AI is not purely digital infrastructure. It is an industrial system.

Last year, went down the rabbit hole on earth materials and how they fit into the theme of supply chain demand, particularly with the growth of compute required AI. Defense contracts heavily depend on lithium, especially for electronic components and warfare applications. Currently, the US mainly rely on outsourcing lithium from countries such as China and Australia.

The interesting realisation is that the bottlenecks increasingly look less like traditional software constraints and more like industrial coordination problems.

A large AI cluster is a physical object. It requires: high electrical load, specialised cooling systems, high-purity quartz, advanced packaging, copper-heavy interconnects, tin solder (?), lithium-backed energy systems, and, globally distributed mineral processing chains.

Which makes the geopolitical structure unusually asymmetric.

Core Components of AI Compute

At the heart of systems; powerful computing units that require a variety of minerals and materials:

1. Central Processing Units (CPUs) and Graphics Processing Units (GPUs):
   • Require silicon, copper, and gold
   • Key players:
     • NVIDIA: Dominates the AI GPU market
     • Cerebras Systems: Developed the world's largest AI chip, the WSE-3
     • Graphcore: Creating Intelligence Processing Units (IPUs) for AI workloads

• High-performance CPUs
• Large amounts of memory and storage
• High-speed interconnects
• Power supply and cooling systems

Raw materials required:

• Silicon for chip fabrication
• Rare earth elements for chip components
• Lithium, cobalt, nickel for batteries
• Copper for wiring and interconnects
• Gold, silver, palladium for connectors

Deeper Dive:

• Central Processing Units (CPUs) and Graphics Processing Units (GPUs): The brains of computing, requiring silicon, copper, and gold.

  • NVIDIA: Dominates the AI GPU

  • Cerebras Systems: Developed the world's largest AI chip, the WSE-3

    

  • Graphcore: Creating Intelligence Processing Units (IPUs) for AI workloads

• Semiconductors:

  • Essential for all electronic devices, relying heavily on silicon, germanium, and gallium arsenide.

• Memory Storage Devices:

  • Utilising materials like cobalt, nickel, and rare earth elements.
  • Western Digital offers high-capacity storage solutions for AI data centers.

• Printed Circuit Boards (PCBs):

  • Incorporating tin, copper, and gold for electrical connections.
  • TTM Technologies: Major PCB manufacturer
  • Flex: Provides PCB assembly services for AI hardware

  

Material Composition in AI Hardware