NEW YORK, 6 MAY 2026 — PayPal's restructuring team has placed its total addressable market at US$860 billion across its three new business units — a figure that underscores just how large the financial infrastructure opportunity is as artificial intelligence begins executing commercial transactions autonomously, without human initiation at each step. The number is not merely an investor relations figure; it is a structural claim about a transformation in how money moves that is already underway, with Stripe, IBM, and a cohort of AI-native challengers all racing to define the architecture of what is being called agentic commerce.

Agentic commerce — the execution of financial transactions by AI systems acting on behalf of users or businesses, without requiring explicit human approval at each step — represents the natural next phase of a digitisation process that has been underway since the first online payment platforms launched in the late 1990s. The difference between digital commerce and agentic commerce is the role of human judgement: in digital commerce, humans make every decision and digital infrastructure executes it; in agentic commerce, AI systems make routine decisions within pre-defined parameters while humans set the parameters and review exceptions. The financial infrastructure required to support this shift is the US$860 billion opportunity that PayPal, Stripe, IBM, and their competitors are competing to own.

Stripe's 288 New Products: The Agentic Infrastructure Play

Stripe's May 2026 product release — 288 new features and products launched in a single announcement — was designed to signal that the company is building the infrastructure layer for agentic commerce rather than competing at the application layer. The partnership with Google for AI-powered trading environments extends Stripe's reach into the machine-to-machine payment use case that agentic commerce requires: when an AI system needs to pay for a cloud API call, purchase a data subscription, or settle a supplier invoice without human intervention, the payment infrastructure must support programmatic initiation, automated compliance checking, and machine-readable confirmation.

Stripe's Link wallet, which now serves 250 million users, is the consumer-facing component of the same infrastructure. Link stores payment credentials that are accessible to merchant applications and, increasingly, to AI agents acting on behalf of consumers. An AI personal finance agent that manages a user's subscriptions, negotiates renewal pricing, and executes payments within pre-approved spending parameters needs a payment credential that the agent can access programmatically — Link provides exactly this. The 250 million user base means that any AI agent designed to execute consumer financial tasks on behalf of its users has immediate access to a payment network covering a quarter of a billion people, without requiring the user to re-enter payment details or approve individual transactions.

Stripe's decision not to cut workforce while competitors are restructuring is a deliberate strategic signal. Where PayPal and Coinbase are cutting headcount to achieve the efficiency ratios their AI-native competitors can match, Stripe was built as an API-first developer platform from inception and does not have the large manual operational workforce that requires replacement. Stripe's expansion while legacy fintechs contract illustrates that the agentic commerce transition is not uniformly painful — it is painful for companies that built human-intensive operating models during the growth era and must now restructure them.

IBM Sovereign Core: Regulated Industries Enter the Agent Era

IBM's announcement of Sovereign Core at its Think 2026 conference on 5 May 2026 addresses the specific challenge of deploying agentic AI in environments where data residency, regulatory compliance, and auditability requirements have made the adoption of conventional cloud AI services legally or operationally impossible. Sovereign Core is designed for governments and highly regulated financial institutions that need AI agent capabilities — automated financial analysis, compliance processing, document review — but cannot deploy those capabilities on US hyperscaler infrastructure without risking regulatory sanction or data sovereignty violations.

The timing of the announcement is directly connected to the emerging agentic commerce landscape. As financial institutions begin evaluating AI agents for treasury operations, AML investigations, and trade settlement, the compliance question becomes central: can the AI agent's decision-making process be audited? Can the data it processed be demonstrated to have remained within the required jurisdictional boundaries? Can the agent's actions be attributed to a specific decision-maker for regulatory accountability purposes? IBM's Sovereign Core is engineered to answer all three questions affirmatively, in environments where the alternative platforms cannot.

US Treasury Secretary Scott Bessent's warning on 3 May 2026 about the increasing threat of AI-powered bank account attacks provides the risk context that makes Sovereign Core's compliance architecture commercially relevant beyond its regulatory appeal. As AI systems on both the attack and defence side become more capable, the financial institutions that can deploy AI agents in compliant, auditable, jurisdictionally-controlled environments have a structural advantage over those deploying on generic cloud infrastructure with limited auditability.

The AI-Native Challengers: Slash and Enter

The incumbents' restructuring creates market space that AI-native challengers are moving quickly to occupy. Slash Financial, the Brazilian fintech that raised €93 million in April 2026, is a representative example. Slash's "Twin" product is an AI financial agent for business clients — managing accounts payable, accounts receivable, and cash flow optimisation through an AI interface that executes transactions within pre-approved parameters. Slash is not trying to be a general-purpose payment processor; it is targeting the specific use case of AI-mediated business treasury operations for SMEs that have never had dedicated treasury teams and would welcome an agent that manages their financial operations intelligently.

Enter, which raised US$100 million at a US$1.2 billion valuation in a round led by Founders Fund, addresses the adjacent domain of AI-mediated legal and compliance operations for financial services companies. Enter's AI handles litigation management for clients including Airbnb, automating the analysis, response, and resolution of legal claims at a scale that human legal teams cannot match cost-effectively. The intersection of AI-mediated financial operations and AI-mediated legal compliance — both moving toward autonomous execution within pre-defined parameters — is the structural territory where agentic commerce and agentic legal services converge.

Singapore: MAS Project Orchid and the Regulatory Foundation

Singapore's Monetary Authority of Singapore has been building the regulatory foundation for agentic commerce since 2022 through Project Orchid, its initiative to develop a framework for Purpose-Bound Money — programmable digital currency that can be configured to be spendable only on specified goods, services, or within defined timeframes. Purpose-Bound Money is the regulatory and technical precursor to agentic commerce: it establishes the legal and technical infrastructure for money that responds to conditions and constraints programmatically, which is the same infrastructure that agentic payment systems require.

The MAS Payments Services Act, which provides the licensing framework for payment service providers in Singapore, is being actively evaluated for amendments that would address the novel agency questions raised by AI-initiated payments: who bears liability when an AI agent executes a payment that the account holder later disputes? What disclosure obligations apply to financial agents? What audit trail requirements are necessary for regulatory oversight of agentic payment flows? Singapore's regulatory clarity in addressing these questions will determine whether Singapore becomes the preferred jurisdiction for agentic commerce pilots in ASEAN — or whether the regulatory ambiguity pushes innovation to less scrutinised jurisdictions.

For ASEAN broadly, the agentic commerce opportunity is large and unevenly distributed. Indonesia's digital economy generated an estimated US$146 billion in gross merchandise value in 2025, with digital payments as the foundational infrastructure. Vietnam's rapidly growing e-commerce sector and the Philippines' large remittance economy are both environments where AI-mediated payment optimisation — finding the fastest, cheapest path for a cross-border payment, automatically handling currency conversion and compliance checking — would deliver immediate value. The ASEAN payment connectivity frameworks being developed through Project Nexus and AFMGM+3 processes create the interoperability infrastructure on which agentic payment applications can eventually run.

Sources

  • Born2Invest — PayPal Fintech Shift: Agentic AI Industry (May 2026)
  • IBM Think 2026 — Sovereign Core Announcement, 5 May 2026
  • Stripe — Product Launch Announcements, May 2026
  • MAS — Project Orchid Purpose-Bound Money Framework, 2024
  • US Treasury Secretary Bessent — Financial Security Statement, 3 May 2026
  • Founders Fund — Enter Investment Announcement, April 2026