Programmable e-Rupee: 6 RBI Pilot Use Cases and What Is Live in 2026

In February 2026, ration-card holders in parts of Gujarat began receiving part of their food subsidy as digital rupees that could only be spent at a fair price shop or a designated merchant. Try to buy a phone recharge with that money and the wallet simply declines. That small refusal is the whole idea behind a programmable e-rupee, and it is the corner of India's central bank digital currency that has moved fastest in 2025-26, even as the headline circulation number went the other way.

What "programmable" means in the RBI's version

The Reserve Bank of India set out its design in the Concept Note on CBDC of October 2022. The retail digital rupee (e₹-R) was to be a token, closer to a banknote than to a bank balance, distributed through banks in a two-tier model and paying no interest. Programmability was not the headline then. It arrived in February 2024, when Governor Shaktikanta Das told the policy press conference that the pilot would add use cases built on programmability and offline functionality, as Business Standard reported.

The examples he gave were concrete. A government agency could ensure that money reaches only a defined benefit. A company could fence spending to something like employee business travel. Rules could also set a validity period, or restrict the geographic area in which the token works. In practice a programmable token carries a small rulebook: who may receive it, which merchant categories may accept it, where, and until when. When the conditions are met, the merchant receives ordinary e-rupee that it can use like any other money.

That last point matters for the economics. Deputy Governor T Rabi Shankar has argued that a restriction is temporary, saying programmability "just puts it on hold" as far as fungibility goes, a framing quoted in a Finshots explainer. The rupee in a citizen's wallet is conditional; the rupee in a shopkeeper's wallet is not.

It is easy to confuse this with other Indian tools. e-RUPI, for instance, is a voucher that runs over the UPI rails and is issued by a bank. A programmable e₹ is central bank money with a rule attached. We cover the voucher route separately and the broader concept in purpose-bound money explained.

From banknote to rulebook: how the pilot got here

The wholesale pilot (e₹-W) started on 1 November 2022 with nine banks settling secondary-market government securities trades. The retail pilot followed on 1 December 2022 in closed user groups. By December 2023 the retail pilot had hit a target of one million transactions a day, a number the RBI and the pilot banks had set out to reach.

Programmability and offline payments were added during 2024. At the RBI@90 conference in August 2024, the Governor described a pilot of more than 5 million users and 16 banks, and named two farm-sector use cases that we return to below, according to a Digital Pound Foundation summary of the event. By March 2025, the RBI's annual report counted 17 banks and about 6 million users, and non-bank firms such as MobiKwik and CRED had been allowed to run CBDC wallets, as MediaNama's digest of the report noted. Reported user numbers reached roughly 7 million across 19 banks by October 2025.

Then came the awkward number. The RBI Annual Report 2025-26, released at the end of May 2026, put CBDC-Retail in circulation at ₹771.66 crore on 31 March 2026, down about 24% from ₹1,016.46 crore a year before, as Outlook Money reported. In the same report, the RBI described a year in which CBDC pilots moved into direct benefit transfer schemes of the Centre and the states. Fewer rupees sitting in wallets, more rupees doing specific jobs. We think those two facts are connected.

What is live: the use cases the RBI has actually run

The list below is limited to use cases that the RBI or the governments involved have described publicly as run or piloted. Most are small. None is a national rollout.

The Odisha case is the clearest example of a government using the rail at scale. Subhadra pays eligible women ₹10,000 a year for five years, and the state paid 88,000 of them in digital rupees under the pilot. The PDS pilots, reported by OrissaPOST from the 2025-26 annual report, are a stronger test of the "programmable" part, because the rule is narrow: food, at a specific class of shop.

Agriculture: the strongest case for programmable money

Subsidy fencing gets the attention. The more interesting idea is credit. Many tenant farmers cultivate land they do not own, so they cannot pledge a land title, and banks have long been reluctant to lend to them under the Kisan Credit Card. The Governor's argument at RBI@90 was simple: if loan money can only be spent on seeds, fertiliser and other inputs, a bank has more assurance that the loan funds a crop and not something else, and that can make it willing to lend without the title.

That is a credit-risk argument, and it is a good one. It does not depend on surveillance of the farmer's whole financial life, only on the narrow question of whether loan money reaches an input dealer. It also creates a natural end point: once the dealer is paid, the money is ordinary e-rupee in the dealer's wallet.

The practical hurdles are just as specific. Input dealers in rural blocks need to accept e-rupee, which means wallets, QR codes and a reason to bother. Connectivity is patchy in exactly the places where tenant farming is common, which is why the RBI tied programmability to offline functionality in the same announcement. We look at how that might work in our explainer on offline CBDC payments. And the rule has to be coded somewhere: a list of eligible merchant categories, kept current, with a route for a farmer to complain when a legitimate dealer is wrongly blocked.

Programmable money is most defensible when the rule protects a lender or a scheme, and least defensible when it simply second-guesses the person holding it.

Worked example: one welfare scheme, two kinds of plumbing

Take Subhadra's published numbers. A beneficiary is entitled to ₹10,000 a year for five years, or ₹50,000 in total. If all 88,000 women paid in digital rupees received a full year's entitlement that way, the state would have pushed about ₹88 crore of e-rupee into wallets in a year (88,000 × ₹10,000). Set that against the ₹771.66 crore of retail e-rupee in circulation at end-March 2026 and one state scheme alone would be equal to roughly 11% of the whole stock. That is how small the pilot still is.

Now compare two ways of delivering the same ₹10,000.

• Bank-account DBT. The money lands as a deposit. It is fully fungible from the first second. The state can see that it was credited; it cannot see or control what it was used for. Leakage, if any, happens after credit and is invisible.
• Programmable e-rupee. The money lands as tokens with a rule. Suppose, purely as an illustration, a ₹5,000 instalment carries a rule that it can be spent only at registered merchants in the beneficiary's district. If she spends ₹3,200 at approved shops in the first month, the merchants receive ₹3,200 of unrestricted e-rupee and the remaining ₹1,800 stays conditional in her wallet.

Two things fall out of the arithmetic. First, programmable welfare money is designed to move quickly to merchants, who are likely to convert it back to bank deposits. That pushes circulation down even as use rises. Second, every rule has a cost to the beneficiary if the list of approved shops is short. A ₹1,800 balance that cannot be spent at the nearest shop is worth less to her than ₹1,800 in a bank account. The design of the merchant list is the scheme.

The Subhadra rule set used in this example is illustrative; Odisha has not published the exact conditions coded into its e-rupee payments.

Why circulation fell while use cases grew

A 24% fall in a year looks bad, and it should prompt questions. But circulation is a stock, not a flow. It measures what is sitting in wallets on one day. Several things can shrink it while the pilot is, in a narrow sense, working.

Some early activity was target-driven. The pilot had a stated goal of one million transactions a day by December 2023, and usage built to meet a target is not the same as balances people choose to keep. The e-rupee also sits in a separate wallet that competes with UPI, which was handling about 802 million payments a day by September 2026. For most people, a separate wallet with no interest and no merchant advantage is a harder sell than a UPI app linked to the account they already use. We compare the two in e-rupee vs UPI.

And programmable money, as the worked example shows, is built to be spent. A scheme that pays ₹88 crore a year through e-rupee might leave only a sliver of that in wallets at any month end. If the RBI wants circulation to rise, it needs reasons for people to hold e-rupee. If it wants use to rise, it needs schemes and merchants. The 2025-26 numbers suggest it has chosen the second path, at least for now.

The risks: coercion, fungibility and who writes the rules

The objections to programmable money are not abstract. Restricted money can force spending a person would rather defer. Expiry dates, an idea sometimes floated for stimulus payments, are a tax on people who save. Rules can be changed. And the same plumbing that limits a subsidy to food can, in principle, limit anything. The Bank of England has said it will not pursue government or central bank-initiated programmable functions for a digital pound, which is a useful marker of how differently central banks read the same technology.

The RBI's concept note leans on "managed anonymity", with small payments more private and large ones traceable. That is a privacy design, not a programmability rule, and the two interact: a rule that restricts merchant category requires the system to know the merchant category of every restricted payment. We go deeper in the CBDC privacy debate.

Our view is that the safeguards matter more than the technology. Rules should be tied to the law or scheme guideline that funds the transfer, published in plain language, and narrow. Restrictions should drop away once money reaches a merchant. Beneficiaries need a grievance route when a payment is wrongly declined. And the default for ordinary e-rupee, the kind a person buys with their own deposit, should be no rules at all.

What to watch next

• 2026-27 pilot expansion. The RBI says it will extend CBDC pilots to more DBT schemes, domestic retail and business applications during the financial year ending March 2027. Watch which central schemes sign up; that is where volumes would move. Our guide to direct benefit transfer explains the scale involved.
• The CAT sandbox. The RBI plans a framework for testing products through a CBDC and Asset Tokenisation sandbox, which is where corporate allowance and supply-chain use cases are likely to be tried.
• Cross-border pilots. The bank signed a digital asset collaboration MoU with the Monetary Authority of Singapore in 2025-26, held talks with the Central Bank of the UAE on a cross-border CBDC pilot, and joined BIS Innovation Hub work including Project Rialto and the second phase of Project Mandala.
• End-May 2027. The RBI Annual Report 2026-27 will give the next circulation figure and, ideally, transaction counts for the programmable pilots. If the RBI wants the public to judge programmability on its merits, publishing use-case-level numbers would help.

For a plain-English primer on the instrument itself, start with the digital rupee explained.

About this researchMuffinPay InsightsIndependent analysis of stablecoins, CBDCs, cross-border payments and digital public infrastructure. Every figure is linked to its source. Read our editorial standards →