Visa’s transformation into an AI-first company has been years in the making. Ramping up this year, the payments giant has woven artificial intelligence into nearly every part of its operations, from fraud detection and data management to software engineering and product design. “AI is part of the fabric of the company,” said Rajat Taneja, Visa’s president of technology, to Fortune. “In everything we do, we’re using it internally, and we’re using it to create intelligent and smarter, more capable products for our customer base.”
That shift has required major investment and architectural rethinking. Over the past decade, Visa has spent $3.5 billion on its AI and data platform, and about $12 billion on technology overall. The result is a system that manages more than 400 AI solutions across risk, cybersecurity, and operations, powering 630 million daily transactions with millisecond precision. Sam Hamilton, Visa’s senior vice president of data and AI, told VentureBeat that the company maintains more than 300 active AI models and “hundreds of petabytes of data,” monitored constantly to prevent drift and bias.
Visa’s internal AI framework includes a hybrid compute strategy that mixes on-premise and cloud infrastructure to balance performance and cost. “We evaluate everything on total cost of ownership and performance per dollar,” Hamilton said. “It’s not about chasing the latest hardware- it’s about optimizing for latency, security, and scalability.”
AI is also changing how Visa’s workforce operates. The company has built more than 100 internal AI applications to support functions such as software development, dispute management, and finance. Since early 2023, every employee has had access to generative AI tools, including Microsoft Copilot and internal large language models, to help with coding, data analysis, and communications. “We’re seeing productivity improvements across the business,” CEO Ryan McInerney said in a CNBC interview. “In engineering, in marketing, in finance, and HR, it’s been a real unlock.”
Governance is a central pillar of Visa’s AI approach. The company maintains an “AI observatory,” a live dashboard that tracks every model’s compliance, security profile, and performance drift. “Just as important as the science of models is the art, the policy and governance,” Taneja said. “We can click on any model, see its risk profile, and whether it meets every privacy law in both letter and spirit.” This oversight structure has allowed Visa to deploy AI responsibly at enterprise scale while maintaining trust across its vast network of partners and regulators.
Turning AI into a Product Engine
While AI has long been part of Visa’s fraud prevention systems, its role is now expanding into product innovation. The company’s latest initiatives: Visa Intelligent Commerce and the Visa Commercial Solutions (VCS) Hub, illustrate how deeply AI is shaping its offerings for consumers, developers, and businesses.
Unveiled at Visa’s Global Product Drop event, Visa Intelligent Commerce integrates AI directly into the payments infrastructure. It allows AI agents, developed by companies such as OpenAI, Anthropic, and Microsoft, to shop and buy on behalf of consumers within trusted limits. “Soon people will have AI agents browse, select, purchase, and manage on their behalf,” said Jack Forestell, Visa’s chief product and strategy officer, to VentureBeat. “These agents will need to be trusted with payments, not only by users, but by banks and sellers as well.”
The system uses Visa’s existing tokenization technology, which replaces card details with secure digital credentials. Each consumer can set spending limits and define which agents are authorized to transact. “You can tell your agent, ‘I want you to spend this amount of money, go to these merchants, and come back in 30 minutes,’” McInerney told CNBC. “We’ve built all that into the Visa network so your bank knows it’s you authorizing the purchase.”
The infrastructure supporting Intelligent Commerce (what Visa calls its Model Context Protocol Server and Acceptance Agent Toolkit) provides developers a standardized way to connect AI agents to Visa’s network without custom integrations. This reduces setup time from weeks to hours and enables what the company calls “agentic commerce.” Early pilots include more than 30 partner companies testing secure, AI-driven purchasing in controlled sandboxes.
In the commercial space, the Visa Commercial Solutions Hub extends these capabilities to businesses and fintechs. The platform automates payables, reporting, and personalization with generative AI at its core. “With GenAI at the heart of the VCS Hub, we’re giving our partners the tools to amaze their clients and unlock new revenue streams,” said Gloria Colgan, Visa’s global head of product for commercial solutions.
Together, these launches reflect Visa’s broader goal to make payments infrastructure “AI-native.” The company is not building its own consumer-facing AI agents; instead, it provides the secure rails that make AI-driven transactions possible. “You want your agent to go buy something for you, and you want it to work everywhere,” McInerney said. “That’s what our network is built for.”
Visa’s approach builds on its longstanding reputation for trust and security. Andres Vives, Visa’s chief data officer, told Forbes that the company’s AI models help prevent more than $40 billion in fraud annually. “Data is our biggest asset at Visa,” Vives said. “We process more than 300 billion transactions a year. We’ve invested heavily to make sure our network can advance at the pace AI is advancing.”
Those same foundations now power consumer-facing experiences like Visa Data Tokens, which allow merchants to personalize recommendations based on AI-generated insights without exposing raw consumer data. “If data is used responsibly, not only can we build products more efficiently, but we can also fuel better commerce interactions,” Vives said.
Building the AI Payments Backbone
Visa’s AI transformation underscores a consistent message: trust remains the company’s differentiator, even as the commerce landscape shifts toward automation. McInerney described the company’s mission as building “AI-ready credentials” that let consumers control their data while giving AI agents a safe way to transact.
Internally, Visa’s disciplined use of AI and observability systems has given it a reputation for operational maturity. Externally, its partnerships across the AI ecosystem (spanning OpenAI, Microsoft, Anthropic, IBM, and Stripe) position it as the payments backbone for the next wave of digital commerce.
As Forestell put it, “The early incarnations of agent-based commerce could shop and discover, but they had tremendous trouble on payments. The payments problem is not something AI platforms can solve by themselves. That’s why we started working with them.”
