Key Takeaways:
The next great platform race is not for your phone or your browser, but for the wallet your AI will use to spend your money.
A race to build the economic infrastructure for autonomous AI agents is heating up, as payments giants Stripe and Ant International launch competing wallet protocols. The moves aim to capture the agentic commerce market, projected to reach $28 billion by 2030, by defining how software pays for goods and services on a user's behalf.
"If AI can solve Nobel level physics problems but can’t buy a domain, something’s gone wrong," Will Gaybrick, Stripe’s president of product and business, said. "Our mantra: empower agents."
Stripe on April 29 launched its Link’s wallet for agents, a system that lets users approve agent-initiated spending via one-time virtual cards or scoped tokens, shielding actual card numbers. Days later, Ant International introduced the Agentic Mobile Protocol (AMP), an open-sourced framework designed for digital wallets and super apps that cuts integration steps by 50% compared to traditional card-binding methods.
The company that owns the wallet, token format, and merchant catalog for agentic commerce could collect a fee on billions of future transactions. The battle for this new payments substrate involves not just Stripe and Ant, but also their partners, including Visa, Mastercard, and Google, with the winner setting the terms for how AI interacts with the global economy.
This payments infrastructure is being built just as tech giants plan the hardware and software where agents will live. OpenAI is reportedly accelerating plans for an "AI agent smartphone" for mass production as early as 2027, potentially using a custom MediaTek Dimensity 9600 chip. Separately, Meta Platforms is developing its own advanced "agent-based" AI assistant for its 3 billion users, powered by a new Muse Spark AI model, signaling a push to integrate autonomous agents directly into its core platforms.
The convenience of autonomous purchasing introduces significant security challenges. Both Stripe and Ant are building identity and control layers to mitigate risk. Ant’s protocol features a “Know Your Agent” framework to certify an agent’s authorized actions and an Agent Trust Rating system for dynamic risk control. Stripe’s architecture, for now, requires per-transaction user approval and uses scoped credentials that limit spending to a specific merchant and amount, preventing an agent from ever accessing the underlying payment method.
This article is for informational purposes only and does not constitute investment advice.
