agents gain several advantages, including enhanced control over customer data and a reduced risk of technology intermediaries displacing the direct banking relationship. However, this approach carries a significant downside: fragmentation. The proliferation of dozens of closed, bank-specific agent systems that cannot seamlessly interoperate across different merchants, payment networks, and competing financial institutions would ultimately lead to a poorer customer experience compared to the open, competitive environment the industry notionally champions. The alternative, where independent technology companies own and operate the agent layer, poses a different risk: it could relegate banks to mere utility infrastructure roles, present in the transaction but largely absent from the direct customer relationship.

Regulators are closely monitoring these developments. In the UK, the Financial Conduct Authority (FCA) has expressed keen interest in how AI systems interact with financial accounts and the appropriateness of existing consent frameworks. The Consumer Duty regime, which mandates that firms deliver good outcomes for retail customers, will apply irrespective of whether the firm is the bank, the agent provider, or the underlying network. Similarly, in the EU, the Digital Operational Resilience Act (DORA) introduces operational resilience obligations that extend across the same multi-party agent architecture. The immediate imperative for the financial industry is to establish clear rules regarding consent, permissions, and liability allocation before agentic commerce scales significantly. Without such a comprehensive framework, the argument of convenience alone will prove insufficient to build and sustain lasting customer trust.