Bengaluru, August 18, 2026: Fintech company Razorpay has launched Vulcan, an AI foundation model developed specifically for the payments ecosystem. The model is designed to make digital transactions more reliable, secure, and efficient by applying artificial intelligence across multiple stages of the payment journey.
Vulcan is a transformer-based foundation model trained on nearly 4 billion payments and around 3 trillion data points, with the ability to analyse approximately 3,000 signals per transaction. It is designed to understand payment behaviour across merchants, payment instruments, banks, and payment gateways.
The model brings together capabilities that traditionally rely on separate systems, including payment routing, fraud detection, risk assessment, and checkout personalisation. By analysing multiple signals simultaneously, Vulcan aims to support real-time decisions and improve the likelihood of successful transactions.
One of its key applications is payment routing, where the model evaluates available routes and identifies the path most likely to result in a successful transaction. It also supports fraud detection by identifying patterns across transactions and can help personalise checkout experiences based on customer and payment behaviour.
The model has been developed using NVIDIA’s accelerated computing technology and Amazon Web Services (AWS) infrastructure, with Amazon SageMaker supporting its training and deployment.
Razorpay plans to expand Vulcan’s applications across additional areas of the payments ecosystem, including authentication, fraud prevention, risk management, and other financial services.
With digital payments continuing to grow in scale and complexity, the launch of Vulcan marks Razorpay’s move towards using a unified AI layer to support payment decisions and improve the overall digital transaction experience.
