Razorpay has launched Vulcan, which it calls India’s first transformer based AI foundation model built specifically for payments. The goal is to make every digital payment in the country more reliable, safer and predictable. Built with NVIDIA and AWS, the model combines Razorpay’s payments data with NVIDIA’s accelerated computing and AWS’s cloud infrastructure, groundwork the company says India’s e commerce market will need as it heads toward a projected $350 billion by 2030.
An internal Razorpay study across 1.5 million shoppers and more than 51,000 businesses found the same payment friction, failed transactions, drop offs and delays, showing up identically from metro high streets to small town markets. That consistency is what convinced Razorpay to build one shared model instead of refining separate systems for each problem.
Ahead of the full launch, early components of the model were already running across 3 trillion data points on Razorpay’s network, testing routing, fraud and risk decisions on live transactions. Customers including Blinkit, Bachatt and redBus have started seeing results: an 8 to 10% improvement in payment success rates, 8 times more international card fraud detected and stopped, 5 times more fraudulent or disputed transactions identified without increasing alert volumes, and 40% more shoppers seeing their preferred UPI app on Razorpay’s Magic Checkout, helping complete 1 to 2 lakh more purchases every month.
India’s payments landscape is unusually fragmented. A single purchase can move through UPI, cards, net banking, wallets or cash on delivery, across hundreds of banks and gateways. Razorpay illustrates the problem with a hypothetical shopper, Meera, who tries to pay Rs 2,400 for running shoes at 9 pm and sees “Payment unsuccessful. Please try again,” even though her card, bank and funds are all fine. Her payment simply had several possible routes, and one happened to be briefly unavailable at that moment. That pattern, repeated across millions of transactions, led Razorpay to build a model that scores every route in real time and picks the healthiest one before a payment is even attempted.
Until now, the industry has tackled this with separate, specialised models for routing, fraud, risk and checkout that don’t talk to each other, even though many of the same signals matter to all of them. Razorpay compares it to several doctors examining a patient, each reading only their own test results. Vulcan instead learns from the entire payments ecosystem at once and keeps improving with every transaction it processes, rather than solving one narrow problem at a time. It’s built on transformer architecture, the same family behind large language models, but adapted specifically to the patterns hidden inside Indian payments data.
Razorpay is careful to draw a distinction here. Vulcan isn’t a traditional machine learning model, which is built for one job and needs retraining for anything new, and it isn’t an LLM either. A foundation model learns how payments move, so that understanding extends to new use cases without starting over. The foundation model concept comes from LLMs, but LLMs understand text. Vulcan understands the movement of money.
By the numbers, Vulcan is trained on roughly 3 trillion data points across 4 billion payments and learns from about 3,000 signals per transaction. Razorpay says both the architecture and the training data are proprietary and built entirely in house. Training at this scale demanded serious computational infrastructure. NVIDIA’s GPUs powered training and live decisioning, while AWS’s cloud infrastructure, including Amazon SageMaker, supported development, training and deployment.
Harshil Mathur, CEO and Founder of Razorpay, said: “India’s appetite for digital payments is real, but it isn’t universal yet, for a large part of the country, going digital still comes down to one thing: does it work, every single time? That’s the customer we built this for: the one still deciding whether to trust a screen over cash in hand. An AI led payments foundation model doesn’t just solve today’s problem and stop there. Every payment teaches the system something that makes the next payment better. That’s what makes this feel less like a product launch, and more like the starting point for how payments in India keep getting better on their own, for years to come.”
Pahal Patangia, Head of Global Industry Business Development and Payments at NVIDIA, said: “India’s rapidly evolving digital economy is creating an opportunity to make payments more intelligent, reliable, and secure. NVIDIA’s work with Razorpay in partnership with AWS on AI payments foundation models has opened up a new frontier, turning complex payments data into real time contextual intelligence.”
Kiran Jagannath, Head of FSI and Conglomerates at AWS India and South Asia, said: “Razorpay is reimagining payments intelligence at India scale with an AI Foundation Model, built on Amazon SageMaker, that consolidates billions of transaction insights into a single, continuously learning intelligence layer, replacing fragmented ML models with unified AI that delivers higher payment success rates, rapid iteration, and enterprise grade security for mission critical payment flows.”
Vulcan’s capabilities include hyper precision routing, which sends each payment down the path most likely to succeed in real time, network level fraud detection that spots stolen cards the moment they’re used across unrelated merchants, RTO risk intelligence that flags risky cash on delivery orders before checkout, and predictive checkout personalisation that recommends the payment method most likely to work for each customer.
Razorpay frames this as a starting point rather than a destination, with the goal of eventually running every payment decision, from authentication to routing to fraud to lending, through one continuously learning model.
Digital Trade Outlook: What makes Vulcan notable isn’t just the volume of data behind it, it’s the shift from siloed, single purpose models to one shared intelligence layer that learns across the entire payment lifecycle at once. In a market as fragmented as India’s, where a single transaction can route through hundreds of banks and gateways, that kind of unified visibility could do more to reduce payment failures than any single point fix, and it may set a template other large payment networks eventually need to follow.