Investment Constitution

Core Thesis

Use Leopold Aschenbrenner's lens: do not start from "which AI app is cool"; start from "if AGI-scale buildout happens, what physical, operational, security, or coordination bottleneck must be solved, and where is supply slow to respond?"

Primary industry to deep-dive:

AI data center power and electrical infrastructure

Why this is the first target:

• AI model progress converts into demand for huge compute clusters.
• Huge compute clusters convert into demand for power, transformers, switchgear, cooling, permitting, interconnection, and site control.
• These constraints have long lead times and cannot be fixed purely with software.
• The broad "AI data center" theme is becoming consensus, but the second-order winners inside the power/electrical stack are less settled.

Research Question

Which companies become structurally more valuable if AI compute demand keeps rising faster than grid connection, transformer supply, cooling capacity, and local permitting can expand?