SShortSingh.
Back to feed

Google, Visa, Stripe and Others Race to Build Payment Infrastructure for AI Agents

0
·1 views

In the first half of 2026, major players including Google, Mastercard, Visa, Stripe, American Express, Ant International, and Circle have all launched or announced infrastructure designed to let AI agents make payments autonomously on behalf of users. The push reflects a convergence of capable large language model agents, legacy payment rails ill-equipped for machine-initiated transactions, and intense competition among tech and finance firms to own the emerging agentic payments layer. Several competing protocols are now available to developers, including Google's Agent Payments Protocol (AP2), Coinbase's x402, Stripe and Tempo's Machine Payments Protocol, Visa's Trusted Agent Protocol, and Circle's Agent Stack, all of which launched or expanded within months of each other. No single standard has emerged as dominant yet, but the tools have moved beyond whitepapers into installable SDKs that developers can use today. Key challenges remain around authorization, identity verification, and spending controls, as existing infrastructure like card networks and PCI DSS was designed with a human present at the point of transaction.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

Log in to join the discussion and vote.

Log in

Related stories

0
ProgrammingDEV Community ·

Developer Breaks Down AI Concepts in New Series Aimed at Simplifying the Technology

A software architect has launched an educational series on DEV Community aimed at demystifying how artificial intelligence works for developers and general users alike. The author notes that AI has become widely used across professions and age groups, fundamentally changing how people seek information compared to older tools like Stack Overflow. Despite its widespread adoption, the author found that understanding the underlying mechanics of AI was difficult due to fragmented and overwhelming resources. The series intends to explain AI concepts in simple terms, covering processes such as input handling and response generation in tools like ChatGPT, Gemini, and Claude. The goal is to help developers not only use AI effectively but also build deeper knowledge to keep pace with the rapidly evolving field.

0
ProgrammingDEV Community ·

How Python Selenium Architecture Works: Layers, Protocols, and Virtual Environments

Python Selenium automation operates through four key layers: the Python client library, the W3C WebDriver protocol, browser-specific drivers, and the web browser itself. Commands written in Python are translated and sent via HTTP to browser drivers like ChromeDriver or GeckoDriver, which then interact with the browser's native API. Selenium 4 modernized this pipeline by adopting the standardized W3C WebDriver protocol, replacing the older JSON Wire Protocol. Python virtual environments play a critical role in Selenium projects by isolating dependencies, preventing conflicts between projects that require different library versions. For example, two projects needing Selenium 3 and Selenium 4 respectively can coexist safely on the same machine only when managed through separate virtual environments.

0
ProgrammingDEV Community ·

AWS EFS Explained: Shared File Storage for Multiple EC2 Instances

Amazon Elastic File System (EFS) is a fully managed, serverless shared file system that allows multiple EC2 instances across different Availability Zones to read and write data simultaneously using the NFS 4.1 protocol. Unlike EBS, which is tied to a single EC2 instance and a single AZ, EFS automatically scales from kilobytes to petabytes and replicates data across multiple AZs within a region. Access is enabled through Mount Targets — Elastic Network Interfaces provisioned in each AZ — which serve as the connection point between EC2 instances and the file system. EFS follows a pay-as-you-go pricing model, billing only for storage actually used rather than pre-provisioned capacity. It is commonly used for shared content, CMS workloads, and machine learning training datasets where concurrent multi-instance access is required.

0
ProgrammingDEV Community ·

CalcMora hits 200 tools with new embed system and static-first architecture

CalcMora, a free online calculator and converter platform, has reached 200 live tools spanning finance, health, math, and sports, marking a milestone toward its goal of 3,000 tools within a year. The site is built on Astro for static output and hosted on Cloudflare Pages, a deliberately lightweight stack that keeps page speeds fast regardless of how many tools are added. Every tool follows a standardised template including a calculator, explanatory content, an FAQ, and schema.org structured data to support search visibility. Alongside the 200-tool milestone, the platform launched an embed system allowing any tool to be placed on third-party sites as an ad-free widget using a simple copy-paste snippet with no sign-up required. Near-term development will focus on scaling the content pipeline while maintaining consistency, with more distribution-focused features planned as the tool count grows.

Google, Visa, Stripe and Others Race to Build Payment Infrastructure for AI Agents · ShortSingh