AI Semiconductors and Geopolitics
This page tracks chips, lithography, national AI strategy, productivity, and the hardware/geopolitics around AI scaling.
Sources in this batch
- Rapidus, EUV reverse-engineering claims, and Nvidia moat analysis are semiconductor/geopolitics sources.
- Taalas’s ubiquitous-AI piece and a video on the world’s most important machine sit in the accelerator/compute narrative.
- Dallas Fed and DeepMind/UK government sources address macroeconomic and policy implications.
Research interest
The surprising angle is that AI progress may be constrained as much by semiconductor supply chains, lithography, and national policy as by algorithms. For research planning, this matters because compute availability shapes which experimental programs are feasible.
Open questions:
- How robust is the Nvidia moat to open models, custom silicon, and national semiconductor programs?
- Does edge/local AI reduce strategic dependence on frontier datacenters or increase total demand?
- Which algorithmic ideas reduce compute dependence enough to matter geopolitically?