AI Infrastructure Supply Chain Map
A guided supply-chain explorer for data-center infrastructure. Choose a facility type, inspect the reference blueprint, click components such as transformers, switchgear, batteries, generators, chillers, cooling loops, fiber, and optics, then see origin countries, OEMs, lead times, and bottleneck shape.
What this proves
The physical AI build-out is a component supply-chain problem, but the user needs context first: choose the data-center type, inspect the blueprint, then map the origins and lead-time risks of each component.
How it works
What it does
AI Infrastructure Supply Chain Map starts where Data Center Atlas stops.
The atlas answers: what kind of data-center cluster is this?
This map answers: what physical parts make each data-center type possible, who makes them, how long they take, and where they come from?
Choose a data-center type first: AI mega-campus, hyperscale metro region, constraint-heavy region, or regional cloud cluster. The build then shows a reference blueprint for that facility class. Click a room or system, choose a component such as HV transformers, MV switchgear, UPS battery banks, diesel generators, chillers, CDUs, optical transceivers, or fiber panels, and only then the map draws origin-country arcs into the selected destination cluster.
Why this is separate from the atlas
A user clicking Northern Virginia expects a facility and region story first: MW scale, likely campus shape, power interconnect pressure, cooling options, and reference blueprint.
A user clicking HV transformer expects a bottleneck story: Mexico, Korea, Germany, Turkey, OEMs, lead times, and why the grid equipment queue slows the AI build-out.
Those are related, but they are not the same job. Splitting the build keeps each interaction honest.
Data posture
This is not a real bill of materials for a specific facility. Real BOMs are private.
The component list is a public-data reference model compiled from Open Compute Project specifications, Uptime Institute and ASHRAE guidance, OEM public materials, trade press, SEC commentary, UN Comtrade, and USITC DataWeb. Origin shares are directional because HS codes are broad and include non-data-center uses.
Pattern worth borrowing
When a map has two valid user questions, split the lenses.
- Atlas lens: place, scale, architecture, local constraints.
- Supply-chain lens: components, origins, OEMs, lead times, concentration risk.
Cross-link them, but do not force one flow to pretend it is the other.
Get new builds, breakdowns, and useful AI updates.