V VISA CLASS A $653.3B

What they are doing

The quarterly report for the period ended 30 June 2026, filed 29 July 2026, does not name Visa Consumer Authentication Service, Visa Advanced Authorization, or Visa Provisioning Intelligence. It says the February 2026 purchase of Prisma and Newpay in Argentina is expected to speed tokenization, biometric authentication, intelligent risk tools, and agentic commerce. That is an expectation about an acquisition, not a result on a score. The 28 July 2026 earnings release talks about value-added services and does not name the three scores. The fiscal 2025 annual report, filed 6 November 2025, is where those scores are described as live. Consumer Authentication Service is an AI risk score for authentication, meant to reduce fraud and false declines. Advanced Authorization uses machine learning to identify fraud on Visa card payments. Provisioning Intelligence is an AI risk score meant to stop a token from being set up fraudulently. That report also records the Featurespace purchase. Employee generative-AI tools in the same filing do not count. Neither filing gives a fraud rate or an approval rate on the scores. On 20 October 2026 Visa, BILL, and Billtrust said a registered agent had started paying supplier invoices in a live pilot, using a tokenized credential and the buyer's payment limits. Visa says participating customers are processing those payments. It gives no count and no fraud rate. The pilot does not move the 9. On the 27 October 2026 call the company did not give a fraud-capture or false-decline number on Consumer Authentication Service, Advanced Authorization, or Provisioning Intelligence. It said Advanced Authorization and Visa Decision Manager are being enhanced with a large transaction model. It signed a $2.4 billion cash agreement to buy BioCatch, a behavioral fraud product covering 1.8 billion devices, 760 million users, and more than 100 large banks. Close is expected no later than fiscal second quarter 2027 and is still in regulatory approval. The deal is not a result on the three scores.

Why the score

• Reach 3 not 2: authentication, authorization, and token setup are three decision points.
• Reach 3 not 4: they sit in one risk suite. The October 2026 invoice pilot is one commercial path, not a fourth scored function. Employee generative AI does not add a function.
• Weight 3 not 2: the scores sit inside authentication and authorization. A rival network without them is weaker in the payment decision.
• Weight 3 not 4: the network is still how Visa goes to market. The filing does not show the scores running the book.
• Proof 3 not 4: the products are named and described as real-time in the annual report. No result on the scores themselves. The June 2026 quarter does not restate them, and it does not retire them. The October pilot gives no result on those scores.

What would move it next quarter: A fraud-capture or false-decline number on one of the three scores. BioCatch closing, by itself, will not move this.

Open these first

Evidence