AI hardware · Layer

Data centers

The buildings that house AI systems, supplying power, networking, and cooling at scale. AI has shifted data center design toward far higher power density per rack and far larger total site power.

Key facts

AI racks draw many times the power of traditional servers, pushing sites toward tens or hundreds of megawatts.
The constraint for new capacity is increasingly the electrical grid connection and power availability, not the building itself.
Hyperscalers (Microsoft, Google, Amazon, Meta) and specialized operators are building dedicated AI campuses.
Siting decisions now hinge on access to cheap, abundant, and ideally low-carbon power.

Where it bottlenecks

Grid power and interconnection timelines are the dominant limit on new AI data center capacity in many regions.

Who dominates it

MicrosoftGoogleAmazonMeta

Companies in this layer

Western Digital

United States

Hard disk drive (HDD) and storage solutions for AI data centers

All companies across the chain

Explore other layers

GPUs and AI accelerators High-bandwidth memory (HBM)Chip foundries and fabrication Advanced packaging Lithography and fab equipment Networking and interconnect AI servers and systems Cooling and thermal management Supply constraints and bottlenecks

Back to the AI hardware map

Let's get to work

Want to build with AI for real?

Beyond the explainer, we design, secure, build and run production AI. Tell us what you have in mind.

Start a project See the offer