The Encyclopedia

The AI Infrastructure
Supply Chain

14 chapters. From the transistor to the money. Every layer of the largest infrastructure build in a generation, explained in plain English.

Start reading Chapter 1 →

The Silicon

Where the compute starts — designing, fabricating, and assembling the chips.

1
The ChipTransistors, lithography, and the companies that design the silicon powering every AI system.
›2
Packaging & MemoryCoWoS, HBM, chiplets, and the server — giving the chip a body and a workspace.
›

The Machine

Connecting, housing, cooling, and powering the hardware.

3
The NetworkSwitches, optical transceivers, and fiber — connecting thousands of GPUs into one machine.
›4
The BuildingSite selection, construction, and the physical structure that houses the compute.
›5
CoolingLiquid cooling, HVAC, and heat rejection — moving energy from the chip to the sky.
›6
Power InsideTransformers, switchgear, UPS, and the 50,000-to-1 voltage cascade to the chip.
›

The Energy

On-site generation and the public infrastructure behind every data center.

7
On-Site UtilitiesGas turbines, nuclear, fuel cells, and renewables — generating power behind the meter.
›8
Grid, Gas & FibreThe public infrastructure a data center still depends on — grid connections, pipelines, and fiber.
›

The Supply

The factories and raw materials that build every layer of the stack.

9
The FactoriesChip fabs, server plants, transformer foundries — the factories that build the buildout.
›10
The MaterialsSilicon, copper, rare earths, and the raw materials behind every layer of the stack.
›

The Operations

The software and companies that run the machines.

11
The SoftwareMonitoring, scheduling, networking, and security — the invisible layer that runs the hardware.
›12
The OperatorsHyperscalers, REITs, NeoClouds — the companies that own and rent out the compute.
›

The Economics

The bottlenecks, capital flows, and whether the math works.

13
The QueueLead times, permitting, labor, and the bottlenecks slowing everything down.
›14
The MoneyCapital spending, financing structures, and whether the math works.
›

Every chapter includes NotebookLM diagrams, company spotlights, and chokepoint analysis. Written in plain English for anyone smart enough to be curious.