AI is becoming a core infrastructure — like electricity — built on five interdependent layers.
At the base sits compute (chips, data centers), followed by data, then models, then applications, and finally distribution. Each layer reinforces the others, creating powerful network effects and high barriers to entry.
What makes this shift structural is that AI is no longer a standalone sector — it’s embedding into every industry, much like electricity did a century ago.
The implication: value will not accrue evenly. Bottlenecks — especially in compute and data — are likely to capture outsized economics, while the application layer drives adoption and demand.
Layers:
- compute (chips, data centers)
- data
- models
- applications
- distribution
