AI Infrastructure
Artificial intelligence is a physical industry. It runs on silicon, buildings, transformers, turbines and crews — and every one of those has a lead time. This is the research on what AI is actually built from, and who gets paid to build it.
The universe
The full map of who supplies the build-out, and how much of each company is genuinely levered to it.
The Powering AI Universe
376 companies across a five-layer infrastructure stack, each scored on exposure purity, input scarcity and time to revenue — so the pure-plays separate from the merely adjacent.
Interactive · IndiaIndia's AI Enablers
The Indian half of the map: 41 listed names across power generation, grid equipment, data centres and electronics, with valuation, quality, cash-conversion and analyst-expectation numbers, and how the basket has traded against the Nifty 50, Nifty 500, Midcap 150 and Smallcap 250.
Valuing the AI labs
The model builders are going public. Price is what the market sets; value is a story you have to be able to defend.
Lumentum — the photonics bottleneck
One company, taken apart in five parts: what it makes, why NVIDIA bet on it, what it is worth, and whether an Indian investor can own it.
Lumentum — Business Study
The full series, from the physics of optical transceivers through to the valuation and the India angle.
Part 3 · Most readNVIDIA's $2 Billion Bet
Why photonics is AI's hidden bottleneck — and what NVIDIA's investment says about where the constraint really sits.
Sector StudySemiconductors
The wider sector view that the Lumentum work sits inside — supply chains, chokepoints and who controls them.
Full reports
Complete research documents and IC-style memos, free to download and cite with attribution.
A note on scope: most of Vatayan Labs looks at how India moves money. This section deliberately does not — the AI build-out is a global supply chain, and the companies that matter are listed in the US, Europe, Japan, Korea and Taiwan. Where there is an India angle, it is called out explicitly.