Data Center Atlas

What it does

Data Center Atlas maps the world's major AI and cloud data-center clusters, then turns each click into a facility brief instead of a generic bubble.

Layer 0 (World). A dark world map with glowing pillars at major data center clusters: Northern Virginia, Frankfurt, Singapore, Mumbai, Dallas, Phoenix, Tokyo, and more. Pillar height is proportional to approximate cluster capacity.

Layer 1 (Cluster brief). Click any cluster and the app shows the likely facility class, MW scale, regional constraint, power architecture, and cooling technology mix. Northern Virginia reads like an AI mega-campus problem. Singapore reads like a land/power/water constraint problem. Smaller clusters read like regional or edge capacity.

Layer 2 (Reference blueprint). From the cluster brief, open a reference blueprint for that facility class. The blueprint is intentionally marked as a public-data model, not a private real-estate drawing.

Supply-chain handoff. Component-origin arcs now live in the separate AI Infrastructure Supply Chain Map, because that is a different question: parts, OEMs, lead times, and origin concentration.

What this is not

This is not a real blueprint for any specific private facility. Actual site plans, MEP drawings, utility interconnect agreements, and equipment schedules are private.

The atlas is a public-data reference model. It uses approximate cluster capacity, public hyperscaler announcements, regional infrastructure constraints, and standard facility patterns to infer a likely facility class.

Why this exists

Most data center visualizations on social media show locations as bubbles on a map. That story is saturated.

The more useful question is what kind of data-center build a location implies. Northern Virginia should feel different from Singapore. A 2,200 MW AI campus should feel different from a 180 MW regional cluster. Power, cooling, land, interconnect pressure, and campus shape change with the place.

That is why the first drilldown is a cluster brief: facility class, MW scale, likely cooling mix, power architecture, and the local constraint. From there, a user can open a reference blueprint or jump into the separate supply-chain lens.

Data sources

All public, all free.

• Public hyperscaler announcements for regional expansion signals
• CBRE, JLL, and Synergy-style market summaries for cluster-scale cross-checks
• Open Compute Project specifications for reference architecture
• Uptime Institute Tier whitepapers
• ASHRAE TC 9.9 guidance for thermal envelope and cooling context
• EPA GHG Reporting Program for facility-level context where available
• Wikipedia "List of data centers" for cluster cross-reference

How to read it

Start at the world view. Pick a cluster you have a reason to care about: Northern Virginia for the AI epicenter, Mumbai for fast growth, Frankfurt for the EU hub, Singapore for constraint pressure.

Read the cluster brief first. The brief should answer: how large is this footprint, what class of facility is it likely to be, what cooling technologies make sense, and what local constraint dominates?

Open the reference blueprint when you want to see the facility anatomy: electrical yard, generator yard, cooling plant, white space, network meet-me room, and operations layer.

Jump into the supply-chain map only when your question changes from "what is this facility?" to "where do its physical parts come from?"

Tech stack

Next.js App Router, Mapbox GL JS for the map, deck.gl for the capacity pillars, hand-written isometric SVG for the blueprint, and a React state machine for the atlas, facility brief, blueprint, and supply-chain handoff. All data is pre-fetched and stored as JSON in public/data/inside-dc/ so there are no live API costs in production.