U.S. military power is bounded by assured access to semiconductors, yet the semiconductor supply chain is global, specialized, and opaque beyond first-tier suppliers. This dissertation examines how the Department of Defense can identify and govern hidden dependencies in that supply chain. It constructs a DoD-anchored, provenance-aware network that combines three evidence layers: disclosed supplier relationships from public and commercial sources, predicted links inferred through leakage-safe link-prediction models, and observed links derived from shipping data. Together, these layers produce a decision-grade map of semiconductor dependence rather than a complete census of the supply chain. Methodologically, the dissertation develops a reproducible workflow for extending visibility under partial observability while preserving the distinction between disclosure, inference, and observation. Using fixed candidate pools, temporal evaluation, calibrated ensembling, and conservative operating rules, it recovers likely hidden tiers and prunes the broader supplier network to the defense-relevant semiconductor subgraph. The analysis shows that vulnerability is not evenly distributed. It concentrates in chokepoints, corridor intermediaries, and brittle paths that can deny or degrade semiconductor support to DoD-linked prime contractors if disrupted. The dissertation evaluates structural consequence conditional on disruption through reachability loss, flow-based impact, brokerage, and path redundancy, while treating disruption probability as a separate concept not estimated directly. It concludes with a policy framework centered on decision-grade visibility, traceability, and acquisition practice, so that Tier-2+ dependencies that matter for continuity and trust become governable before crisis exposes them.