The Protocol Was Never the Constraint: What Delhi’s Peer-to-Peer Pilot Suspended

The four charges DERC switched off are the exact mechanism by which an Indian distribution company recovers the cost of the wire and the cross-subsidy the wire carries, so the pilot’s favourable economics are an accounting identity rather than a finding. Reading the primary specification sharpens the point: the India Energy Stack Architecture Document v0.4 names Beckn once, in a bibliography, as a pattern IES "can adopt", and never uses the phrase cross-subsidy at all. Beckn genuinely solves coordination, which was blocking things; it cannot decide who pays for the wire, which is what actually decides whether peer-to-peer scales.

In February 2026 a farmer in Meerut named Arun Singh sold six units of surplus rooftop solar to Lakshmi, who runs a garment shop in Delhi, and was paid thirty rupees. He did it by talking to a WhatsApp voice agent built by Pulse Energy. Within days the demonstration had been shown to the Prime Minister at the MeitY Pavilion and to the Union Power Minister at the REC Pavilion. Every part of that sentence is a genuine engineering achievement: two consumers, sitting under different distribution companies in different states, contracted directly with each other, and the delivery was evidenced and settled without either of them joining a common platform.

Now read the order that made it legal. On 12 February the Delhi Electricity Regulatory Commission cleared a six-month pilot for peer-to-peer solar trading, intra-discom, intra-state, and inter-state into Uttar Pradesh. Its conditions, as reported across the trade press, were these. Wheeling charges waived for transactions inside Tata Power Delhi Distribution's licensed area. Open access charges waived within Delhi. The twenty per cent capacity utilisation factor restriction on solar plants relaxed, so a prosumer could trade an entire output rather than a slice of it. Penalties for under-injection and under-drawl waived for the duration. Against those four suspensions, one charge added: forty-two paise per kilowatt-hour, split equally between buyer and seller. The trading price itself was left to the parties.

Those waivers are not the fine print of the experiment. They are the experiment. The charges that were switched off are the precise mechanism by which an Indian distribution company recovers the cost of the wire, and the cross-subsidy that the wire carries. Suspend them and a peer-to-peer trade clears easily. That is not a discovery. It is an accounting identity.

It is worth being exact about what Beckn is, because the exactness is where the argument lives. Beckn is an open protocol: standard APIs and message formats for decentralised discovery, ordering, fulfilment and status. It is the same pattern that runs underneath ONDC. The Unified Energy Interface is the energy specification built on it, now carried by a UEI Alliance of some eighty organisations. What it solves is real and was genuinely blocking things: a buyer registered on one platform can transact with a seller registered on another without both of them being herded onto a single platform owned by a third party. In electric vehicle charging that has produced something you can count, with reporting citing 5,386 charge points across ten companies reachable through one interface. Anyone who has watched a sector fragment into mutually illegible apps should recognise how much work that does.

Then go to the primary document. The India Energy Stack is the Ministry of Power's digital public infrastructure for electricity, run through REC Limited, and pitched publicly as the UPI for power. Its Architecture Document, version 0.4, dated 27 March 2026, runs to sixty pages and defines four architectural layers: Data, Identity, Exchange and Consent. Where in it does Beckn appear? In section 7.4, inside a bibliography of standards, sitting near the Account Aggregator framework and above a list that includes HTTP, JSON, MQTT and X.509. The entry credits Beckn with standard APIs for decentralised discovery, ordering, fulfilment and status, notes its usage in ONDC, and states its relevance to IES as a "pattern for open, federated service marketplaces" with "potential interoperability with Beckn". A cross-cutting note elsewhere says IES "can adopt" Beckn-based energy specifications such as DEG and UEI.

That is a considered and conditional endorsement of a design pattern. It is not the same claim as "the India Energy Stack is built on Beckn", which is roughly how the ecosystem describes it. The distance between those two statements is small in a press release and large in an accountability question, because they place responsibility for scaling in different places.

Now search the same sixty pages for the vocabulary of money. "Settlement" appears eighteen times. "Tariff" fifteen. "P2P" nine. "Open access" three. "Wheeling" twice. "Cross-subsidy" does not appear at all.

That absence is the whole matter. In Indian distribution the retail tariff is not a price; it is a transfer instrument. Commercial and industrial consumers pay well above the cost of supplying them so that agricultural and domestic consumers can pay well below it, and the per-unit charge on a commercial connection carries the wire, the subsidy and the utility's fixed obligations in a single number. A commercial consumer who buys directly from a prosumer at something like forty-three per cent below retail tariff, which is what earlier Uttar Pradesh pilots recorded, is not merely buying cheaper electricity. That consumer is stepping out of the transfer while continuing to use the wire the transfer pays for. Multiply by enough participants and the question is not whether the protocol works. It is who is left holding the subsidy.

None of which is a failing of the architecture document, and it would be lazy to read it as one. Setting cross-subsidy is a regulator's instrument, not a specification's, and the document is careful exactly where carelessness would matter. Its first reference use case is inter-DISCOM peer-to-peer trading, which it calls "the hardest system-of-systems case", and it describes what it delivers in bounded language: IES "makes P2P repeatable by standardising discovery plus contracting interactions and defining shared evidence and receipt artefacts". Repeatable, not viable. Coordination, not allocation. The specification understands the difference. The commentary around it frequently does not.

There is precedent for the confusion. The Brooklyn Microgrid was the emblem of peer-to-peer energy for most of a decade and it did not stall on cryptography or on protocol design. It stalled on regulation, in a country where peer-to-peer trading is largely confined to microgrids that do not use central grid infrastructure. Read that restriction again and it says something plainer: peer-to-peer is permitted where it does not raise the question of who pays for the wire. India's pilots are more ambitious than Brooklyn's precisely because they run across the shared network. That ambition is why the charge question arrives sooner here, and harder.

So: is Beckn useful for energy trading? Yes, and the yes deserves to be stated with its edges intact. It removes platform lock-in from a sector that was heading directly into it. It gives multi-utility settlement a clean evidentiary trail through signed, meter-derived actuals that every party can reference. It makes discovery across organisational boundaries cheap where it had been a bilateral integration project each time. Those gains are real, they are not small, and they were genuinely obstructing things. What does not follow is the inference that solving coordination solves peer-to-peer trading. Coordination was never the binding constraint. It was an enabling condition, and enabling conditions are the ones you notice only when they are missing.

Which makes the next few weeks the part to watch. Both DERC and the Uttar Pradesh commission said they would review the pilots after six months in order to frame durable rules. The Delhi order cleared in February. The review window is now. When that order arrives, read it the way this piece has read the February one: not for what it says about protocols or blockchains, but for which of the four suspended charges come back, and at what rate. If wheeling returns at full tariff while the forty-two paise stands, the arithmetic that made the Meerut trade work is a different arithmetic. If the waivers hold at scale, then the question simply relocates, because the wire that carried six units from Meerut to Delhi was paid for by somebody, and the order will tell you who.

The protocol will function identically under either outcome. That is what a well-designed protocol does, and it is also the exact limit of what one can do. Arun Singh's thirty rupees was a demonstration of plumbing, and the plumbing is good. Whether it becomes a market is a decision, and the decision is not in the specification.