Types of data centers and the land each one needs

What the types of data centers are

The types of data centers are five common categories: hyperscale, colocation, enterprise, edge, and powered shell. The labels describe a business arrangement more than a building. Two facilities can look alike from the road and still belong to different types. The type depends on who owns the facility and who runs computers inside it.

How the types differ by owner and user

The types differ in two answers: who owns the facility, and who uses it.

• Hyperscale. One very large operator, such as a cloud or AI company, uses the whole facility. The operator often owns it, and sometimes leases it from a developer. See hyperscale data center.
• Colocation. A provider owns the facility and leases space, power, and cooling to many customers. The customers bring their own servers.
• Enterprise. A company owns a facility and uses it for its own business systems. Banks, hospitals, and manufacturers are typical owners.
• Edge. An operator places a small facility close to users, so data travels a shorter path.
• Powered shell. A developer builds the structure and delivers power. A tenant fits out the inside and runs it.

Some facilities fit more than one label. A powered shell leased to a hyperscale operator is both.

What land each type needs

Each type needs a different kind of land, and power decides most of the difference.

• Hyperscale needs large connected tracts with high-voltage power and room for later phases.
• Colocation needs sites near fiber and customers, with power for a building or a small campus.
• Enterprise needs a site near the company’s operations, often on land the company already holds.
• Edge needs small parcels close to population and fiber, with modest power.
• Powered shell needs land where a utility can serve a building before the tenant is known.

When the type matters for a site

The type matters as soon as you ask which buyers a tract can serve. A small parcel near a town center with strong fiber may suit edge or colocation and never suit hyperscale. A large rural tract near transmission may suit hyperscale and never suit edge.

Reading the type first keeps a search on the buyers who can use the land. It also keeps a seller from pricing a tract against the wrong market.

Who decides the type

The buyer or developer decides the type, not the land. The same tract can draw a colocation provider in one year and a powered shell developer in another. Each buyer reads the site against its own business. See who buys data-center land.

What the public record shows about type

The public record shows parcel size, zoning, nearby transmission, and mapped fiber in some areas. From those facts you can judge which types a tract might suit. The record cannot show whether a utility will serve the load, whether a board will approve the use, or which buyers are searching nearby.

The serving utility, the local planning office, and the buyer answer those questions. We read the record, mark what it cannot settle, and name who does. See what we check for power and substations and data-center site selection criteria.

Questions

Is a powered shell a type of data center or a way to build one?

It is both. A powered shell describes how a developer delivers a building. The tenant inside can be a hyperscale operator, a colocation provider, or an enterprise.

Which type of data center needs the most land?

Hyperscale campuses usually need the most land, because they draw the largest loads and hold room for later phases. The acreage depends on the design and the load.

Can one campus hold more than one type?

It can. A developer can lease powered shells to different tenants on one campus, and a colocation provider can sell a whole building to one large customer. Zoning and power decide what fits.