How to Find Second-Order Supply-Chain Exposure
August 2, 2026
Altsets
Research by Altsets Research
Expand beyond direct suppliers and customers to identify economically meaningful two-hop paths without turning network proximity into a false exposure score.
Data used:Altsets Supply Chain Intelligence: 90k+ entities, 400k+ relationships, 20+ years of history.
Key findings
- Second-order analysis is most useful when both hops are economically meaningful and the business context plausibly connects the path.
- SK ecoplant to SK Hynix to Nvidia and equipment suppliers to Micron to Nvidia provide real examples of meaningful two-hop paths.
- Supplier revenue and customer cost percentages should not be multiplied into a synthetic indirect exposure without a separately justified model.
To find meaningful second-order exposure, move one relationship beyond a company's direct supplier or customer, then rank the longer paths by economic weight and verify how the companies are connected. First-degree supply-chain maps answer who directly supplies or buys from a company; second-order analysis asks what matters one layer further out.
Second-order exposure asks what sits upstream from a major supplier or downstream from a major customer, then uses relationship data to decide whether the longer path is worth investigating. Two Altsets networks show why this matters.
One path runs from SK ecoplant to SK Hynix, then from SK Hynix to Nvidia. The first relationship is estimated at 36.49% of SK ecoplant revenue, 3.1B USD, and 18.81% of SK Hynix COGS. The second is estimated at 27.88% of SK Hynix revenue, 21B USD, and 27.33% of Nvidia COGS.
Another path runs from semiconductor equipment suppliers into Micron Technology, then from Micron to Nvidia. The ASML to Micron relationship is estimated at 7.64% of ASML revenue, 3B USD, and 11.91% of Micron COGS. Micron to Nvidia is estimated at 17.62% of Micron revenue, 9.8B USD, and 14.00% of Nvidia COGS.
The point is not to multiply the percentages. The point is to identify economically meaningful paths that a one-hop company screen would miss.
What second-order exposure is trying to find
A first-degree supplier screen might tell you that Nvidia depends on SK Hynix or Micron.
A second-order screen asks: Who matters upstream from those suppliers? Which of those upstream relationships is economically meaningful? Could a disruption, demand signal, or capacity constraint travel through more than one mapped relationship? Which paths are strong enough to justify deeper research? This is useful because the most visible company in a theme is not always the origin of the economic dependency.
Example: looking beneath an AI memory supplier
SK Hynix is a direct supplier to Nvidia in the displayed network. Stopping there misses the SK ecoplant to SK Hynix relationship one level upstream.
The upstream edge is estimated at 36.49% of SK ecoplant revenue and 18.81% of SK Hynix COGS. Those are large enough to make the relationship worth investigating on its own.
The direct SK Hynix to Nvidia edge is also large, at 27.88% of SK Hynix revenue and 27.33% of Nvidia COGS. Together, the two edges create a path with meaningful economic exposure at both hops.
That does not prove that an SK ecoplant event would affect Nvidia. It says the path is economically substantial enough to investigate rather than dismiss as graph noise.
Example: looking beneath Micron
The same method works in a different network.
Micron is mapped as a supplier to Nvidia. Several equipment and materials companies sit upstream from Micron. Among the quantified edges:
- ASML to Micron: 11.91% of Micron COGS, 3B USD;
- Lam Research to Micron: 5.52% of Micron COGS, 1.2B USD;
- KLA to Micron: 2.84% of Micron COGS, 693M USD;
- Applied Materials to Micron: 3.84% of Micron COGS, 990M USD;
- Shin-Etsu Chemical to Micron: 1.34% of Micron COGS, 59.1B JPY.
Micron to Nvidia is then estimated at 17.62% of Micron revenue, 9.8B USD, and 14.00% of Nvidia COGS. A second-order screen can use those first-hop cost percentages to prioritize which upstream Micron relationships deserve attention when researching a downstream Nvidia exposure.
Why you should not multiply the percentages
It is tempting to turn a two-hop path into one synthetic exposure number by multiplying percentages. That is usually not justified.
Supplier revenue percentage and customer cost percentage use different company-specific denominators. They do not behave like portfolio weights. For example, multiplying ASML's share of Micron costs by Micron's share of Nvidia costs would create a number, but that number would not be a validated estimate of Nvidia's indirect cost exposure to ASML.
The safer use is ordinal: is the first hop economically meaningful? is the second hop economically meaningful? is the business connection plausible? does the path deserve further work?
The network prioritizes paths. It does not automatically quantify complete transmission.
When second-order analysis is useful
Second-order exposure is especially useful for disruption research, geopolitical or geographic shocks, semiconductor capacity constraints, earnings propagation, commodity exposure, portfolio hidden concentration, identifying overlooked beneficiaries or losers, and tracing a product cycle through upstream suppliers. It becomes less useful when the graph contains weak or unquantified relationships, when the intermediate company has many alternative inputs, or when the path depends on unrelated business segments.
A disciplined two-hop workflow
A practical workflow is:
- Start with the company or event you care about.
- Map direct suppliers or customers.
- Rank first-degree relationships using the metric that matches the question.
- Expand only the economically meaningful first-degree nodes.
- Rank the second-degree relationships.
- Check whether the business context connects the two hops.
- Do not multiply percentages unless a separate model justifies it.
- Record the path as a hypothesis for deeper investigation.
This keeps the graph from turning into a giant collection of technically connected but economically irrelevant nodes.
What a strong two-hop path looks like
A useful two-hop path usually has several qualities: both edges are mapped and directional, at least one relationship metric is available on each hop, the values are economically meaningful relative to the companies involved, the business roles make sense across the path, and there is a concrete event or research question motivating the expansion.
The SK ecoplant to SK Hynix to Nvidia path satisfies those conditions in the displayed data. The equipment supplier to Micron to Nvidia paths provide another set of examples with different industries and relationship sizes.
The SK ecoplant and SK Hynix analysis examines the upstream relationship itself. The Micron and Nvidia analysis examines the downstream customer connection. This guide explains the method for combining those kinds of observations into a second-order research queue without pretending the path is a single measured exposure.
For relationship direction and metric limitations, read the Altsets supply-chain data methodology. Browse Supply-Chain Data Use Cases for other research methods.