ONDC 2.0: The Real Fight Is Over Who Owns Commerce Data

The cheque that was never about the cheque

On May 12, 2026, a small company in New Delhi held a board meeting and allotted 2.2 crore equity shares at a face value of Rs 100 each.

The buyers were not venture funds. They were Zoho, Uber, Paytm's parent One97 Communications, and BSE, the technology arm of the Bombay Stock Exchange. Together they put in Rs 220 crore, roughly 23 million dollars. It was the first time strategic corporates took equity in ONDC, India's government-backed Open Network for Digital Commerce.

The headlines called it the next chapter for "the UPI of ecommerce." That framing is comfortable and mostly wrong.

These companies did not write cheques to ride a transaction wave. They wrote cheques to sit at the table where the rules for commerce data get written. The money is small. The seat is not.

What ONDC actually is, for anyone outside India

Picture Amazon. One company runs the app you shop in, the listings you browse, the search that ranks them, the ads that jump the queue, the payment, and the logistics. It is one stack, owned end to end. That bundle is the business.

ONDC takes that bundle apart.

It is an open network, backed by the Indian government, that lets any buyer app discover any seller registered on any seller app. You could search inside Paytm, find a shop that lists through a completely different seller app, and buy from it. A standard messaging protocol called Beckn carries the search, the offer, and the order between the apps. A gateway in the middle broadcasts your request and collects the replies.

Nobody owns the whole pipe. ONDC the company does not run a store, does not rank products, and charges no central commission on a sale. It runs the plumbing and the registry that says who is allowed to connect.

So the thing Amazon sells as one product, ONDC splits into parts that anyone can supply: the buyer app, the seller app, the catalog, the logistics. That split is the entire point. And it is also where the real argument starts.

The moat was never the storefront. It was the data.

Here is the part the "next UPI" story skips.

When one company owns the rails, it owns three things that compound. It owns the demand signal: every search, every add-to-cart, every abandoned basket. It owns the pricing data: what sold, at what price, against which rival. And it owns the ad inventory: the right to charge sellers to be seen by shoppers the platform already understands.

That last one is not a side business. It is the business.

Look at the numbers. Flipkart's advertising income reached about Rs 6,317 crore (around 716 million dollars), up 27 percent year over year, and advertising is now roughly 31 percent of Flipkart's total revenue. Amazon India reportedly pulled in about Rs 8,342 crore from ads in the same window. Globally, Amazon's ad business runs north of 90 billion dollars a year, and it holds close to 80 percent of the United States retail media market.

None of that comes from selling products. It comes from owning the shopper graph and renting access to it. A data moat is exactly this: a dataset so specific and so hard to rebuild that it keeps improving your product while locking everyone else out. Cross-category purchase behavior across thousands of sellers is the cleanest example there is.

ONDC's design attacks that moat at the root. Unbundle the stack, and first-party commerce data stops collecting in one company's warehouse.

Walled garden versus open network: follow the data

The difference is easiest to see side by side.

EY India put the architecture plainly: business data is available across the ecosystem instead of concentrating with one party, through an anonymized open-data framework. As ONDC's own framing goes, the network itself "does not get any data." Individual transaction records stay with the intermediaries. Only aggregated, anonymized signals flow into the open layer everyone can see.

Read that table top to bottom. Every row that used to belong to one company is now contested. That is the fight the May round bought into.

Why a Zoho or a BSE actually pays in

Now the equity makes sense.

These investors are not buying a slice of 218 million transactions. They are buying influence over the standards that decide how an open commerce network stores, shares, and anonymizes data. When the rails are open, the value moves up a layer, to whoever shapes the catalog format, the data schema, and the tooling sellers depend on. That layer is what the cheques are really for.

The stated rationales fit that read:

This is tranche one. ONDC is raising Rs 430 crore in total, with the rest expected from State Bank of India, NSE, Punjab National Bank, Amul, the dairy cooperative, the farmer-cooperative body NACFMIL, and the Quality Council of India. The money is earmarked for AI-led and agentic commerce, a national cataloging effort called DigiCatalog, and quality assurance, all under the banner of "ONDC 2.0."

Vibhor Jain, who became managing director and CEO in April 2026 after T Koshy stepped down, framed the phase as sharpening focus on "removing systemic market frictions that limit adoption of digital rails." The roadmap even names a sovereign commerce model, Vyapar LM, meant to let a shopper buy in their own language through an agent.

A buyer app investing in the agent layer is investing in the next place demand will be captured. Same instinct as Amazon Ads. Different address.

The contrarian kicker: open rails make intelligence harder, not optional

Here is the catch nobody on the celebration side wants to say out loud.

An open network does not remove the need for commerce intelligence. It explodes it.

When Amazon owned everything, a seller had one dashboard. Painful, expensive, controlled by the platform, but singular. You logged into one place and saw your demand, your rank, your ad spend, your competitors' prices.

ONDC deletes that single pane of glass. Your products now move across more than 100 buyer apps. The network spans about 7.64 lakh sellers and service providers across 616 cities, and it handled roughly 21.8 crore (218 million) transactions in FY26 across retail, mobility, logistics, and financial services. There is no master view of any of it. Each leg of a single order can touch a different buyer app, a different seller app, and a different logistics provider, and the data sits with each of them.

Unbundling solved a monopoly problem and created an aggregation problem.

That fragmentation already bites. As ONDC trimmed its participant incentives from about Rs 3 crore a month to around Rs 30 lakh, retail order momentum cooled, and reporting through early 2026 noted uneven adoption, local-merchant friction, and order volumes below the hype. Sellers complain they cannot stand out without the ad boosts the walled gardens sell, and that managing orders across disconnected apps is its own tax. Open does not mean easy.

For a brand, the strategic question flips. In a walled garden you ask how to win the platform's algorithm. On an open network you ask a harder one: how do I see myself at all, when my demand signal is smeared across a hundred apps and no one is required to hand it back? That is precisely the gap a commerce intelligence layer exists to close, the work behind every BrandBaazar solution. The network gives you reach. It does not give you sight.

What to watch as ONDC 2.0 lands

The equity round is not the story. It is the signal.

Watch who shapes DigiCatalog, because the catalog standard quietly decides whose data format becomes the default. Watch whether the anonymized open-data layer ships with real, queryable depth, or stays a brochure. Watch whether agentic buying through something like Vyapar LM concentrates demand back into a few smart assistants, rebuilding the very gatekeeper ONDC was built to break.

India is running a live experiment the rest of the world is watching. Can you keep open rails and still get usable intelligence out of them. Can you separate the network from the data owner without leaving sellers blind.

The cheques from Zoho, Uber, Paytm, and BSE were not bets on transactions. They were bets on owning the answer.

When the rails go open, the people who win are not the ones who move the most goods. They are the ones who can still read the map.