Razorpay launches Vulcan AI payments model, citing payment and fraud gains at Blinkit

Built with Nvidia and AWS, Razorpay’s India-hosted payments foundation model is being deployed across merchant transactions. The company says Blinkit saw a 10% payment-success lift, while Vulcan also improved fraud detection and UPI-app access.

— Source published Tue, 18 Aug, 2026, 07:00 IST · First seen Tue, 18 Aug, 2026, 07:30 IST · Source Hindustan Times · Business

What happened

Razorpay launched Vulcan, an India-hosted payments foundation model built with Nvidia and AWS. Blinkit and other merchants report higher payment success and

Key facts

  • 23 billion UPI transactions per month
  • 590 million credit-card payments per month
  • 3 trillion training data points
  • 4 billion digital payments
  • 1.5 million tested transactions
  • 50,000 merchants
  • 8-10% improvement in payment success rates
  • 5x reduction in fraud
  • 10% payment-success improvement for Blinkit
  • 8x more international-card fraud detections
  • 40% more shoppers accessing preferred UPI app
  • 200,000 additional purchases per month

Why this matters

Razorpay’s Nvidia- and AWS-backed, India-hosted payments model makes it a potential partner for transaction optimisation, fraud intelligence and embedded-credit distribution.

What to watch

  • Independent merchant results confirming sustained payment-success gains beyond Blinkit.
  • Razorpay disclosure of paid Vulcan adoption, pricing structure, processed GMV or merchant retention impact.
  • Launch of AI-assisted underwriting, credit or risk-scoring products and associated loss-rate data.
  • UPI ecosystem changes affecting app access, transaction routing or third-party payment-app rules.
  • Competitive AI-payment announcements from PayU, Cashfree, PhonePe, Juspay, banks or global PSPs.
  • Regulatory scrutiny of automated fraud decisions, lending models, consent and India-hosted payments data.
  • Package Vulcan as a premium conversion-and-risk module with merchant-specific ROI reporting.
  • Prioritise pilots with quick-commerce, food delivery, travel, ticketing and high-failure-rate subscription merchants.
  • Use model outputs to offer selective working-capital and merchant-credit products, initially to proven low-loss cohorts.
  • Secure UPI, issuer-bank and network data partnerships to improve routing and fraud signal coverage.
  • Strengthen model governance, explainability and data-localisation controls before broader underwriting deployment.