Razorpay

Fintech · Payments · AI

Launch

Razorpay launches Vulcan, India's first AI payments foundation model

The launch positions the fintech against global rival Stripe's own payments model, arriving as it preps a lower-valued IPO in India.

• Razorpay launched Vulcan, a transformer-based AI foundation model for payments, built with NVIDIA and AWS and trained on roughly 3 trillion data points across 4 billion transactions.
• Razorpay is a Bengaluru-based payments gateway that powers checkout, routing and fraud detection for merchants and banks across India's digital economy.
• Unlike narrow machine-learning models built for one task, Vulcan learns from roughly 3,000 signals per transaction to score routing, fraud and risk across the entire payments ecosystem at once.
• Early deployment across 1.5 million transactions and more than 51,000 businesses lifted payment success rates 8-10%, detected 8x more international card fraud and identified 5x more disputed transactions.
• On Razorpay's Magic Checkout, the model helped 40% more shoppers see their preferred UPI app, adding an estimated 100,000-200,000 extra purchases a month.
• The launch lands as Razorpay confidentially filed for an IPO in June 2026, with a public offer expected to value it at $5-6B, below its last private valuation of $7.5B.
• Global payments processor Stripe already runs its own foundation model trained on tens of billions of transactions, showing AI-driven routing and fraud tools are becoming standard infrastructure at scale.
• As UPI and other rails push India's transaction volumes past 24,000 crore a year, foundation models that generalize across fraud, routing and checkout could become essential infrastructure rather than a competitive edge.