The news: Mastercard is launching its own generative AI model largely aimed at thwarting fraud, per a press release. Nvidia’s computing platform will help power this model.

How it works: Mastercard’s large tabular model (LTM) is trained on structured data—large-scale tables and datasets—unlike large language models (LLMs) like Claude and ChatGPT, which were built through unstructured data. 

Why this matters: Mastercard will feed its LTM billions of anonymized payments data, merchant location data, fraud data, and other data sets to deliver predictive insights about future transactions. 

Mastercard is building out its value-added services (VAS), as net revenues from VAS have grown more than twice as fast as net payments revenues in its most recent Q4 2025 earnings.

This could help Mastercard offset losses from Capital One shifting its card program onto the Discover network: Mastercard’s US debut and prepaid cardholder purchase volume growth plummeted from 10.8% to 2.3% YoY, per its most recent earnings, and Capital One’s staggered credit migration has yet to fully impact the network.

Implications for payment networks: While Mastercard’s LTM launch initially targeted fraud, wider use cases for AI beckon, especially for targeted ads. 

Mastercard’s new financial media network could turn this LTM into a gold mine of insights for marketers and merchants. This dovetails with our forecast for FMN ad spending, which we anticipate will reach $1.78 billion by 2027. With Visa yet to launch its own FMN, this also gives Mastercard an opportunity to power past its largest rival.