Current AD staging frameworks (A/T/N) assume a universal amyloid-first cascade, but clinical heterogeneity in biomarker ordering suggests AD comprises mechanistically distinct subtypes. This hypothesis proposes that SORL1/BIN1/PICALM variants generate endosomal trafficking failure that drives amyloidogenic APP sorting as the primary event, while NTRK1/APOE variants generate cholinergic trophic failure and tau phosphorylation without obligate early amyloid accumulation. Confirming this would reframe staging, trial stratification, and intervention timing. The challenge requires: (1) unsupervised clustering of ADNI/BioFINDER longitudinal biomarker trajectories stratified by polygenic risk for each axis; (2) demonstration that SORL1-risk carriers accumulate amyloid before tau, while high NTRK1-pathway risk carriers show tau and cholinergic markers before amyloid; (3) survival analysis for MCI conversion by subtype × intervention (anti-amyloid vs. neuroprotection). Falsifiable prediction: in ADNI-3 (n ≥ 500 MCI), patients in the top quartile of SORL1/BIN1/PICALM polygenic risk should show amyloid-positive/tau-negative prevalence ≥2× higher at baseline than NTRK1/APOE4-high patients matched for MMSE. Bounty tier: $750K biomarker stratification / AD subtype discovery.