Real Estate Thinking

Data Centers

A data center looks like a warehouse but is priced in megawatts, not square feet. It teaches that every technology rests on physical infrastructure, and when demand for the digital abstraction surges, the bottleneck migrates to the scarcest physical input, which right now is power.

  • Advanced
  • 8 min total
  • 11 chapters

What decision this helps you make: How to read data-center economics (power as the binding constraint, region concentration, obsolescence), and how the bottleneck migrates to the scarcest physical input in any tech boom.

What this topic is

Buildings housing the servers behind the internet, cloud, and AI, priced and constrained by power and cooling capacity (megawatts), not by square footage.

Why it matters

It reveals the physical bottleneck under the digital economy: when compute demand surges (AI), value migrates to the scarcest physical input, which is power, making that input, not the technology, the durable position.

Who should learn it

Anyone thinking about infrastructure investing or the physical substrate of technology, and anyone learning to find where a boom's bottleneck actually sits.

What you will understand

  • The constraint is power (megawatts), not space
  • Value concentrates where power, cooling, and fiber are cheap and available
  • Tenants sink huge capital in equipment: sticky, but buildings are specialized
  • Obsolescence is fast as compute density and cooling evolve

Prerequisites

Common misconception

"A data center is just a high-tech warehouse, priced per square foot." It's priced per megawatt. The scarce input is reliable, affordable power (and the cooling and grid capacity to use it), which is why data centers cluster in a few power-rich regions rather than near cities, and why securing grid interconnection, not finding land, is the defining bottleneck of the AI era.