“The scientific community has a precise empirical observation of AI scaling laws but lacks a fundamental, first-principles understanding of why they work.”