Semiconductors & energy policy: The link driving India's growth | Energonomics

ELI5/TLDR

India loses about a fifth of all the electricity it generates just by moving it around — wires leak power on the way from the plant to your house. A power-semiconductor executive argues that instead of rebuilding the whole creaky grid, India should put smart chips into the system to claw back those losses, and skip ahead to a more modern setup the way it once skipped landlines and went straight to mobile phones. The catch: most of the specialised chips needed are still imported, so India has to build its own chip supply to actually pull this off. His company is betting on this exact gap.

The Full Story

The episode pairs two of India’s loudest industrial-policy ambitions — making semiconductors at home, and overhauling its energy system — and asks why they keep getting mentioned in the same breath. The guest, Anuj Narayan of Sine Semiconductors, gives two answers.

Why people think chips and power are linked (and why that misses the point)

The obvious link is appetite. Chip factories are electricity gluttons. A modern fab — the plant that etches circuits onto large silicon discs, called 300 mm wafers for the disc size — can pull 200 to 300 megawatts. For scale, he notes some of India’s nuclear plants run around 500 megawatts and the Srisailam hydro plant about 1,200. So a single fab can swallow a meaningful slice of a real power station’s output. For thirty years, he says, the worry that India’s grid couldn’t feed such beasts is part of why large fabs never got built here.

“That’s been the narrative for the last 30 years, which has really prevented us from having large-scale wafer fabs in India.”

But he thinks the more useful link runs the other way — not chips needing power, but chips fixing power. India has wired up nearly every village, an achievement he credits, yet it runs one of the leakiest delivery networks anywhere: 20 to 22% of electricity is lost just in transmission and distribution. His pitch is that smart power-handling chips can recover a big chunk of that waste, instantly freeing up supply without building a single new plant.

The leapfrog argument

His central analogy is telecom. India never bothered upgrading its sprawling copper landline network to fast fiber. Instead the government wrote private-sector-friendly rules and the country jumped straight to mobile — cell towers running on their own generators and car batteries, independent of both the landline and power grids.

“Why don’t we create policy that’s friendly to the private sector to go a leapfrog… We’ll go directly to mobile.”

Power, he argues, can do the same. Trying to rebuild the national grid overnight is a decades-long slog; better to jump ahead with self-contained, modern systems.

Where the losses actually come from: AC versus DC

This is the technical heart of it, and worth going slow on. Electricity comes in two flavours. The grid runs on alternating current (AC) because AC is easy to push over long distances at high voltage. But solar panels produce direct current (DC), batteries store DC, and most modern loads — your phone charger, LED lights, data-centre servers — ultimately run on DC. Every time you flip between AC and DC, or step voltage up and down, you lose a sliver of energy, roughly 2 to 3% per conversion.

Today’s chain is absurd when you trace it: solar makes DC, it’s converted to AC, transmitted, stepped down repeatedly, then converted back to DC at the appliance. Each hop bleeds power. His fix is the DC microgrid — a small, local power network that keeps electricity as DC from generation through storage to use, skipping the wasteful conversions. He frames it across three angles:

Generation — India is building roughly 500 gigawatts of renewables, much of it solar, which is born as DC.
Storage — batteries are inherently DC, so a DC system avoids the double conversion of charging and discharging.
Load — the biggest consumers are heavy industry and the 20–30 million agricultural water pumps running at any moment, plus the looming giant: data centres, which are pure DC inside. Convert their power ten times and you’ve burned 20% before processing a single AI “token.”

The supply-chain trap

Here’s the honest caveat he keeps returning to. You can design a brilliant leapfrog, but if the chips that make it work are imported, you move only as fast as global supply allows. A worldwide shortage of one part stalls the whole plan. So he insists this is simultaneously an infrastructure problem, a chip-design problem, and a product problem — solve one without the others and it breaks.

What his company is doing about it

The back third is essentially a corporate update wearing a policy hat. Sine’s three bets: building custom high-voltage power chips to manage electricity from the grid down to the home; a partnership with Navitas to make wide-bandgap semiconductors domestically — these are next-gen chip materials (gallium nitride, silicon carbide) that handle high voltage far better than ordinary silicon, and India imports nearly all of them today; and the April acquisition of Kinetic Technologies (250 products, 100-plus patents) to round out lower-voltage power management. He credits government chip-incentive schemes (DLI and others) for making this viable.

The host gently registers a doubt worth keeping: even Andhra Pradesh’s microgrid deal for Google still falls back on the main grid when the local one can’t cope — so “independence” has limits.

Key Takeaways

India loses 20–22% of its electricity in transmission and distribution — among the worst rates globally — making recovery a bigger near-term prize than new generation.
A single modern chip fab can draw 200–300 MW, comparable to a real power station; this fear stalled large fabs in India for decades.
The pitch is to “leapfrog” grid modernisation rather than rebuild it, mirroring how India skipped landlines for mobile.
DC microgrids cut the repeated AC/DC conversions that each waste 2–3%; especially relevant for solar (DC), batteries (DC), and data centres (DC inside).
The leapfrog only works if the underlying chips are made domestically — otherwise modernisation runs at the speed of import supply chains.
Wide-bandgap semiconductors (gallium nitride, silicon carbide) are critical for high-voltage systems and almost entirely imported by India today.
Government schemes (DLI, etc.) are explicitly encouraging private semiconductor players; Sine is targeting the power-electronics niche specifically.

Claude’s Take

The framing is genuinely useful: the insight that India’s bottleneck is delivery loss, not generation, and that power chips are a cheaper lever than new plants, is the kind of reframe worth carrying around. The AC/DC conversion-loss explanation is concrete and correct, and the telecom-leapfrog analogy does real work.

That said, this is a sponsored-feeling conversation. The guest sells exactly the products he’s describing as the solution, and the second half slides into a company press release — Navitas, Kinetic, “we’ve already made announcements on the parts.” Treat the technical diagnosis as solid and the prescriptions as a vendor’s view. A few claims wobble too: “millions of megawatts” for a fab is a slip (it’s hundreds), and the host’s own pushback — that microgrids still lean on the main grid for backup — quietly punctures the independence narrative without being resolved. No data centre actually runs purely on a DC island today.

Score 6: clear, well-explained, a couple of takeaways with real shelf life, but it’s a short interview that’s part education, part sales pitch, and never gets pressed hard on the gaps.

Semiconductors & energy policy: The link driving India's growth | Energonomics | EP 18