Razorpay launches Vulcan AI model to lift payment success and fraud detection
India-focused fintech Razorpay has launched Vulcan, a proprietary payments foundation model trained on 4 billion transactions. The company says early deployments improved payment success rates by 8–10%, increased preferred UPI-app visibility and strengthened international-card fraud detection.
What happened
Razorpay launched Vulcan, a proprietary AI payments foundation model trained on 4 billion transactions to improve payment routing, UPI checkout visibility and
Key facts
- ~3 trillion data points
- 4 billion payments
- ~3,000 signals per transaction
- 8-10% improvement in payment success rates
- 8x increase in international card fraud detection
- 5x rise in fraudulent/disputed transactions identified
- 40% more shoppers shown their preferred UPI app
- 1-2 lakh additional purchases per month
- ~200 businesses using its AI offerings
- $600-700 million expected IPO size
- $5-6 billion expected IPO valuation
- $7.5 billion last private valuation
- over $800 million raised since inception
Why this matters
Razorpay’s payments AI push makes it a more strategic partner or competitor for fintechs seeking India-scale checkout optimisation, UPI visibility and fraud capabilities.
What to watch
- Independently verified uplift in payment success rates, net of incentives and traffic-mix changes.
- Merchant adoption rates and evidence of payment-volume consolidation onto Razorpay.
- UPI ecosystem or NPCI guidance on app visibility, routing, data use and consumer choice.
- Changes in card fraud rates, false-positive declines, chargebacks and cross-border approval rates.
- Pricing announcements for AI-powered payment optimisation and competitive launches from Indian PSPs, banks or global gateways.
- Package Vulcan performance gains into premium enterprise payment-optimisation tiers with outcome-linked pricing.
- Expand merchant-facing controls that explain routing, retry and fraud decisions to reduce regulatory and partner concerns.
- Use transaction intelligence to bundle fraud prevention, tokenisation, reconciliation and cross-border acceptance.
- Target high-failure categories such as travel, digital goods, subscriptions and high-value retail with vertical-specific models.
- Competitors and banks are likely to accelerate proprietary payment AI, seek alternative data partnerships, or compete on transparent routing guarantees.
Also reported by
- Inc42 · Buzz — Same time