Issue IV: The Compute Constraint

This thesis examines the structural case for power-advantaged inference compute infrastructure — the bottleneck that determines competitive positioning in the next phase of AI deployment.

Preface available below. Full thesis available to members.

PREFACE  The Inference Transition

The AI infrastructure narrative has been dominated by training. The next phase is inference, and inference has a different cost structure and a different set of winners. Training compute is becoming a commodity. Inference compute located where power is structurally cheap is the bottleneck that does not resolve.