Last week, at an industry conference in New Delhi, renewable energy secretary Santosh Kumar Sarangi announced a number that made headlines. The latest auction run by SECI (Solar Energy Corporation of India), the government company that buys renewable power in bulk, discovered a price of Rs 5.25 per unit for what is being called round-the-clock renewable energy.

Discovered is auction jargon, and for once the jargon is accurate.

Nobody fixed this price. It emerged from a reverse auction, developers bidding against one another, each undercutting the last, until no one was willing to go lower.

“We expected the price to be very high,” Sarangi reportedly said, “but thanks to the competitive spirit among our developers, we discovered a rate of Rs 5.25 per unit.”

The auction was for 1,000 MW, roughly the output of a large coal plant, supplied under 25-year contracts. It drew sixteen bidders, of whom seven won. The tender carries its ambition in the name: Round-the-Clock Thermal Mimic.

A bundle of solar farms, wind farms and batteries, working together, is meant to imitate a coal-fired station, supplying steady power day and night. The government says the price should be compared with firm sources such as thermal and nuclear, and that the tender is a milestone for reliable green energy, especially for data centres.

What does it mean for the person paying an electricity bill?

A fixed Rs 5.25 for 25 years is cheaper than what new coal capacity costs (recently approved coal plants are coming in at Rs 5.85 to over Rs 6 per unit, with fuel bills that rise for decades) and far cheaper than new nuclear.

Every unit bought this way is a unit whose price cannot be inflated by imported coal or a weak rupee. For consumers, this is clean power at a fair, stable price, and that is not nothing.

What it is not, yet, is a guarantee that the lights stay on. The gap between those two things is where this story lives.

The 70% Clock

The winning developers have promised to deliver, in every 15-minute slice of the day (the grid measures everything in these slices, called time blocks), a minimum share of the contracted power. But the minimum changes with the hour.

During peak hours, the evening stretch when everyone's lights, fans and televisions come on together, they must deliver at least 90%. During the sunny hours, only 50 to 60%. During the remaining hours, the floor is 70%.

Average that across a full day and the guaranteed supply works out to roughly 70% of round-the-clock demand. The clock is only about seven-tenths wound. For nearly a third of the promised power, the buyer has no assurance, only a hope that the weather cooperates.

In the hours the floors do not cover, the contract is deliberately quiet. Nothing in the tender obliges the developer, or SECI, to procure replacement power. SECI is an intermediary: it buys from the winners and resells back-to-back to the discoms that sign up, passing the product through as contracted.

A developer's only liability for a shortfall is the penalty; the job of actually filling the unserved hours rests with the buying discom, which also chooses, a day in advance, which six hours of its day count as peak.

To be fair to SECI, this is sensible design rather than sleight of hand. The obligation is strongest in the evening, when the grid is desperate, and weakest at midday, when India's solar farms already produce so much that wholesale prices routinely crash.

Nobody needs guaranteed delivery at noon. But sensible design is still a concession. This is shaped energy with a safety floor. It is round-the-clock only in name.

The Hidden Backup

A coal station is paid in two parts. An energy charge covers fuel for units actually drawn. A capacity charge, the larger fixed part, is paid whether or not a single unit is drawn. Think of it as a retainer: it pays the plant to stand ready; under the central regulator's rules, the plant recovers it in full only by proving it was available at least 85% of the year.

In return, the buyer commands the machine. The plant declares what it can offer each day, the buyer schedules it, and the plant runs to that schedule whether in sunshine or in a week of cloud.

So when Rs 5.25 is placed beside a coal tariff, the comparison mixes two purchases. One buys energy with a partial floor. The other buys energy plus a costed, penalised promise of being there. The promise is most of what the retainer pays for.

That price mix-up is only half the trouble. The other half is the weather. Coal plants fail too, but they fail one at a time, and the grid absorbs scattered outages easily. Renewable shortfalls are caused by weather, and weather does not strike one solar farm at a time. A spell of monsoon cloud dims every developer's panels across entire states in the same fortnight.

The unassured 30% will therefore tend to go missing all at once, for everyone, precisely when replacement power on the spot market is most expensive. The contract's penalties compensate the buyer with money, but money does not light bulbs.

Someone must still keep backup ready for those hours; that someone is the distribution company, which pays for standby power through separate contracts or emergency purchases on the exchanges. That expense is incurred because of these megawatts, and it appears nowhere in the Rs 5.25.

How large is that hidden bill?

Exchange prices allow a rough range. Spot power on the IEX averaged Rs 4.99 a unit in July 2026, and scarcity evenings settle well above the average. Assume the discom fills the unguaranteed fifth to third of this product at Rs 7 to Rd 10 a unit: the effective cost of genuinely round-the-clock supply then rises by roughly 35 paise to Rs 1.45 a unit, taking Rs 5.25 to between about Rs 5.60 and Rs 6.70. The bottom of that range still beats new coal. The top does not.

The Producer’s Side Of The Bet

Who took the other side of this trade? Seven developers.

Juniper Green Energy won the largest share, 230 MW at Rs 5.26 per unit, and Hexa Climate Solutions took 150 MW at the same price. The other five won at Rs 5.25: Kengeri Prime Solar (180 MW), Resolven Four Energy (150 MW), Hero Solar Energy (120 MW), the investment vehicle EMIF II Holding (100 MW), and Purvah Green Power (70 MW), a unit of the Kolkata utility CESC.

Their price is locked for the full quarter century, against obligations that are anything but flat. To guarantee 90% of contracted capacity through every evening block, a developer must build far more than 1 MW of generation for every megawatt sold, plus hours of battery storage, and the tender's own clarifications acknowledge that installed capacity can be a large multiple of contracted capacity.

Shortfalls against the floors are tallied and penalised monthly. The contract closes the obvious escape hatch: a developer caught selling power to outsiders while its own contracted blocks go unfilled pays a penalty at twice the prevailing exchange price.

The weather risk moves onto the developer's balance sheet, priced into Rs 5.25 alongside batteries whose costs are still falling and penalties whose incidence nobody can yet predict.

India has seen this movie before: the aggressively won solar bids of the last decade that later returned as pleas for renegotiation. Whether Rs 5.25 is a sustainable price or this cycle's optimism will only be known a few monsoons in. The winners' lenders will be watching the first year's penalty statements more closely than any press release.

Why bid this low at all? Three things explain the aggression. First, batteries: standalone storage tariffs have fallen by more than 70% between 2022 and 2025, to a record low of Rs 1.48 lakh per MW per month, and a bidder pricing a 25-year contract is betting that the decline continues. Second, surplus: the oversized portfolio needed to hit the evening floors produces spare power through most hours of most days, and the tender permits its sale once the contracted obligations are met, so merchant revenue subsidises the contracted tariff. Third, the counterparty: SECI pays on time, which discoms frequently do not, and 25 years of assured offtake is the kind of revenue a lender will finance.

mThe caution comes from the same analysts who track storage economics: tariffs have fallen roughly twice as fast as battery pack prices, and three-quarters of the two-hour storage capacity awarded in 2025 is classed as at risk. The aggression may be rational, but not riskless.

Power Math Problem

Every state must now show its electricity regulator, a few years in advance, that it has lined up enough dependable power to cover its busiest evenings.

When states do this math, they should count only what the contract guarantees: every 100 MW bought from this tender as 70 MW, no more.

Whether they will is another matter, and their finances are the reason to doubt it.

Most state distribution companies are chronically loss-making and habitually late in paying the generators they already owe.

For a discom in that condition, honest counting is expensive, because admitting that 100 MW is really 70 MW means buying more firm capacity with money it does not have.

Counting at full value is free today. The cost arrives later, on a hot evening, as emergency power bought at whatever the exchanges charge that week, and it lands on household bills, the same households the cheap tariff was supposed to protect.

This is precisely why the counting rule needs to come from the centre rather than be left to each state's self-restraint.

For industry, the math is the same whether the buyer is a data centre or a factory, comparing this against rooftop solar and a diesel set. Data centres sell uptime measured in decimal places; a 70 to 90% assurance is nowhere near their standard.

Any large consumer signing up for this product still needs firm backup for the unassured hours, and its true cost of round-the-clock green power is Rs 5.25 plus that backup. Only that full sum is honestly comparable with coal or nuclear.

What Could Have Been And Still Can Be Done

None of these gaps required a different technology. They required different paperwork, and fixes are available.

First, pay green power the way coal is paid. Split the payment in two, one part for standing ready and one part for electricity actually delivered, and let developers bid on both. The auction would then reveal, in the open, what standing ready costs, instead of hiding it inside a single number.

Second, fix the arithmetic in advance. Put the hundred-counts-as-seventy rule into the planning rulebook that states follow, written down before the first shortage, so no broke discom can flatter its own plans.

Third, drop the misleading name. Call this what it is: renewable power guaranteed most of the time, and fully guaranteed only in the evening peak. “Mostly-round-the-clock” would sell fewer headlines. It would also mis-sell nothing.

The Price Of Reliability, Six Years Apart

The deeper achievement of Friday's auction emerges from comparing it with the first of its kind. In May 2020, SECI's first round-the-clock tender was won by ReNew Power at a headline Rs 2.90 a unit, and ministers celebrated it much as they are celebrating now.

The fine print told a different story even then. The Rs 2.90 was only the first year's price, rising 3% every year for fifteen years before flattening, so the average over the contract works out closer to Rs 3.60. The reliability promise was measured loosely, over months and the full year, with no hour-by-hour floors at all.

And delivery was slow: the contract itself was signed more than a year after the auction, and supply was expected only around three years after the win, from a project that needed 1,300 MW of wind and solar, and a modest battery, to serve 400 MW of contract.

Set against that, Friday's Rs 5.25 is a different animal: flat for 25 years, policed every 15 minutes, reconciled every month.

Each turn of the screw on reliability, from yearly averages to monthly checks to hourly floors, has raised the price, and faster each time. The first 70% of a dependable clock now costs Rs 5.25.

The last 30%, the part that would make this a true substitute for coal or nuclear, is the hardest and dearest part, and no auction has yet asked for it. Until one does, Rs 5.25 buys a very good imitation, and the price of the real thing remains undiscovered.