India's rooftop solar problem | Recapping 2025 in the energy world

ELI5/TLDR

India doesn’t have a “people don’t want solar panels” problem anymore — most people know about rooftop solar and a lot want it. The real problems are on either side: most willing households never figure out how to start, and most installed panels quietly underperform because nobody cleans the dust off them. Separately, 2025 was the first year in 140 years that the world made more electricity without burning more fossil fuels — basically because China had a monster solar year. But emissions barely fell, because fossil fuels still make more than half the world’s power.

The Full Story

This is a two-story energy brief from Zerodha’s Daily Brief, plus a few tidbits. The first half is India-specific (two new reports from the Council on Energy, Environment and Water). The second is the global picture (Ember’s Global Electricity Review 2026).

Story 1: The rooftop solar problem isn’t the one we thought

For years the assumed problem was awareness — Indians didn’t know about rooftop solar, didn’t trust it, found it foreign. CEW surveyed 17,000 households across 308 districts and demolished that idea. 57% already knew rooftop solar was an option. A quarter were willing to install. And of the 43% who didn’t know about it, half wanted it once it was explained.

So demand is largely solved. The problem is two gaps that bracket that desire, and neither shows up in the one number everyone quotes — installed capacity.

Both of these gaps often seem invisible because the only number anyone usually quotes is installed capacity, which tells you nothing about either.

Gap one: wanting it vs. installing it. Of every 10 willing households, 8 never even start. An 80% drop-off between intent and action. The single biggest reason — cited by 68% — is that they simply don’t know how to begin. The process is a fog. Only 12% of people who knew about rooftop solar knew that step one is registering on a national portal.

This reframes the fix. If the barrier were money, you’d throw subsidies at it (which India has, via PM Surya Ghar). But the barrier is information, which is cheap to fix. CEW’s line is that one well-designed handholding campaign could move more people into solar than a big subsidy hike.

The people who do install share one trait: they’re rich. High-income households adopt at roughly 10x the rate of low-income ones. The typical adopter has a 50% higher electricity bill and runs ACs and geysers, so the economics are a no-brainer. The catch is that solar spreads by word of mouth, word of mouth travels within social networks, and social networks cluster by income. Left alone, solar circulates among the affluent and skips exactly the households policy wants to reach.

There’s also a self-defeating subsidy design. PM Surya Ghar caps the subsidy at a 3 kW system. So wealthy adopters who actually need more deliberately undersize to milk the subsidy, while low-consumption households — the target — have little incentive because small bills mean a long, unattractive payback.

Gap two: installing it vs. actually generating power. Installed capacity is easy to celebrate and easy to fake — a panel that does nothing but gather dust still counts.

A solar panel can make for installed capacity even if it does nothing but accumulate dust.

The villain here is mundane: dust and bird droppings can cut a panel’s efficiency by as much as 60%, and unaddressed, cause localized overheating that cracks glass or, in bad cases, starts a fire. PM Surya Ghar mandates 5 years of free maintenance, but the scheme never defined what maintenance means — no minimum scope, no required visit frequency. So in practice cleaning (the most performance-critical task) is dumped on the consumer, and the “free” maintenance was baked into the upfront price, meaning the vendor already has the money and zero reason to show up.

Most people don’t even notice the rot. Of the 93% of adopters who say they’re satisfied, half don’t know they have a maintenance contract, and nearly half never check how much power their system produces. Bad maintenance doesn’t announce itself — output bleeds away slowly and quietly.

Some vendors are building the opposite model: proactive maintenance, real-time generation apps, output guarantees. Zerodha calls out SolarSquare’s “Good Zero” plan, and notes that Zerodha’s Rainmatter has invested in SolarSquare. (Worth flagging that disclosure.)

The economic stakes cascade to three parties. A well-run 3 kW system pays back in ~5 years, beating a fixed deposit. A neglected one can stretch payback past the panel’s entire 25-year life — all lifetime savings gone. For the discom, failed rooftop solar causes a “grid rebound effect”: households go back to the grid, and the discom, having planned around power that never arrived, scrambles to buy replacement supply. For the government, it’s wasted subsidies plus, since many states subsidize residential power, paying a subsidy twice.

The flip side: every panel needs servicing for 20+ years, which is a recurring market. At the 30 GW residential target, CEW sizes the O&M market at tens to low hundreds of billions of rupees a year and ~3.3 lakh dedicated jobs.

Story 2: 2025 — the year the world stopped adding fossil fuels

Ember’s Global Electricity Review 2026 had one deceptively simple headline: the world generated more electricity last year without burning more fossil fuels. That’s happened four times before, but always alongside an excuse — a recession, a pandemic, a mild winter. 2025 was the first normal, anomaly-free year it happened.

For ~140 years the pattern was iron: economies grow, demand rises, fossil fuels fill the gap. That grip has been loosening. Fossil fuels’ share of new demand went from 73% in the late 2000s, to 36% (2015–19), to 29% (2020–24), to zero last year.

Solar did almost all the heavy lifting — 75% of all new global electricity demand, with wind covering most of the rest. Together, 99% of new demand. The scale is hard to hold in your head: in 2015 the world made 256 TWh of solar; last year, 2,778 TWh — over 10x in a decade. Growth hit 30% in 2025, the fastest in 8 years despite a much bigger base. Solar output has been doubling every 3 years. The upshot: for the first time in over a century, renewables generated more electricity than coal. The last time that happened was 1919, when demand was 300x smaller and “renewables” basically meant big hydro dams.

But strip out China and the story changes. Outside China, fossil generation had already been flat since 2018 — Europe retiring coal, the US switching coal to gas. China was the one country pushing the global number up, adding 1,145 TWh of fossil generation between 2018 and 2024 (4x what India added). Then in 2025 China flipped: its fossil generation fell for the first time since 2015 — not from a slowdown (energy use grew), but because clean energy grew faster. Chinese solar grew 40% in a year. China installed 378 GW of new solar in 2025; the US has built 274 GW in its entire history.

China added more in one year than the US has built in its history.

So the honest version is: China had a historic clean-energy year, and that alone tipped the global number.

India’s 2025 milestones. Coal’s share of generation fell from 75% to 71% in one year. Solar overtook hydro to become India’s largest clean source for the first time, now 9.4% of the mix (up from 5.3% three years ago). India installed more solar than the US in 2025 for the first time. But demand grew only 2.4% — third lowest in 20 years — not from a weak economy but a cool year; mild summers (vs brutal 2024) cut demand by ~32 TWh. The report calls the demand dip an outlier: demand grew 100+ TWh in each of the four prior years, but only 49 TWh last year.

The deeper point echoes Story 1: building capacity is half the job, using it is the other half. India has a big pipeline (101 GW solar under construction, 24 GW wind, 23 GW hybrid). But at noon, solar generates 50+ GW; from 6pm–6am, coal still meets over three-quarters of demand growth. There were power shortfalls on 13 of 15 nights in late April while midday solar was being wasted. The cause is coal physics: coal plants have a minimum output they can’t dip below without a slow, expensive shutdown, so they keep idling through the solar-flooded noon to be ready for the evening peak. Solar gets curtailed at midday, coal runs at night, both wasted in different ways. The telling sign: in 2025 Indian coal plants started inviting bids to bolt batteries onto themselves, because there isn’t enough storage yet to cover the night.

The emissions buzzkill. Despite all of the above, global power-sector emissions in 2025 fell only ~0.4%. Fossil fuels still generate 57% of the world’s electricity, and the system is so large that a historic clean-energy year barely moves the needle. The direction changed; the pace hasn’t. Last year’s coverage of the IEA outlook found that even if all stated climate policies land, we’re heading for 2.5°C of warming — and 2025’s data doesn’t change that.

The most interesting observation: the countries with growing fossil generation aren’t the ones with no alternative — they’re often the ones with world-class sun. Egypt gets twice the sunlight of the Netherlands, yet the Netherlands makes 1,553 kWh of solar per person a year and Egypt makes 67. Saudi Arabia, Indonesia, Algeria, much of Africa — same. Cost isn’t the excuse anymore: in 2024, 90% of new renewable projects were already cheaper than the cheapest fossil alternative. What’s left are the hard, un-technological problems — grids not built for variable solar, financing not reaching the right markets, regulations written for a coal-and-gas world.

Tidbits

Nvidia unveiled an AI-focused “RTX Spark” superchip for personal AI agents, creative work and gaming; commentators see AI-first PCs as an opening for India in chip design and semiconductor manufacturing.
Oil India made another natural gas discovery (offshore), feeding hopes of cutting import dependence.
India’s economy grew 7.7% in FY26 (up from 7.1% in FY25), led by manufacturing and services; Q4 eased slightly to 7.8% from 8%.

Key Takeaways

Demand for rooftop solar in India is largely solved: 57% awareness, a quarter willing, and half of the unaware become willing once told.
The killer gap is intent-to-action: 8 of 10 willing households never start. 68% say the reason is not knowing how to begin; only 12% knew step one is registering on a national portal.
The binding constraint is information, not money — CEW argues one handholding campaign beats a subsidy hike.
Adoption skews rich: high-income households adopt ~10x more; the typical adopter has a 50% higher bill. Solar spreads by word of mouth within income-clustered networks, so it skips the poor.
PM Surya Ghar’s 3 kW subsidy cap makes the wealthy deliberately undersize systems while giving low-bill households too weak a payback to bother.
Dust and bird droppings can cut panel efficiency by up to 60% and, untreated, crack glass or start fires.
The mandated 5-year free maintenance has no defined scope or visit frequency, so panel cleaning (the critical task) gets pushed onto consumers; maintenance cost is prepaid into installation, killing any vendor incentive to show up.
Of adopters who say they’re satisfied, half don’t know they have a maintenance contract and nearly half never check generation.
A well-run 3 kW system pays back in ~5 years; a neglected one can take longer than its 25-year lifespan.
2025 was the first anomaly-free year in 140 years where electricity grew without more fossil fuels.
Solar covered 75% of new global demand, wind most of the rest; solar output rose 10x in a decade (256 → 2,778 TWh) and doubles every 3 years.
Renewables beat coal globally for the first time since 1919.
China drove the flip: 378 GW of new solar in 2025, more than the US has installed in its entire history; Chinese fossil generation fell for the first time since 2015.
India: coal’s generation share fell 75% → 71% in a year; solar overtook hydro as top clean source, now 9.4% of the mix; India out-installed the US on solar for the first time.
India’s demand grew just 2.4% — an outlier driven by a cool year (~32 TWh less demand), not a weak economy.
The intermittency problem: solar floods the grid at noon and gets curtailed, while coal runs minimum-load all night; 13 of 15 late-April nights had shortfalls despite wasted midday solar. Indian coal plants began bidding to add batteries to themselves.
Global power-sector emissions fell only ~0.4% in 2025 — fossil fuels still make 57% of electricity, so the system is too big for one good year to move emissions.
The biggest fossil holdouts often have world-class sun (Egypt: 2x the Netherlands’ sunlight, 1/23rd its per-capita solar). In 2024, 90% of new renewables were already cheaper than the cheapest fossil option.

Claude’s Take

A genuinely good brief — both stories share the same backbone idea, which makes the episode feel coherent rather than a news dump. The backbone: installed capacity is a vanity metric. India can win the demand fight, hit the subsidy targets, and still get nothing because panels gather dust and coal runs all night. The world can have a historic clean-energy year and barely dent emissions. “Crossing a milestone is not the same as solving the problem” is the through-line, and it’s the right one.

The numbers mostly hold up against what I’d expect — solar doubling every ~3 years, renewables passing coal, China’s outsized role — these track Ember’s and IEA’s framing. The “0.4% emissions fall” reality check is the most valuable bit, because it’s the one number that cuts against the triumphant headline, and most coverage buries it.

Two honesty flags. First, the SolarSquare mention is a Rainmatter (Zerodha) portfolio company, which they disclose — fine, but worth keeping in mind that the “proactive maintenance is the future” framing conveniently flatters a company they own a piece of. Second, the dust-causes-fires line is real but presented at its scariest; it’s the bad-case tail, not the median outcome.

Docked from a 9 only because the source-of-numbers gets fuzzy in places (the transcript’s ASR mangles a few figures) and the tidbits are filler. Solid 8 — the kind of explainer that actually reorganizes how you think about a topic rather than just updating a stat.

India's rooftop solar problem | Recapping 2025 in the energy world | The Daily Brief #482