The race to build the payment infrastructure of the AI-agent economy is entering a new phase, with global payment networks and technology companies competing to define how machines will identify, authorize and execute transactions.
Unlike the consumer-facing AI shopping race, where companies compete to help users purchase products through assistants, the emerging battle is taking place at the infrastructure layer: who controls the standards that allow autonomous software agents to exchange value.
The industry is increasingly resembling previous technology format battles, where competing standards fought to become the foundation for an entire agentic payments ecosystem. This time, the prize is the future of machine-to-machine payments.
At the centre of the competition are four major approaches: the open x402 protocol, Visa’s Trusted Agent Protocol, Mastercard’s Agent Pay framework and UnionPay’s Agentic Payment Open Protocol.
Agentic Payments Protocols: Key Players and Approaches
x402 Foundation
- Main backers: Linux Foundation, Visa, Mastercard, Google, AWS, Circle and others
- Region focus: Global open standard
- “Know Your Agent” approach: Enables agent payments through internet-native payment infrastructure and interoperable payment rails.
Visa Trusted Agent Protocol
- Main backers: Visa
- Region focus: Global payments ecosystem
- “Know Your Agent” approach: Uses cryptographic agent signatures, intent verification and consumer recognition to establish trust between AI agents and payment networks.
Mastercard Agent Pay
- Main backers: Mastercard
- Region focus: Global payments ecosystem
- “Know Your Agent” approach: Focuses on agent credentialing, verified intent, permissions management and transaction controls to ensure secure autonomous payments.
UnionPay Agentic Payment Open Protocol
- Main backers: UnionPay
- Region focus: China and international markets
- “Know Your Agent” approach: Combines AI-driven payment intelligence, privacy-preserving verification and financial models to support agent-led transactions.
The x402 Foundation has emerged as the strongest push for an open industry standard. Launched under Linux Foundation governance in July, the initiative aims to create a neutral framework for payments between AI agents, applications and digital services. The protocol uses the long-reserved HTTP 402 “Payment Required” mechanism to enable machine-driven transactions.
Its supporters represent nearly every layer of the future payments stack, from card networks and cloud providers to stablecoin companies and commerce platforms.
Meanwhile, Visa and Mastercard are focusing on trust and identity — a critical challenge as AI agents gain the ability to spend money without direct human involvement.
Visa’s Trusted Agent Protocol introduces agent-specific cryptographic signatures designed to confirm an agent’s identity, intent and relationship with consumers during commerce interactions.
Mastercard’s Agent Pay takes a similar approach by creating a framework where agents are credentialed, authenticated and limited by predefined permissions before they can transact. The company describes “Know Your Agent” capabilities as essential for ensuring autonomous payments remain accountable and transparent.
China’s UnionPay is also positioning itself in the emerging standards race. During the World AI Conference in Shanghai on July 17, the company introduced its Agentic Payment Open Protocol framework alongside AI technologies designed for financial transactions and privacy-preserving inference.
The competition highlights a fundamental question for the next generation of commerce: will the future belong to a single dominant payment protocol, or will multiple standards coexist?
Traditional payment networks bring decades of regulatory experience, merchant acceptance and security infrastructure. Open protocols offer flexibility, developer adoption and interoperability. Stablecoin-based systems add another layer by enabling programmable, always-on settlement.
The winner may not simply be the company with the largest payment network. It could be the ecosystem that solves the hardest problem in autonomous finance: proving that an AI agent is legitimate, authorized and acting within human-defined boundaries.
As AI agents move from recommending actions to executing transactions, the battle for the machine payments standard has officially begun. The companies defining agent identity, permissions and settlement rules today could shape the financial infrastructure of the AI economy for years to come.
