Mastercard Incorporated (NYSE:MA) is moving quickly to keep AI shopping agents on its payments network. The company is partnering with Alchemy to give developers the ability to integrate Mastercard-backed virtual cards directly into AI agents, positioning the network for the growth of agentic commerce.
On the surface, this provides Mastercard with a clear way to participate in the growth of AI-driven shopping. However, the more important question for institutional investors is whether the company can retain those transactions over the long term. If AI agents are designed to optimize purely for cost, they could eventually divert purchases off card rails and onto lower-cost stablecoin payment rails, eroding the transaction volume on which Mastercard’s network fees depend.
We recently examined how Visa is positioning itself for the next wave of stablecoin payments, raising a similar question about whether traditional payment networks can capture growth even as alternative rails expand. Read here: Visa Positions Itself for the Next Wave of Stablecoin Payments.
Mastercard and Alchemy Build Infrastructure for AI-Driven Purchases
Alchemy’s AgentCard will bring Mastercard payment credentials to AI agents through an integration with Mastercard Agent Pay. The partnership will allow developers to issue virtual cards to AI agents while setting built-in spending limits and restrictions on what and where those agents can purchase. Cardholders can determine how much control the agent has over each transaction. They can pre-approve purchases that fall within defined parameters or require the AI agent to seek approval before completing a transaction. The partnership is part of a broader industry effort to build payment infrastructure for AI agents. Visa’s Intelligent Commerce and similar toolkits from PayPal, Stripe, and Coinbase are also targeting this emerging market. Nikil Viswanathan, co-founder and CEO of Alchemy, said:
AI agents are quickly moving from simple assistants to software that can take action for people. To make that possible in everyday commerce, agents need a trusted way to pay, prove authorization, and operate within clear user-defined limits.
Limited Adoption and a Bigger Structural Risk
Despite the growing number of announcements around agentic commerce, adoption remains limited. A January Omnisend survey found that only 8.29% of U.S. shoppers were fully comfortable with AI completing online purchases for them. In 2025, Visa’s risk unit reported a sharp increase in dark web references to “AI Agent”, with such posts rising more than 450% over six months compared with the prior six months. More importantly, Citrini Research’s February scenario, which it framed as a thought experiment rather than a forecast, points to a potential long-term challenge for traditional card networks.
It imagines AI agents built to prioritize cost above other considerations; they could eventually bypass card rails, where interchange fees are commonly cited at 2% to 3%, and shift transactions toward stablecoin rails, where the cost can be just a fraction of a cent per transaction. Mastercard does not collect that interchange, which goes to issuing banks. However, it earns network fees on the volume flowing through its system, so any spending that migrates off card rails reduces its revenue directly.
Hedge fund interest in Mastercard was largely unchanged, with the number of funds holding the stock slightly increasing from 157 at the end of Q1 2026 to 158 at the end of Q2 2026. Short interest stood at just 0.87% of float as of August 31, 2026, showing that bearish positioning remains minimal.
For now, Mastercard’s fee model looks protected, but the long-term risk is whether AI agents eventually route more payments toward cheaper stablecoin rails.
READ NEXT: Bernstein’s Defense of GE Vernova Doesn’t Answer the Valuation Question and Nvidia’s CUDA-Q Logical Extends Its Quantum Software Moat, But the Hardware Timeline Remains Uncertain
Hedge Fund Resource Center