India's DPI stack in 2026: a builder's guide to what's production-ready

The phrase 'India Stack' gets used as if it's a single thing. It's not. Five distinct infrastructure layers (Aadhaar, UPI, Account Aggregator, ONDC, and OCEN) were built by different regulatory bodies, launched years apart, with dramatically different adoption curves and integration realities.

Most coverage treats them as a unified triumph. That framing is partly right and partly misleading. UPI and Aadhaar eKYC are genuinely production-proven infrastructure. Account Aggregator has enormous potential but sits at an awkward activation gap. ONDC is real but still finding product-market fit at the network layer. OCEN is the most ambitious and the farthest from production-ready for most builders today.

If you're building in India in 2026, the question isn't 'should I use India Stack?' It's 'which layers are actually ready, and what is the honest integration complexity of each?'

Aadhaar: reliable, but not invincible

Aadhaar has three sub-services that matter for product builders:

• eKYC: verify a user's identity (name, address, date of birth, photo) with OTP consent, in seconds
• eSign: allow a user to sign a document using their Aadhaar OTP, producing a legally recognised electronic signature under Section 3A of the IT Act
• OTP Auth: simple authentication against Aadhaar without fetching KYC data

eKYC is the most-used of the three. It's what most fintechs use for account opening, what HR platforms use for employee onboarding, and what lenders use for loan origination. The alternative is video KYC or physical document upload. Both are slower and more error-prone. For Indian-market products that need identity verification at scale, Aadhaar eKYC is the clear choice.

eSign is similarly mature. A growing number of signing platforms (FlowVerify among them) offer it as the primary signing method for high-trust, India-specific workflows. It produces a signed artifact that includes the signing certificate and a full audit trail, giving stronger provenance than an OTP-over-email signature.

Integration path: UIDAI directly (requires licensed AUA status), or through a certified Authentication User Agency or eSign service provider. Most startups go through an intermediary.

UPI: the layer that disappeared into the background

UPI is the infrastructure win that's so complete it barely needs explaining. In January 2026, NPCI processed 21.7 billion UPI transactions worth ₹28.33 lakh crore. That's not a pilot — it's the default payment method for most of urban and semi-urban India.

For builders, UPI matters in three specific ways:

1. Payment collection: accept UPI payments via a payment gateway or aggregator. Table-stakes for any Indian consumer product.
2. UPI AutoPay: recurring mandates for insurance, SaaS subscriptions, and EMI flows. Had friction in earlier years; now mature enough for production use.
3. UPI for B2B: corporate UPI and enterprise settlement rails. Less discussed but increasingly real for mid-market invoice and payout flows.

The constraint: UPI access for most startups is through intermediaries (Razorpay, Cashfree, PhonePe for Business, and similar). If you're building a payment product, not just accepting payments, the path to becoming a certified Payment Service Provider carries significant regulatory overhead. Most startups should not attempt it. Use a PSP.

Integration path: NPCI through certified PSPs. No direct integration for startups.

Account Aggregator: powerful concept, incomplete activation

The Account Aggregator framework is the India Stack layer that gets the most talk and the least actual adoption in product decisions. That is slowly changing, but the activation gap is real and worth understanding before you build on it.

What it does: AA lets a user share verified financial data (bank statements, insurance policy details, tax information, investment portfolios) with any licensed Financial Information User (FIU), without uploading PDFs or sharing login credentials. The user consents through an AA app (Setu, Finvu, or Onemoney), and data flows directly from the Financial Information Provider to the FIU.

The numbers look impressive: as of March 2026, 2.88 billion financial accounts are enabled for data sharing on AA, and 410 entities are registered as FIUs. But here's the honest constraint: approximately 38% of borrowers have AA-enabled accounts. That means 62% of your potential users may not have the flow set up, and you still need a PDF fallback for them.

The framework is genuinely powerful for fintech — lending decisions that used to take three days of document chasing can happen in minutes. But for non-fintech builders, the question is whether you're actually handling financial data flows. If you're not in lending, insurance, or financial planning, you likely don't need AA today.

Integration path: Sahamati manages the AA ecosystem. Integration requires a licensed AA service provider such as Setu, FinBox, or Onemoney.

ONDC: real orders, fragmented network

ONDC is the most visible and most misunderstood of the newer layers. The network has genuine scale: over 1.16 lakh retail sellers across 630+ cities and 154 million cumulative orders as of late 2025. Real transactions are happening at scale.

The complication is the network model. ONDC is not a marketplace. It's a protocol that lets buyers on one platform discover and transact with sellers on another. A buyer on Paytm can order from a restaurant listed on Magicpin. The seller's onboarding, logistics, and dispute resolution happen across multiple network participants.

This creates a fragmentation problem at the product layer. Buyer experience can be inconsistent depending on which Buyer Network Participant (BNP) they're using. Seller onboarding varies across Seller Network Participants (SNPs). Logistics coordination across multiple participants introduces failure modes that wouldn't exist in a centralised platform.

For D2C brands, ONDC is increasingly worth a presence. The distribution reach is real and unit economics can be better than large marketplace fee structures. If you're building a network participant, the ONDC protocol has matured significantly in the last 18 months and technical documentation is substantially better than it was in 2022.

Integration path: ONDC network directly, typically through a certified SNP or BNP.

OCEN: the most ambitious promise, the most patience required

OCEN (Open Credit Enablement Network) is the DPI layer for credit: a protocol that lets any platform (a GST portal, an accounting app, a supply chain tool) originate loan requests and route them to licensed lenders, without being a lender itself. The vision is compelling: contextual credit, offered at the moment of need, using data from the platform context.

The honest 2026 picture: OCEN has working implementations, but adoption is thin compared to the scale of the problem it's trying to solve. The lenders on the network are primarily NBFCs and smaller digital lenders; large banks are present but cautious. Loan origination via OCEN is happening, but volumes are not yet comparable to traditional lending flows.

The bottleneck is trust and data quality. Lenders on the network need confidence in the quality of applications coming through Loan Service Providers (LSPs). When that confidence builds, partly through better underwriting data from Account Aggregator, OCEN has a real shot at becoming the credit equivalent of UPI. It isn't there yet.

If you're building a platform with a large SMB user base and thinking about embedded lending as a feature, OCEN is worth watching closely. If you need it to work reliably at volume for a core product feature today, the dependency on network participant depth is a real production risk.

Integration path: iSpirt runs the OCEN working groups; implementation is through NBFC partnerships or tech providers on the network.

Which layers to build on in 2026

The five layers are not equally ready. Here's a practical decision map:

The India Stack's promise — that public digital infrastructure removes the cost of building and maintaining private identity, payment, and credit rails — is largely delivered for identity and payments. For commerce and credit, 2026 is the point where you can start building seriously, not where you should assume everything just works.

The practical framing: UPI and Aadhaar eKYC are table stakes. Account Aggregator is worth building toward now if you're in fintech, with a clear fallback strategy. ONDC is worth a presence if you're in commerce. OCEN is worth monitoring, not depending on for core product flows. The gap between 'announced' and 'production-ready at scale' remains meaningful for the newer layers, and that gap is closing faster than most outside observers realise.