Can India's Grid Handle More ACs, Data Centres & EV Charging? | Govindraj Ethiraj

ELI5/TLDR

India’s electricity problem used to be simple: not enough power. Now there’s plenty of power on average — the trouble is having it at the right moment. The country built a mountain of solar, which only works in daylight, but everyone wants to crank their AC in the evening when the sun is gone. So the grid runs a daily scramble to buy expensive power for a few evening hours, while during the day electricity is sometimes nearly free. The fix isn’t more solar — it’s storage (batteries to time-shift the daytime surplus into the evening), more wind (which conveniently blows in the evening), and smarter wires.

The Full Story

This is a 46-minute conversation between Govindraj Ethiraj of The Core and Vishal Mehta, who runs the energy practice at Boston Consulting Group in India. It’s a tour of why a country with surplus generation capacity still finds itself panicking on hot evenings.

The day used to be the problem; now it’s the evening

A power grid is a balancing act that has to clear every fifteen minutes. India splits the day into 96 such slots, and a utility has to tell a central dispatcher, a day in advance, how much power it will draw in each slot. Guess wrong in either direction and you pay a penalty — because if everyone’s forecasts don’t add up, the grid’s frequency drifts off its target of 50 hertz and the whole system gets shaky. Think of frequency like the pressure in a water pipe: too many people drawing at once and the pressure sags; the grid operator’s job is to keep generation and consumption matched so the needle stays put.

Twenty years ago the headache was the daytime peak, and engineers worried about what to do with coal plants at night that couldn’t be switched off fast enough — prices would collapse in the dark hours. Today it’s the mirror image.

“10 20 years back when I would work with clients, we would find the peak used to be obviously in the daytime… Now we have an exact opposite situation.”

Urbanisation pushed the peak into the evening. People come home, switch on the AC, and the load spike lands right as the sun — and India’s vast solar fleet — clocks off.

The solar paradox: surplus you can’t use

Here’s the mismatch in numbers. Last year India added roughly 50 gigawatts of solar. Solar generates for about six useful hours a day. Six hours times 50 gigawatts is around 300 gigawatt-hours of energy crammed into the middle of the day. But daytime demand only rose by about 50–60 gigawatt-hours in that window. So India manufactured a daily surplus of roughly 240 gigawatt-hours — energy generated in the afternoon with nowhere to go — and it does this every single year.

A gigawatt-hour, loosely, is how much energy a million homes’ worth of one-kilowatt appliances would chew through in an hour. The point isn’t the exact figure; it’s that the surplus is enormous and it’s daily.

“It’s not just about average amount of electrons I can generate, but it’s at what time I can generate, with what flexibility I can generate and dispatch them.”

Solar and wind have what’s called “must-run” status: whatever they produce, the grid is obliged to take. You can’t dial them up or down on command — the weather decides. The only resource India can truly summon at will (“dispatchable” power) is thermal, mostly coal — but coal is slow to ramp up or down, expensive now, and useless for a sudden two-hour evening spike. So the daily drama is the evening hunt: as solar fades around 5–6 pm and load climbs, every utility piles into the spot market at once, paying up to 20 rupees a unit for power they’d normally buy far cheaper.

”Spot” isn’t spot, and the cash problem

The exchange where this happens is called the day-ahead market — already a slight misnomer, because you’re buying a day in advance, not on the spot. There’s also a newer real-time market (RTM) that lets you correct course about 90 minutes before a slot. Price discovery happens, bids get matched, a congestion check confirms the contracted power can physically be routed through the wires, and it gets scheduled. Mehta says there’s essentially no friction in the mechanics.

The friction is elsewhere. A private urban utility can move fast — buy when it needs to, with light internal approvals. A government-owned distribution company (a “discom”) hits a wall: spot purchases demand cash that day (no month-end credit like long-term contracts give), and every expensive purchase has to be justified later to a regulator. So a cash-strapped state discom dithers, leaves it to the last minute, or quietly sheds some load — the old escape hatch that’s mostly gone now, because once you’ve given customers 24/7 power, demand becomes “inelastic”: you can’t take it back.

Why your bill doesn’t move even when wholesale prices swing wildly

India runs regulated tariffs. A discom is a “cost-plus” entity: tot up all its allowed costs, add a guaranteed 15.5–16% return on equity, divide by total units sold, and that’s the tariff. So the price you pay is an average, and a backward-looking one — today’s bill reflects last year’s costs, not this evening’s 20-rupee scramble. That’s why the wild wholesale swings (near-free at noon, brutal at 6 pm) never reach the household.

The tool meant to fix this is “time-of-day” pricing — cheaper in the surplus afternoon, dearer in the scarce evening, with the average unchanged. Mumbai has begun rolling out time-of-day meters, working up from North Bombay to Andheri. The nudge is to get people to align their consumption with when power is plentiful. The catch: most household load is stubborn. You’re going to run the AC when you get home, not at 2 pm. Industrial and agricultural users are more flexible — factories can shift shifts, and states like Gujarat and Maharashtra have already moved subsidised farm power from night to afternoon, since if you’re subsidising it anyway, you may as well do it with the cheapest (daytime solar) electrons. Those farm loads are big, so shifting them genuinely helps match demand to generation.

In an ideal world every 15-minute slot would carry its own price (96 prices a day), but no household can navigate that. Western markets solve it with retail energy contracts — third parties package the volatility into a fixed or banded price you can buy. India can’t easily do this yet because it hasn’t “unbundled” the last mile: the company that owns the wires and the company that sells you the power are the same entity. Regulation has actually allowed separating “carriage” (the wires) from “content” (the power) since the 2003 Electricity Act — but discoms resist, because they lose money on the wires and make it on supply. Give up supply and the cost of the wires alone would have to rise. The one crack in the wall: large consumers (above 1 megawatt) increasingly sign power-purchase agreements straight with a renewable generator — the “C&I” (commercial and industrial) market — paying the discom only to carry the power, not for the power itself.

The missing card: gas

Any well-designed grid juggles four goals at once — affordability (cheapest electron, which today is solar), reliability (will it be there when I want it?), dispatchability (can I send it on command?), and now the “colour” of the power (how green it is). No single source ticks every box, and which goal ranks highest shifts as a country develops. For years India just wanted power for everyone, colour be damned; now it has green targets too, which piles pressure onto flexibility and dispatchability.

India has solar, wind, coal (reliable, dispatchable, but slow and now pricey), a little nuclear (clean, reliable base load, but costly), and limited hydro. The “joker in the pack” it lacks is gas — globally predictable, relatively clean, and fast to dispatch, exactly what an evening peak demands.

“One important joker in the pack which is missing in India, which is gas.”

India has almost no domestic gas, and imported gas is expensive. Worse, the cheap domestic “APM” gas that exists is rationed by priority: first to city gas distribution (CNG for vehicles, piped cooking gas), then to fertilisers (which guzzle gas to make ammonia, and tie directly to food inflation). Power sits below both. So gas plants flicker on and off depending on scraps of allocation or pricey imported LNG — and you can’t conjure LNG overnight; a cargo needs 45 days’ notice, or up to two months if it’s coming from the US. Gas as a molecule is liquid and globally available (even with Qatari supply disruptions, it exists somewhere and can be rerouted, just at higher transport cost) — but building enough gas-fired capacity in India just doesn’t pencil out. Mehta, “as a power systems engineer,” wishes India had more, but doesn’t see it as the answer.

So what does solve it? Storage, wind, and accepting expensive thermal

Three levers. Storage is the obvious one, and Mehta thinks everyone is underestimating how much is needed — a mix of technologies (pumped hydro, batteries), because each has trade-offs. That 240 gigawatt-hour annual surplus is roughly the entire storage build India’s official projections target for 2030–32 — and it’s being created in a single year. The simple, brutal arithmetic: if next year’s peak adds ~20 gigawatts lasting two hours, that’s 40 gigawatt-hours of storage needed. The year after, maybe 60. It compounds annually — “like a national campaign now” to keep adding it.

There’s a live risk hiding in the storage tenders. Developers have bid battery projects very aggressively, betting that Chinese battery prices keep falling between contract signing and actual purchase. If prices U-turn instead, the projects go underwater. This already happened in solar around 2021: bids near 15–16 cents a module collapsed to ~2 rupees, then prices reversed and shot up, and many projects were quietly never built. Mehta has his fingers crossed the battery commitments actually get delivered.

Wind is the under-loved second lever, and Mehta thinks India should chase far more of it — because wind tends to pick up in the evening, exactly when the load peaks and solar dies. On that 270-gigawatt-peak day, wind quietly carried a lot of the evening load. The problem is wind is slower and pricier to build: solar projects commission with an average 3-month delay, wind with a 9-month delay, and wind costs more upfront. So every investor wants wind in the system but would rather someone else build it.

“Everybody wants wind in the system. But I will not do it. Let somebody else do it… I want a Bhagat Singh but not in my house, but in my neighbour’s house.”

Thermal (coal) rounds it out, but in a new role: running at low utilisation, unable to compete with cheap daytime solar, switched on mainly to relieve the evening-night crunch. That low utilisation makes each unit it produces expensive — but it’s the backstop.

Demand-side management: India has barely tried

Asked whether people can be taught to use less, Mehta is blunt: India runs demand-side management as campaigns, not incentives, and campaigns don’t last. He remembers a Mumbai power crisis in his childhood and hoardings everywhere telling people to keep their AC at 24 degrees — it worked so well that 24 still lives in everyone’s head. But campaigns fade as generations turn over. Rooftop solar is great generation, but it doesn’t touch the evening problem because it, too, only produces in the afternoon.

The real lever is a price signal strong enough to bite. Today’s time-of-day price gap is too small to change anything. Make the evening-versus-afternoon difference 60–70%, and behaviour shifts: people install home batteries to charge cheap in the afternoon and run the AC off them at night; buyers suddenly care whether an AC is five-star or three-star (today the efficiency arbitrage isn’t worth paying for); architects design homes around energy. India’s inverted tariff — you pay more per unit as you consume more, and industry pays more than households, which Mehta says is the opposite of most countries — is a blunt instrument that hasn’t worked, because the problem isn’t total units, it’s which slot.

Notably, Mehta isn’t a fan of just consuming less. He flips the usual logic:

“People think that with increasing affluence people consume more energy. It’s also the other way around. With more energy consumption creates more growth and more affluence in the economy.”

He recalls the old “Pune model,” where the industrial city staggered weekly holidays across regions — your factory’s day off might be Thursday — to smooth the load.

The looming new loads, and the wire that costs more than the power

Looking ahead, the demand surge won’t only come from ACs. Data centres are a “non-trivial, concentrated load.” A steel plant needing 250 megawatts traditionally built its own captive power plant on-site; abroad, data centres run on captive gas plants. In India a data centre has no captive plant — it pulls 200–250 megawatts straight from the grid, and the national demand forecast gets revised upward every month partly because of them.

Then Mehta delivers the line that reframes the whole problem. In any commodity, trade happens when the cost of transport is lower than the cost of the thing itself — that’s why LNG made gas globally tradeable, why crude tankers change hands a hundred times before docking. Electricity has crossed the opposite threshold:

“In electricity you have reached a point where cost of transport is actually higher than the cost of generation.”

The wires now cost more than the electrons. The way to fix that is to drive up utilisation of the “pipe” — you can’t justify infrastructure used only 20–25% of the day. Hence a huge role for technology, and not just the fashionable AI-for-forecasting kind (though that matters too — India now has to forecast generation as well as demand, because weather-dependent solar and wind didn’t need forecasting 20 years ago). Mehta stresses the unglamorous hardware: the electrical equipment industry has stagnated, and the cheapest way to upgrade a transmission corridor is not to fight for new land and right-of-way but to squeeze more capacity out of existing towers — more circuits, higher line ratings.

He closes with a reliability benchmark called SAIDI — the average minutes of power interruption a customer suffers over a year. Tokyo sits at 2 minutes. Indian urban utilities like Mumbai run into the low hundreds of minutes; rural ones, far worse. And — a useful myth-buster — seven out of ten outages aren’t because supply ran short. They’re local: a relay tripping to stop equipment overheating in the heat, a transformer at its limit. It’s a side-street problem, not a highway problem.

His own next focus as a consultant: the neglected middle layer, transmission. The old “4:2:1” planning rule — 4 years to build generation, 2 for transmission, 1 for distribution — has broken. Renewables crashed generation from 4 years to 2, but transmission is still stuck at a tight 2 years and can’t keep up. So generation now races ahead of the wires that carry it. India needs transmission built faster, used more fully, and made more intelligent.

Key Takeaways

India no longer has a power-quantity problem; it has a power-timing problem — surplus on average, scarce in the evening peak.
The grid clears every 15 minutes (96 slots a day); utilities must forecast each slot a day ahead, with penalties for both under- and over-draw, to keep frequency pinned at 50 Hz.
The 270 GW national peak (hit ~2 weeks before recording) was a single 15-minute slot, not an all-day level.
50 GW of new solar a year generates ~300 GWh in a 6-hour window but daytime demand only rose ~50–60 GWh — creating a ~240 GWh daily surplus with nowhere to go.
That single year’s surplus (~240 GWh) roughly equals India’s entire planned storage build for 2030–32.
Solar and wind have “must-run” status — the grid must take whatever they produce; only thermal (coal) is truly dispatchable, but it ramps slowly.
“Day-ahead market” (DAM) is the misnamed “spot” market — you buy a day in advance; a newer real-time market (RTM) allows corrections ~90 minutes before a slot.
Evening scramble economics: state discoms struggle not on mechanics but on cash (spot needs same-day payment, no credit) and regulator justification; private urban utilities move faster.
Indian household tariffs are a backward-looking average (cost-plus + 15.5–16% guaranteed RoE ÷ total units), which is why wild wholesale swings never reach the consumer’s bill.
Time-of-day pricing (rolling out in Mumbai) aims to nudge load toward the cheap afternoon, but most residential load is inelastic; today’s price gap is too small to change behaviour.
A 60–70% afternoon-vs-evening price gap would trigger home batteries, five-star AC demand, and energy-first home design; current gaps don’t.
Gujarat and Maharashtra have shifted subsidised farm power from night to afternoon to soak up cheap daytime solar — big, flexible loads that help match demand to generation.
Gas is the “missing joker”: fast, clean, dispatchable — but India has little domestic gas, expensive imports, and rations cheap APM gas to city-gas and fertilisers above power.
LNG can’t be summoned fast — a cargo needs ~45 days’ notice (up to 2 months from the US).
Three real fixes: storage (a mix of pumped hydro + batteries), more wind (peaks in the evening, aligned with load), and low-utilisation expensive thermal as backstop.
Storage tender risk: batteries bid aggressively on the bet that Chinese prices keep falling; a price U-turn makes projects unviable — exactly what burned solar developers around 2021.
Wind commissions with a 9-month average delay vs solar’s 3 months, and costs more upfront — so developers avoid it despite its evening-aligned value.
Data centres are a major new concentrated load (200–250 MW each, grid-fed in India with no captive plant), pushing demand forecasts up every month.
Electricity has crossed a threshold where the cost of transport (wires) exceeds the cost of generation — the fix is driving up pipe utilisation, not just adding generation.
SAIDI (average annual minutes of customer interruption): Tokyo = 2 minutes; Indian urban utilities run into the hundreds. Seven of ten outages are local equipment/heat trips, not supply shortfalls.
The “4:2:1” planning rule (4 yrs generation, 2 transmission, 1 distribution) has broken — renewables cut generation to ~2 years, leaving transmission as the new bottleneck.

Claude’s Take

This is the rare infrastructure interview that earns its 46 minutes. Mehta is a consultant, so there’s the usual fondness for clean frameworks (four objectives, three sources, 4:2:1), but the frameworks actually clarify rather than decorate, and he keeps grounding them in numbers and field observation — the visit to the North India discom, the agonising over the evening slot, the SAIDI benchmark. No vendor pitch, no “AI will solve everything” hand-waving; in fact he pointedly says the boring hardware (transmission towers, electrical equipment) matters more than the digital glamour.

The single most useful idea here is the reframe: India’s power story has flipped from a scarcity problem to a timing and logistics problem, and the cost of moving an electron now exceeds the cost of making it. That’s a genuinely clarifying lens, and it explains a pile of otherwise confusing facts — why near-free noon power coexists with 20-rupee evening power, why building yet more solar doesn’t help, why everyone’s suddenly obsessed with batteries. The wind-vs-solar asymmetry (9-month vs 3-month commissioning delays creating a free-rider problem on the resource that’s actually best-aligned to demand) is the kind of non-obvious detail that justifies listening to a specialist.

Where I’d push back gently: the conversation stays comfortably inside the technocrat’s worldview. The discoms’ chronic losses, the politics of free farm power, and the unbundling stalemate are all treated as engineering puzzles awaiting the right price signal — but those are deeply political, and “just send a stronger price signal” runs straight into the reason tariffs are inverted and farm power is free in the first place. The optimism that next year’s peak “will get taken care of” rests entirely on storage tenders actually getting built, which Mehta himself flags as shaky given the solar precedent. Still, it’s honest about its own uncertainty. An 8 — substantive, specific, and it leaves you understanding a system you probably took for granted.

Can India's Grid Handle More ACs, Data Centres & EV Charging? | Govindraj Ethiraj | The Core Report