NextFin News - In payments, every extra security check costs conversions, and every removed check invites fraud. That trade-off - fraud versus friction - is the single most important engineering problem in the industry, and it is the lens through which Sheetal Lalwani, co-founder and chief operating officer of Indian payments infrastructure firm Juspay, views the entire sector. As real-time payments scale globally and AI-driven fraud accelerates, the question is no longer whether to choose security or speed. It is whether a company can keep moving the frontier outward on both at once.
Juspay's own trajectory is a test case. Founded in Bangalore in 2012, the company now processes more than 300 million payment transactions a day in India for merchants including Amazon, Swiggy, Google, Microsoft, Ola and HSBC. In January 2026 it closed a $50 million follow-on funding round that valued the business at roughly $1.2 billion - the second liquidity event for early investors and employees within a year. Operationally, Juspay reported record fiscal 2025 results: revenue of 514 crore rupees, about $61 million, up 61% year on year, and profit after tax of 62 crore rupees, about $14 million, turning a 97.5 crore rupee loss in the prior year into one of the few profitable payments-infrastructure businesses operating at genuine global scale.
The central judgment here is that payments is not a winner-take-all race for transaction volume. It is a persistent optimisation problem, and the economic prize sits not on the payment rail itself but in the infrastructure layer that tunes the fraud-friction balance for each merchant, each transaction type, and each market. Juspay's early, unpaid bet on India's Unified Payments Interface - four years of building with no monetisation when UPI was still "a concept in a Word document," in the company's telling - is the clearest evidence that the winners in this industry are the players willing to commit to foundational work before the payoff is visible.
The Mechanism: Why Fraud and Friction Can Never Be Solved, Only Balanced
The mechanism is deceptively simple and brutally consequential. A payment that is too easy to make is too easy to steal. A payment that is too hard to steal is too hard to make. Every authentication step - a one-time password, a biometric check, a two-factor prompt - reduces fraud losses but also increases the probability that a legitimate customer abandons the checkout. Every removed step does the reverse. The optimal point is not fixed; it shifts with the transaction value, the merchant's risk appetite, the device, the geography, and the sophistication of the fraudster.
India's 2011 mandate for two-factor authentication on all digital payments crystallised the problem. Card details plus an OTP sent to the customer's phone made transactions materially more secure - and materially more cumbersome. While much of the world was still focused on removing friction, India had deliberately added it. Juspay's founding thesis was that the two goals were not mutually exclusive: "We believed there was a way to do both: keep payments secure, but also make them easy to use," Lalwani has said. The company's early products focused on making two-factor authentication feel seamless - the first practical expression of a philosophy that still shapes its roadmap.
This is why the fraud-friction frontier is a structural feature of the industry, not a cyclical problem that resolves. Each technological shift - real-time rails, open banking, tokenisation, AI agents initiating payments - moves the frontier outward rather than settling it. Fraud adapts: synthetic identities, account takeover, authorised push-payment scams. Defences adapt in turn: behavioural biometrics, device fingerprinting, risk-based authentication that steps up only when a transaction looks anomalous. The equilibrium is a permanent, escalating optimisation loop. What is cyclical is the loss rate around specific adoption waves - UPI credit, cross-border instant payments, agent-initiated transactions - which spike as new channels open and then settle as rules and models mature. The underlying tension does not mean-revert.
"We ended up investing four years into building infrastructure for UPI with no monetisation, at a time when its future wasn't certain."
The practical implication for any payments business is that scale alone is not an advantage unless it feeds the optimisation engine. More transactions generate more labelled data; more data improves fraud models; better models allow more legitimate transactions through with fewer checks; higher approval rates attract more merchants. That is the data network effect that separates infrastructure platforms from simple gateways. Juspay's claim that it now touches nearly every Indian making a digital payment - "Around eight years ago, we set a goal internally: that every Indian making a digital payment would touch one of our products in some way. Today, that's more or less the case" - is significant not as a vanity metric but as a data-advantage statement.
The UPI Bet: Committing Before the Payoff Was Visible
Juspay's involvement with UPI illustrates the structural nature of the commitment required. When founder Vimal Kumar first encountered the design for India's national real-time payments system, UPI was not a product. It was a concept document. The company nonetheless invested four years building authentication-layer infrastructure for a system whose commercial future was unproven and whose economics - India's zero merchant-discount-rate framework for UPI - would explicitly forbid monetising the rail itself.
That decision reads differently in hindsight than it did at the time. Today UPI is one of the largest real-time payment systems in the world. National Payments Corporation of India data reported by India's Press Information Bureau shows the system processed 241.6 billion transactions worth 314 lakh crore rupees in fiscal 2025-26 - a 12,000-fold volume increase since its April 2016 launch, when just 21 banks were live on the network. Average daily volume reached 660 million transactions in 2025, and the average ticket size was roughly 1,300 rupees. Volume grew 30% year on year while value grew 21%, evidence that the network has migrated decisively toward routine, small-value retail spending. Research firm IMARC projects the India UPI market to expand from about $24.5 billion in 2025 to more than $600 billion by 2034.
The lesson is not that betting on unproven public infrastructure is always right. It is that in payments, the durable value accrues to the layer that becomes indispensable to the rail's operation - and that layer is rarely the rail's owner. NPCI operates UPI; Juspay built parts of the authentication plumbing that makes it usable at scale. The rail is a public good with capped economics. The infrastructure wrapped around it - routing, orchestration, compliance, fraud tooling, analytics - is where pricing power survives.
Where the Money Actually Is: Infrastructure, Not the Rail
This is the second-order point that volume headlines obscure. The market's instinct is to track transaction counts and gross payment value. But in a zero-MDR environment, moving money generates little to no direct revenue. The economics live in the value-added services layered on top: smart routing across providers, checkout design that lifts conversion, reconciliation, compliance tooling, and fraud scoring that lets merchants accept more transactions without accepting more losses. Juspay's financials are consistent with that model. At 514 crore rupees of revenue on more than 300 million daily transactions, the company is not monetising volume per se; it is monetising the complexity of managing that volume for large merchants who would otherwise have to build the capability in-house.
"In many cases, we effectively become the merchant's payments team, so they don't need to build that capability in-house," Lalwani has said. That positioning - infrastructure-as-a-service rather than payment facilitator - is what allows profitability at a scale where a pure processor would be trapped by per-transaction margins too thin to cover engineering. The company's annualised total payment volume crossed $1 trillion in fiscal 2025, up 150% from $400 billion a year earlier - yet revenue of $61 million is a tiny fraction of that flow. The spread between what moves across the network and what the network pays the infrastructure layer is precisely where the fraud-friction optimisation business lives.
The company's international expansion follows the same logic. Juspay now has teams in Singapore, Dublin, São Paulo, San Mateo and Dubai, and its open-source orchestration platform, Hyperswitch, is the vehicle for reaching merchants who want control over their payments stack without vendor lock-in. Open source is a deliberate wedge: it lowers adoption friction for engineers evaluating alternatives to Stripe, Adyen or in-house builds, and it converts that evaluation into a managed-services relationship once the platform is embedded. The priority geographies - the EU, UK, US, Asia-Pacific, the Middle East and Latin America - are precisely the markets where real-time rails are either maturing or being built, and where the fraud-friction problem is newest and most acute. In Brazil, for example, the company is developing biometric payment solutions for the Pix ecosystem.
The January 2026 round - $50 million led by WestBridge Capital, split between primary and secondary components - was the second such liquidity event in twelve months, following an April 2025 Series D that raised $60 million at the same $1.2 billion valuation with Kedaara Capital leading and SoftBank Vision Fund and Accel participating. A valuation at that level implies investors are underwriting not just India's UPI growth but the thesis that Juspay's optimisation layer can be exported. The company employs more than 1,500 people and describes itself as scaling while remaining profitable - a combination that remains uncommon in payments infrastructure. For fiscal 2026, management has said it will focus on "agentic commerce" - context-aware, end-to-end purchase flows in AI-native environments - alongside authentication tools such as passkeys and biometrics.
The Counter-Case: Middleware Is a Commodity in Waiting
The strongest argument against this thesis is that payment orchestration is being commoditised, and fast. Stripe, Adyen, PayPal and the card networks are all building orchestration and routing capabilities in-house; for a large enough merchant, building internally is a credible option. Hyperswitch's open-source licence invites fork-and-run competition: any sophisticated engineering team can take the code, customise it, and bypass Juspay entirely. In a world of zero-MDR rails and price-sensitive merchants, middleware risks becoming a feature bundled into a broader stack rather than a standalone product with pricing power. If that happens, Juspay's valuation multiple cannot hold, no matter how elegant its fraud-friction engineering.
The counter-thesis has force, and it reflects the broader trajectory of financial infrastructure: layers that start as differentiated products tend to become commoditised utilities as standards mature. But it underestimates two sources of durability. First, localisation depth. India's payments environment - mandatory two-factor authentication, UPI's idiosyncratic protocols, the interplay between cards, wallets and bank-led rails - is not something a global player can replicate cheaply from a distance. The fraud-friction optimum is locally specific, and it is learned through operating at scale in that market, not through reading documentation. Second, the product is moving up the stack. Orchestration alone is vulnerable; orchestration plus proprietary routing intelligence, fraud scoring, reconciliation and analytics is a stickier, harder-to-replace proposition. The open-source layer is the top of the funnel, not the whole product.
The signal that would prove the counter-thesis right - and the core bullish view wrong - is observable: if Juspay's revenue growth falls below 20% year on year for two consecutive years while headcount continues to rise, that would indicate commoditisation pressure is compressing pricing power faster than scale can offset it. A second, equally telling signal would be the loss of a marquee client to an in-house or competitor build. Either would suggest the optimisation layer is becoming a feature, not a platform.
What to Watch: Three Horizons, Three Scenarios
Short term (6-18 months): The near-term drivers are India-specific. UPI volume growth, the rollout of credit-on-UPI, and festival-season throughput will determine how much incremental value the authentication and orchestration layer captures. A base case has UPI volumes continuing to grow at 25-30% annually, keeping Juspay's India business on its current trajectory. The downside is a fraud spike on new UPI credit products that forces regulators to add friction - which would raise the value of Juspay's optimisation tooling even as it slows raw volume. The upside is faster-than-expected cross-border UPI linkage, which multiplies the complexity - and the need for infrastructure - without changing the underlying rail economics.
Medium term (2-4 years): The swing factor is Hyperswitch's international adoption. Success looks like a meaningful share of non-India revenue, sustained profitability, and gross margins that hold as the business scales. Failure looks like open-source adoption that never converts to paid relationships, leaving the company competing on price against well-capitalised global orchestration providers. This is the horizon on which the $1.2 billion valuation is actually decided.
Long term (5+ years): The structural question is whether AI-initiated payments and always-on real-time rails raise the stakes on the fraud side faster than defences can respond. If they do, the fraud-friction optimisation layer becomes more valuable, not less - because the cost of a wrong decision rises with transaction speed and autonomy. If, instead, authentication becomes invisible and near-perfect, the optimisation problem shrinks and the layer's pricing power erodes with it. The base case is the former: fraud is an adaptive adversary, and invisible authentication is a moving target, not a destination.
For investors and operators, the asymmetry is clear. The exposed are the pure-play processors and gateways competing on per-transaction price in markets where the rail itself is free. The beneficiaries are the infrastructure layers that can demonstrably move the fraud-friction frontier - accepting more good transactions while losing less to fraud - and can prove it with data. Juspay's profitability at scale is the early evidence that such a position is attainable.
The payments industry does not reward choosing security over speed, or speed over security. It rewards the companies that refuse to choose - and that treat the tension between the two as a permanent engineering discipline rather than a problem to be solved. Juspay's four-year unpaid bet on UPI, and its profitability a decade later, suggest the market is finally pricing that discipline accordingly. The fraud-friction frontier will keep moving. The question is which layer captures the value of moving it.
Explore more exclusive insights at nextfin.ai.
