I started putting this together while preparing my Kalaari Fellowship application. I kept finding more material than the application could hold — infrastructure timelines, funding data, gap analysis, global comparisons. Rather than leave it in a notebook, I've tried to put the full synthesis here. It's research in progress, not a finished document.
Three simultaneous shifts are changing what companies can build in India and who they can serve. This is my attempt to understand why they're happening, what they imply economically, and where the earliest interesting investment opportunities might be.
The short version explains what I think is changing. The longer version is where I tried to understand the mechanisms, the gaps that persist, and what those gaps imply for where new companies can be built.
Indian AI funding in H1 2026 reached $1.34B across 66 rounds (Tracxn) — more than 2.4× the same period in 2025. But most capital is going to application-layer companies, not infrastructure. The infrastructure gaps are therefore both the constraint on the sector's growth and the most underinvested investment category.
India holds ~49% of global real-time payment volume (NPCI, 2026). But UPI had demonetisation as a forcing event. ONDC — the open commerce protocol — has no equivalent shock. Infrastructure existing and infrastructure being used are different things. The adoption gap is the business problem.
| Capability | India | US | China | Benchmark | India's strategic question |
|---|---|---|---|---|---|
| Real-time payments / DPI | ★ Leader | Lagging | Strong | Estonia (identity) | How to move from transaction infrastructure to programmable economic substrate |
| Digital identity at scale | ★ Leader | Lagging | Strong | Estonia (X-Road) | Identity exists; interoperability and consent infrastructure lag |
| Frontier AI / foundation models | Early | ★ Leader | Strong | US (OpenAI, Anthropic) | India's path is application layer + sovereign models, not frontier model competition |
| AI software / applications | Growing | ★ Leader | Strong | US (software depth) | India's advantage is cost + DPI rails + domain expertise in specific sectors |
| AI compute / GPU infrastructure | Lagging | ★ Leader | Strong | US (Nvidia) | IndiaAI Mission targets 10,000+ GPUs but distribution remains a problem |
| Chip design talent | ★ Strong | Leader | Strong | US (design ecosystem) | 20% of global chip design workforce is Indian. The talent exists. Ecosystem depth doesn't. |
| Semiconductor fabrication | Building | Strong | ★ Leader | Taiwan (TSMC) | India's bet: non-frontier 28nm sovereignty, not competition with TSMC at 2–3nm |
| Private space / launch | Emerging | ★ Leader | Strong | US (SpaceX) | Skyroot orbit (Jul 2026) confirms viability. 285 active startups. Scale is next. |
| Deeptech commercialisation | Weak | ★ Leader | Strong | Israel (8200 ecosystem) | IP filing is high. Commercial exploitation infrastructure does not match. |
| Industrial robotics | 4 per 10k workers | Strong | Strong | S. Korea (631 per 10k) | Massive productivity gap = massive opportunity for low-cost collaborative robots |
| Govt procurement speed | Slow | Better | Medium | Israel / US (DARPA) | iDEX helps but structural reform needed for startups to survive on government contracts |
India's structural advantage is not in any single technology layer. It is in the combination: open public digital infrastructure at population scale, 20% of global chip design talent, a domestic market large enough to validate at scale without needing immediate international expansion, and — uniquely — a policy environment that is actively building the physical technology stack underneath the digital one.
The US has better models. China has more compute. Neither has open public infrastructure at this scale without a private platform gatekeeper extracting rent at every layer. Estonia has better government digital services but a population of 1.4 million. South Korea has better manufacturing but no open commerce rails.
India's opportunity is therefore not to replicate Silicon Valley, Shenzhen, or Taipei. It is to build companies at the intersections of these unique attributes — where the combination creates advantages that don't exist in exactly the same form anywhere else.
The opportunity I see is not simply "AI in India," "digital India," or "deeptech in India." It is the layer where these systems begin interacting — where intelligence, execution infrastructure, and physical sensing converge.
The companies I find most interesting are not necessarily the ones building the most impressive technology in isolation. They are the ones that turn these infrastructure shifts into dramatically better economic outcomes for businesses and people who previously couldn't access those outcomes at all.
India's unusual combination — open public infrastructure at population scale, growing physical technology capacity, 1.4 billion people as both a validation market and a distribution channel, and engineering talent at competitive cost — creates a set of structural advantages that don't exist in exactly the same combination anywhere else. That combination is not inevitable to compound. But the companies that figure out how to sit at the intersections will build something that is very hard to replicate.