where India actually sits in the next tech wave - by Jaimin

Poems, Short stories and other things..

SubscribeSign in

where India actually sits in the next tech wave<br>Jaimin<br>Jun 08, 2026

Share

There’s a question worth sitting with, uncomfortable as it is: when the AI-and-robotics build-out of the next decade is finished and the supply chains have hardened, will India own any part of it, or will it once again be the place that serves the people who do?<br>I want to weigh this honestly, because the easy version (”India is hopeless”) is wrong, and the comfortable version (”we have a thriving startup ecosystem, we’re fine”) is also wrong. The truth is more specific, and more urgent.<br>The layer cake nobody can opt out of -

Strip the next wave down to its physical and computational stack, and a small number of countries own each layer outright:<br>Frontier models: the United States. The labs setting the capability frontier are American, and so is the company designing the chips they train on.

Leading-edge lithography: the Netherlands. ASML is the only company on Earth that makes EUV machines. Every advanced chip in the world is, at one remove, downstream of one Dutch firm. You can listen to this March podcast to get more information to get deeper understanding of the stack.

Leading-edge manufacturing: Taiwan. TSMC turns those machines into the chips. This is the single most consequential chokepoint in the global economy, and the reason “the Taiwan question” is really a semiconductor question. As Gavin Baker mentioned in Invest like the best.

High-bandwidth memory: South Korea. SK Hynix and Samsung make the HBM that AI accelerators can’t run without. They are having generational run. (including Mircon from US)

Industrial robotics and upstream chip materials: Japan. FANUC and Yaskawa on the robot side; Tokyo Electron, Shin-Etsu, and the photoresist makers on the materials side. Quiet monopolies that the whole stack depends on.

Open-weight models and manufacturing scale: China. The strongest open-weight model families now come out of China, and it installs more industrial robots than the rest of the world combined. Its chip manufacturing is real but stuck at the trailing edge, because EUV export controls keep it off the leading node, a gap that takes many years and possibly a Taiwan crisis to change.

Now find India on that list.<br>“No skin in the game”? Not quite

It would be tidy to say India is simply absent. It isn’t, and the honest accounting matters more than the rhetorical punch.<br>India has, finally, bought a ticket:<br>The IndiaAI Mission (about ₹10,000 crore, roughly $1.2B) has onboarded more than 38,000 GPUs into shared compute and funded a dozen teams (Sarvam, BharatGen, Soket, Gnani) to build indigenous, largely open-source foundation models, some targeting 120 billion parameters.

Dholera will be India’s first real fab. Tata, partnered with Taiwan’s PSMC and equipped by ASML, is targeting first silicon by December 2026. Micron’s assembly-and-test plant in Sanand is already operational. India joined a US-led semiconductor supply-chain alliance in early 2026.

Space is the one bright spot the bears underrate: ISRO’s cost-per-launch is genuinely world-class, and a private layer (Skyroot, Agnikul) is forming underneath it.

So the seat exists. The problem is where the seat is.<br>Look closely and a pattern repeats at every layer. The fab is targeting 28nm and above, mature nodes for cars and appliances, not the 2-3nm frontier that AI runs on. Sanand does packaging and test, not fabrication. The sovereign models are competent and useful for Indian-language public services, but they are fast-followers built on imported architectures and imported silicon, not capability-frontier bets. The launch business sells services, not deep platform technology.<br>India has entered every layer at the trailing edge, under-scaled by roughly an order of magnitude, and is present in exactly zero of the chokepoints that confer real leverage. The IndiaAI Mission’s entire multi-year budget is smaller than what a single US hyperscaler spends on AI infrastructure in a single quarter. That is the gap, stated plainly.<br>weighing the causes, and rejecting the easy one

Why is India structurally a step behind at the exact moment the next platform is being poured?<br>The temptation is to reach for culture: Indians play it safe, society doesn’t reward risk, too much energy goes into social and religious life instead of building. There’s a grain of truth buried in there, but as a primary explanation it doesn’t survive contact with data.<br>The R&D gap is the heavy variable. India spends about 0.65% of GDP on research and development. China spends roughly 2.4%, the US around 3.5%. In absolute terms in 2024, the US and China each spent close to $780 billion; India spent about $76 billion. Worse, India’s private sector funds only about a third of national R&D, versus more than two-thirds in the US, China, and South Korea. Frontier innovation is downstream...