✓ Source-verified - Verified from SEC filings, annual reports, official disclosures
Inferred - Industry-standard relationships based on structural dependencies
Confidence levels: High (company disclosure), Medium (source-backed), Low (industry-inferred)
Filters
Filter by relationship confidence (High/Medium/Low)
Filter by entity type (Supplier/Service/Channel/Demand)
Filter by verification status (Source-backed vs Inferred)
About
Top 100 companies by market capitalization. Data sourced from companiesmarketcap.com with supply chain relationships verified from primary sources including SEC 10-K filings, annual reports, and official company disclosures.
All 100 companies have source-verified profile data. 103 automated tests ensure data quality.
Methodology & Data Sources
Data sources
Profiles for the top 100 companies cite 407 sources in total. Of these, roughly 131 sources carry a parseable publication year, which is what powers the age-decay signal below. Sources include SEC 10-K filings, annual reports, and official company disclosures.
Confidence weighting
Every figure inherits a confidence tier from the source backing it. The dataset carries two real disclosure tiers — 120 high-qualifier figures (direct company disclosure) and 3,447 medium-qualifier figures (source-backed). A figure with no resolving source falls to an explicit Unknown floor rather than being scored upward.
Observed — high (company disclosure): base weight 90.
Estimated — medium (source-backed): base weight 65.
Source-backed figures decay with the age of their source: an exponential half-life of about 4 years measured against the dataset generation date, floored at half weight so old-but-real sources are never zeroed. When no publication year can be parsed, no decay is applied — absence of a date is not treated as evidence of staleness.
What the badges mean
Observed — backed by a high-disclosure source.
Estimated — backed by a source-backed reference, not direct disclosure.
Unknown — no resolving source; shown explicitly rather than guessed.
Supplier concentration (derived)
Each company gets a 0–100 Supplier concentration score — a derived aggregate, badged Derived (never Observed). It is a composite of two real graph signals:
C = round(100 · (0.6 · HHI + 0.4 · sharedFrac))
HHI is the Herfindahl index over a company’s suppliers. The dataset has no per-supplier volume field — every supplier link weight is the constant l.v = 2 — so HHI is taken under the equal-weight assumption, where it reduces exactly to HHI = 1/k for a company with k distinct suppliers. This equal-weight limit is stated explicitly: without volumes, all k-supplier firms share the same raw HHI, so HHI alone cannot differentiate them.
sharedFrac is the fraction of a company’s suppliers that are shared single-points (a supplier depended on by more than one company), which breaks the equal-weight HHI tie using real structure.
Critical chokepoints (derived)
The chokepoints ranking is the supplier fan-in: the real count of companies depending on each supplier, sorted descending. The top chokepoint is depended on by 4 companies. This is a derived aggregate computed from the graph, kept separate from the curated editorial bottleneck flag.
Scenario stress-tests (derived)
A scenario disables one or more real suppliers and lists the companies that lose at least one supplier as a result. Like the concentration and chokepoint figures, scenario output is a derived aggregate, badged Derived — never Observed. It is a structural what-if over the graph, not a forecast.
Bounded multi-hop cascade. The model starts from the direct dependents (hop 1) of the disabled suppliers, then propagates: a firm impacted at one hop that is itself a real supplier disrupts its dependents at the next hop (hop 2+). Traversal is cycle-safe — a visited set guarantees each firm is reached once, so the cascade always terminates — and bounded to at most N hops (the UI runs N=3). Real second-order edges are sparse, so the multi-hop result often equals the direct-dependent set; no edges are fabricated.
Concentration shift is reported as HHI = 1/k, which rises when a supplier is removed (fewer distinct suppliers → more concentrated). The composite concentration score above is not used here because it is not monotonic under removal — dropping a shared supplier can lower its sharedFrac term even as the firm becomes more concentrated.
“Market cap exposed” is exposure, not a loss estimate. It is the combined market capitalization of the impacted firms — the dollar value of companies touched by the disruption, not a projected loss or write-down.
The Taiwan semiconductor preset disables the real TSMC supplier-label variants found in the data (TSMC is fragmented across several normalized labels), so the headline reflects the true union of dependents rather than a single label.
Known limits
About 75 dangling source references (source FKs that do not resolve to a listed source) remain in the dataset; figures behind them score at the Unknown floor.
Many relationships are estimated from structural dependencies rather than directly disclosed.
Market caps are auto-updated weekly; the freshness indicator reflects the live update timestamp.
Compare Companies
Supply Chain Intelligence
Part of Monarch Castle Technologies.
0nodes0links0countries0layers$0Tcap
Click any company node to open its dedicated supply-chain profile.
Estimated relationships are labeled and source-linked when evidence is available.
Deterministic scenario — not a prediction.
Top 10 by Market Cap
Filters
--
Company
SYMB
0 suppliers
0 services
0 channels
--
City, Country
Credit Ratings
Risk & Resilience
Low RiskProfile risk summary
Supplier Concentration
Shared Supplier Overlap
Evidence Timeline
Critical Chokepoints
Top suppliers by dependent-company count (derived fan-in)
Scenario Stress-Test
Run a what-if disruption — multi-hop downstream cascade (bounded, cycle-safe)
No scenario run yet.
Interactive supply-chain network. Use Tab to focus a node, then Enter or Space to open it.
Global: click company to open profile · Scroll zoom · Drag pan · Press H for help
