What is Visa's AI strategy?
Visa treats AI as both its oldest risk technology and, in CEO Ryan McInerney's words on the July 28, 2026 call, the force "transforming the front end" of commerce while stablecoins reshape the back end. On the April 28, 2026 call McInerney said AI and agentic commerce would expand Visa's addressable market in four ways, citing third-party estimates of 80 to 150 basis points of extra GDP growth. Rajat Taneja, president of technology, told Fortune in October 2025 that Visa spent $3.5 billion building its AI platform and about $12 billion on technology over five years, and that every employee had access to a generative AI model by February 2023. Visa works with models from OpenAI, Anthropic, Google, IBM, Meta and Mistral AI, Fortune reported in February 2025; in April 2026 Taneja said employees used 1.9 trillion AI tokens a month and 89% were active AI users. In October 2025 CFO Chris Suh said every leader has AI targets to drive efficiencies. In July 2026 Visa said it had more than 150 AI-powered applications and was reorganising product teams into "agentic squads" of two to four people, the same day it announced about 2,600 job cuts in which AI was a factor. Its products run from Visa Advanced Authorization to Visa Intelligent Commerce (April 2025) and a Large Transaction Model (June 2026). The FY2025 10-K says agentic commerce "presents novel and complex regulatory, privacy and cybersecurity risks". Visa does not disclose AI revenue or an AI headcount.[1][2][3][4][5][6][7][8][9][10]
Data as of Oct 22, 2026 · Visa profile updated Oct 22, 2026 · 24 sources cited on this page · 52 fintechs profiled on this site · Markdown
What have Visa's CEO and technology chief said about AI, and when?
In date order. Earnings-call lines come from published transcripts (The Motley Fool, Investing.com); interview lines are as reported by the outlet named.
| Statement or move | Who | When | Source |
|---|---|---|---|
| Visa spent a few years and $3.5 billion rebuilding its data platform; it works with models from OpenAI, Anthropic's Claude, Google, IBM, Meta and Mistral AI, and nearly all staff have a secure internal version of ChatGPT | Rajat Taneja, president of technology | February 5, 2025 | Fortune[4] |
| "We're making a very deliberate effort to focus on open banking in the markets that have the greatest potential, such as Europe and Latin America" | Ryan McInerney, CEO | July 29, 2025 | Q3 2025 earnings call[11] |
| Visa spent $3.5 billion building its AI platform and around $12 billion on technology over five years; every employee had model access by February 2023; Visa joined Microsoft's early Copilot pilot | Rajat Taneja, president of technology | October 7, 2025 | Fortune[3] |
| Over half of the new code base for the next generation of VisaNet was built with the assistance of generative AI | Ryan McInerney, CEO | October 28, 2025 | Q4 2025 earnings call[6] |
| "Every leader at the company has AI targets to drive efficiencies that we intend to invest back in the business" | Chris Suh, CFO | October 28, 2025 | Q4 2025 earnings call[6] |
| Employees used 1.9 trillion AI tokens a month as of March, double February; 89% are active AI users and 44% power users; Claude is the most popular chatbot internally, often swapping places with ChatGPT | Rajat Taneja, president of technology | April 10, 2026 | Business Insider[5] |
| AI and agentic commerce "will expand our addressable market" in four ways: digitising commerce, agents paying for resources in micro transactions, digitising B2B payments, and 80 to 150 basis points of extra GDP growth per third-party estimates | Ryan McInerney, CEO | April 28, 2026 | Q2 2026 earnings call[2] |
| "If stablecoins are reshaping the back end of commerce, we see AI as transforming the front end"; more than 150 AI-powered applications; product teams of 10 or more reformed into "agentic squads" of two to four | Ryan McInerney, CEO | July 28, 2026 | Q3 2026 earnings call[1][12] |
| About 2,600 positions, roughly 7% of the workforce, to be cut, mostly in technology and product; a person with direct knowledge told CNBC AI was a significant factor but not the sole driver | Visa | July 28, 2026 | CNBC[7] |
| Teams on the agentic tool chain saw "80% more code commits" and requirement definition cut from 30 days to five, translating to "65% plus faster feature development" | Ryan McInerney, CEO | July 28, 2026 | Q3 2026 earnings call[1] |
Which AI products has Visa shipped, and in what order?
A selection of named AI products and AI-related acquisitions; fraud models are covered in depth on Visa's fraud and risk AI page.
| Product | What it does | Built on | Since |
|---|---|---|---|
| Visa Advanced Authorization | Machine-learning fraud scoring of Visa card payments | Neural networks Visa first deployed for risk in 1993 | 1993 onward[13][10] |
| Featurespace | Real-time behavioral analytics and AI payments protection, folded into Visa's risk unit | Acquisition, $946 million purchase consideration | December 19, 2024[14][10] |
| Employee AI assistant and GenAI Hub | Internal assistant using agentic AI for self-service; nearly 26,000 employees launched over 261,000 AI-powered chats from March 2025 to September 30, 2025 | Visa's internal GenAI Hub of models and tools | March 2025[10] |
| Visa Intelligent Commerce | Opens Visa's network to developers of AI agents that browse, select and buy for consumers | Tokens and APIs; Anthropic, IBM, Microsoft, Mistral AI, OpenAI, Perplexity, Stripe and Samsung named as collaborators | April 30, 2025[8] |
| MCP server | Gives AI systems access to Visa Intelligent Commerce APIs | Model Context Protocol | September 2025[15][6] |
| Trusted Agent Protocol | Open framework for agents to pass signed intent and payment data so merchants can tell trusted agents from bots | Developed with Cloudflare | October 14, 2025[16] |
| Large Transaction Model | Fraud model trained on billions of transactions, announced with an OpenAI partnership and an Agentic Directory | Visa transaction data | June 10, 2026[9][17] |
| Visa Vulnerability Agentic Harness | Open-sourced harness directing AI models through vulnerability discovery, verification and remediation | Tested with Anthropic's Claude Mythos under Project Glasswing | June 10, 2026[18] |
| BioCatch (agreed) | Behavioral, device and network AI to detect account takeover, scams and mule accounts | Acquisition agreed for $2.4 billion | August 3, 2026[19] |
How is AI governed inside Visa, what does the 10-K say, and what is not disclosed?
Technology and AI sit under Rajat Taneja, president of technology, who told Fortune the approach treats "the sciences of models" and "the art, which is the policy and the governance" as equally important. He described codifying data-use principles into observability systems and an "AI observatory" dashboard that watches models in production for misuse and drift.[3]
Taneja said in February 2025 that he picks models on transparency, fairness efforts, the ability to fine-tune and hallucination rates, and that Visa matches models to software development, risk modelling and agentic workflows. In April 2026 he said Visa rewards employees for AI use, with software engineering the heaviest user.[4][5]
The FY2025 10-K says Visa is investing in "generative AI (GenAI), stablecoins and agentic commerce", that its Smarter Stand-In Processing uses AI to authorise transactions for issuers during outages, and that its risk products are "underpinned by real-time AI-driven scores".[10]
The 10-K risk factors say there is no single global regulatory framework for AI, list risks from training data, intellectual property, personal data and safe deployment, and warn that as agentic commerce scales Visa "may see increased instances of erroneous or disputed payments, increased chargebacks and reputational harm".[10]
What is not disclosed: an annual AI budget (the $3.5 billion platform figure covers several years), AI headcount, revenue from AI-embedded services, how many of the July 2026 job cuts were attributed to AI, and which of the more than 150 AI applications are customer-facing.[3][1][7]
What does the Visa profile record on this topic?
The rows below are the entries of the Visa AI strategy profile that bear on this page, as of Oct 22, 2026; the profile holds the full record of 17 dated moves, 11 use cases and 24 sources.
| Use case | What the record shows | Status | Source |
|---|---|---|---|
| Large Transaction Model | An AI model trained on billions of transactions, announced June 2026, intended to improve fraud detection while raising authorization performance and cutting false declines. | Announced | [9] |
| A2A Protect (with Featurespace) | Real-time risk scoring for account-to-account payments using transfer learning and network-level signals; Visa says it increased fraud detection by 75% in the first six months of deployment, and the September 2026 version adds a unified Featurespace-based score with plain-language alert explanations. | In production | [20] |
| BioCatch behavioral intelligence (pending) | Behavioral-biometrics models covering account takeover, scams, money mules and application fraud; Visa agreed to buy BioCatch for $2.4 billion in August 2026, with closing expected by the end of fiscal Q2 2027. | Announced | [19][21] |
| Visa Vulnerability Agentic Harness | A governed multi-model pipeline that directs frontier models, including Anthropic's Claude Mythos, through threat modelling, verification, exploit-chain analysis and remediation of Visa's own code; open-sourced in June 2026 with human review of findings, and extended to remediation and validation in August 2026. | In production | [18][22] |
| Date | Move | Detail | Source |
|---|---|---|---|
| Sep 1, 2026 | A2A Protect adds first Featurespace-based unified score | Visa launches an enhanced A2A Protect with a unified fraud score, its first in-market integration of Featurespace technology; each alert carries a plain-language explanation of why a payment was flagged. | [20] |
| Aug 27, 2026 | Vulnerability harness extended to remediation | Visa releases a new version of the Visa Vulnerability Agentic Harness that adds closed-loop remediation and validation and lets organizations deploy approved Anthropic and OpenAI models or other models by configuration; Visa says it has been downloaded by tens of thousands of developers since June 2026, and launches cybersecurity advisory services around it. | [22] |
| Aug 3, 2026 | Agreement to acquire BioCatch for $2.4 billion | Visa signs a definitive agreement to buy BioCatch, whose AI and machine-learning models analyse behavioral, device and network signals to detect account takeover, scams and mule accounts, from funds advised by Permira; closing is expected by the end of Visa's fiscal Q2 2027, subject to regulatory approval. | [19][21] |
| Jul 28, 2026 | About 2,600 jobs cut; AI cited as one factor | Visa plans to eliminate roughly 2,600 positions, about 7% of its workforce, mostly in technology and product. A person with direct knowledge told CNBC AI was a significant factor but not the sole driver. Fiscal Q3 results the same day include $563 million of severance costs. | [7][23][24] |
| Jun 10, 2026 | Visa Vulnerability Agentic Harness open-sourced | After testing Anthropic's Claude Mythos against its own infrastructure under Project Glasswing, Visa releases the harness that directs AI models through vulnerability discovery, verification and remediation, with human review at key points. | [18] |
| Jun 10, 2026 | OpenAI partnership, Agentic Directory and Large Transaction Model | At Visa Payments Forum Visa announces a strategic collaboration with OpenAI to enable Visa payments in agentic commerce, an Agentic Directory of verified agents and merchants, and a Large Transaction Model trained on billions of transactions. | [9][17] |
| Apr 30, 2025 | Visa Intelligent Commerce launches | At the Visa Global Product Drop, Visa opens its network to developers building AI agents that browse, select and buy for consumers, naming Anthropic, IBM, Microsoft, Mistral AI, OpenAI, Perplexity, Stripe and Samsung as collaborators. | [8] |
| Dec 19, 2024 | Featurespace acquisition completes | Visa completes its acquisition of Featurespace, a Cambridge, UK developer of real-time AI payments-protection technology, folding it into the Risk and Identity Solutions unit. Visa's FY2025 10-K puts the purchase consideration at $946 million. | [14][10] |
How much has Visa spent on AI?
Rajat Taneja told Fortune in October 2025 that Visa spent $3.5 billion building its AI platform and about $12 billion on technology overall over five years. In 2023 Visa said it had spent more than $3 billion on AI and data infrastructure over 10 years. Visa does not report an annual AI budget.
Which AI models does Visa use?
Fortune reported in February 2025 that Visa works with OpenAI, Anthropic's Claude, Google and open-source models from IBM, Meta and Mistral AI. In April 2026 Taneja told Business Insider that Claude was the most popular chatbot among employees, often swapping places with ChatGPT, with Google's Gemini also widely used.
Did Visa cut jobs because of AI?
Partly, by its own account. On July 28, 2026 Visa said it would cut about 2,600 roles, roughly 7% of its workforce, mostly in technology and product. A person with direct knowledge told CNBC AI was a significant factor but not the sole driver, and McInerney described reorganising product teams into smaller agentic squads.
What does Visa's CEO say about AI and agentic commerce?
McInerney said on the April 28, 2026 call that AI and agentic commerce will expand Visa's addressable market in four ways, including micro transactions by agents and faster B2B digitisation. On July 28, 2026 he said AI is transforming the front end of commerce and called agentic commerce "a when, not an if".
What AI risks does Visa list in its 10-K?
The FY2025 10-K says AI regulation is fragmented with no single global framework, lists risks from training data, intellectual property and personal data, and says agentic commerce presents novel regulatory, privacy and cybersecurity risks, including more erroneous or disputed payments, more chargebacks and reputational harm as it scales.
- Earnings call transcript: Visa tops Q3 2026 estimates, raises outlook
- Visa (V) Q2 2026 earnings call transcript
- How Visa wove AI into every facet of the company by approaching it as both a science and an art
- How Visa places big bets on AI and gen AI to secure 639 million daily transactions
- Visa says it's burning through almost 2 trillion AI tokens per month
- Visa (V) Q4 2025 earnings call transcript
- Visa is cutting 7% of employees in efficiency push as AI reshapes work
- The Future is Here: Visa Announces New Era of Commerce Featuring AI
- Visa Announces New AI, Stablecoin and Token Innovations to Power Intelligent, Programmable Commerce at Visa Payments Forum
- Visa Inc. Form 10-K for the fiscal year ended September 30, 2025
- Visa (V) Q3 2025 earnings call transcript
- Visa Expands Money Movement as Card Spending Accelerates
- 30 years of AI and counting (Rajat Taneja, President of Technology)
- Visa Completes Acquisition of Featurespace
- Visa advances agentic commerce with MCP Server, Acceptance Agent Toolkit
- Visa Introduces Trusted Agent Protocol: An Ecosystem-Led Framework for AI Commerce
- Visa Partners with OpenAI to Power the Next Generation of AI Commerce
- Visa Releases Its AI-Powered Cyber Defense System to Open Source
- Visa to Acquire BioCatch
- Visa Launches Enhanced A2A Protect Innovations to Help Financial Institutions Stop Fraud Before Money Leaves Accounts
- Visa to buy cybersecurity firm BioCatch for $2.4 billion amid surge in AI-powered scams
- Visa Expands Support for its Clients and the Industry as Organizations Navigate New AI Era of Cybersecurity
- Visa says AI is helping to enable workforce cuts
- Visa Reports Fiscal Third Quarter 2026 Results (Form 8-K, Exhibit 99.1)
When Visa moves on AI, you'll read it here first.
when a bank, fintech or regulator moves on AI, it is in the next morning's brief · six sourced stories · 7 am PT · free
plus every regulation, bank and agent page update, the morning after · leave any morning