Skip to main content
Invest AI
From market noise to cited evidence

AI Industry Map

24 AI supply-chain layers · 16 public-equity covered · 3 private-market · 5 classification-only.

Descriptive positioning only. The positioning label says whether a company's business is exposed to a layer, never whether to buy it. Layers with no exposure are not recommendations to add any.

1AI Energy

2Semiconductor Materials & Equipment

3Compute & Memory

4Networking & Optical

5Power & Cooling

6Data Center & GPU Cloud

7Data & Model Infrastructure

8Agent & Security

9Enterprise & Vertical AI

10Physical AI

How demand propagates (optional diagram)
AI supply chain flow: power, data centers, chips, cloud, robots

Most AI investing stops at the visible end — a GPU designer and a chatbot. The chain runs much further in both directions. 24 AI supply-chain layers in 10 tracks: 16 with public-equity companies named, 3 private-market (pure plays not listed), 5 classification-only (we will not publish a contestable ticker list as fact). Empty does not mean unresearched.

  1. AI demand
  2. More accelerators
  3. More HBM → advanced packaging → wafers → equipment → materials
  4. More interconnect and optics
  5. More racks → power → cooling
  6. More datacenter capacity → rentable compute
  7. More models, data and agents
  8. More enterprise, vertical and physical applications

Watch It Decide — Four AI Stocks, One Map

I Checked Four AI Stocks on One Map — One Sits on Three Layers, Two Sit on Just One

Four real runs on this map, filmed exactly as they happened. Micron sits on one layer — HBM / Memory — and its page cites zero public 13F rows. TSMC sits on one layer, Foundry, with seven public 13F rows from seven filers. Broadcom spans three layers — Compute, Custom Silicon and Networking — and its own page insists three layers is position, not a ranking. NVIDIA, the giant, sits on one layer, Compute — the layer that cannot run alone. Layers locate; they do not rank. Research and education only — not financial advice, no price target, no buy or sell call.

Watch on YouTube

Watch It Decide — Three on One Layer, One on Two

I Checked Four AI Stocks on One Map — Three Sit on One Layer, the Two-Layer One Makes No Chips

Four more real runs on this map, filmed exactly as they happened. Micron sits on one layer — HBM / Memory — with zero public 13F rows cited, and the map covers US-listed tickers and ADRs only, so SK hynix and Samsung are left off it entirely. Vertiv is the only one on two layers — Power Infrastructure and Cooling — with four products on file, and its own page insists two layers is position, not a ranking; it makes no chips at all. ASML sits on one layer, Semiconductor Equipment, and its page says some steps have effectively one supplier worldwide. TSMC sits on one layer, Foundry — the narrowest point in the entire chain — with seven public 13F rows from seven filers. Layers locate; they do not rank. Research and education only — not financial advice, no price target, no buy or sell call.

Watch on YouTube

Common questions

What is the AI Industry Map?

It splits the AI supply chain into 24 layers across 10 tracks, from energy and materials through chips, networking, power, cloud and applications. Each layer names companies whose primary business sits there, with a public source and a checked-on date. It is research context, not a buy list.

Does being on the map mean a stock is recommended?

No. Layer placement is XingAI's own taxonomy, not a rating or ranking, and the filers did not endorse it. Companies that span several constraints appear on more than one layer. Nothing here is a price target or investment advice. Each company role links a public source.

Where do the fund holdings on the map come from?

From public SEC Form 13F-HR filings, which are lagged, long-only US equity reports filed after quarter-end. A listed position may already have been sold. Each company role links its public source with the date it was checked. It is research context, not a current holdings book.

How is the map organized?

Layers are grouped into 10 tracks, and each layer is its own page listing the companies on it. A company whose business spans several layers appears on each layer it spans, and every company role links its public source with the date it was checked.

Why do some layers list no companies?

3 layers are private-market pure plays that are not listed, and 5 layers are classification-only — we will not publish a contestable ticker list as fact. An empty layer is a deliberate boundary, not missing research.

Is the AI Industry Map free to read?

Yes. This page and the layer pages linked from it are public — no account is needed to browse the layers, the companies on them, and the sources cited. It is research context, not investment advice.

Weekly digest: AI-map layer changes and new public 13F filings, each with its source. No score, no grade, no trade signal. Unsubscribe anytime.