Portfolio Supply Chain Concentration: Internal Links vs. Outside Overlap
Published August 25, 2026
Altsets
Research by Altsets Research
Audit ASML, Micron and Nvidia across 17 observed relationships. Separate two internal links from one shared external customer before adding portfolio weights.
Dataset reference:Altsets supply chain relationships dataset covers 115k+ entities, 550k+ relationships, and history since 2006 across the full dataset; individual analyses use a narrower scope.
Key findings
- The selected July 20, 2026 network contains 17 first-degree edges: two internal portfolio links and 15 outside connections.
- ASML to Micron to Nvidia creates a chain of different company decisions, rather than one equal event exposure.
- Quanta is the sole visible shared external customer in the bounded map, with incomplete economic weights.
To audit portfolio supply chain concentration, separate commercial links between holdings from shared companies outside the portfolio. Those are different routes through which one event can affect several positions. In a hypothetical basket of ASML, Micron and Nvidia, Altsets shows ASML supplying Micron and Micron supplying Nvidia. The portfolio owns both ends of two relationships before any external overlap is counted.
The July 20, 2026 maps contain 17 visible first-degree directed relationships touching this holding set: two internal links and 15 external connections involving 14 outside entities. Quanta Computer is the only visible shared external customer. Second-degree suppliers beneath Shin-Etsu Chemical are excluded. These boundaries matter because widening the graph would change both the count and the question.
The evidence supports a network decomposition, rather than a portfolio loss estimate. There are no account weights, and the Micron-to-Quanta connection lacks comparable displayed economic metrics. Nevertheless, the graph changes how the basket should be monitored: Nvidia purchasing is a direct Micron sales question, while any subsequent effect on ASML requires a separate Micron equipment-investment decision. Treating all three stocks as equally exposed to the same headline would discard that distinction.
Count internal links before external overlap
| Relationship pattern | Observation | Decision connecting the positions |
|---|---|---|
| Owned supplier to owned customer | ASML to Micron | Micron purchasing relevant equipment from ASML |
| Owned supplier to owned customer | Micron to Nvidia | Nvidia purchasing relevant products from Micron |
| Two holdings to one external customer | Micron and Nvidia to Quanta | Quanta purchasing from each supplier through the observed relationships |
Portfolio supply chain concentration adds commercial structure to position weights, sectors and factors. It does not replace those views. A basket may contain different securities and business models while repeating a buyer, a supplier or a chain of investment decisions. The direction of each link determines what additional evidence belongs in a stress test.
The two owned relationships have different financial bases
ASML to Micron displays 7.64% of ASML revenue, a 3B relationship size and 11.91% of Micron cost of goods sold. Micron to Nvidia displays 17.62% of Micron revenue, a 9.8B relationship size and 14.00% of Nvidia cost of goods sold.
Owning both companies does not cancel either relationship. A purchasing cut can affect the supplier's sales while also reflecting a weaker growth outlook for the buyer. A price increase might redistribute economics, but the account result depends on position weights, margins and product scope. The graph alone supplies none of those offsets.
Nor should the analyst multiply the two revenue shares to estimate ASML exposure to Nvidia. Nvidia orders can affect Micron's use of existing capacity without changing equipment purchases. Conversely, a Micron technology transition can require investment without a corresponding immediate rise in memory sales. The intervening capacity decision belongs in the model.
Micron's supplier cluster supplies context, not a common-supplier result
The overview contains eight visible first-degree suppliers feeding Micron. Five are named as ASML, Applied Materials, Lam Research, KLA and Shin-Etsu Chemical; three additional nodes are ticker-only in the supplied view. The quantified named relationships are:
| Relationship | Supplier revenue | Relationship size | Customer COGS |
|---|---|---|---|
| ASML to Micron | 7.64% | 3B | 11.91% |
| Lam Research to Micron | 5.61% | 1.2B | 5.52% |
| KLA to Micron | 4.59% | 693M | 2.84% |
| Applied Materials to Micron | 2.96% | 990M | 3.84% |
| Shin-Etsu Chemical to Micron | 1.83% | 59.1B JPY | 1.34% |
These suppliers converge on one holding. That observation does not establish that they also supply the other two holdings in this bounded map. It is also insufficient to claim redundancy: equipment and materials vendors need not perform interchangeable functions.
The size labels are reproduced as displayed. Shin-Etsu is explicitly labeled in JPY; the other labels are not assigned an assumed currency here. Supplier revenue percentage, customer cost percentage and monetary size answer different questions and remain separate throughout the audit.
Quanta is one observed shared customer, with incomplete weights
Both Micron and Nvidia connect downstream to Quanta Computer (2382). The Nvidia-to-Quanta edge displays 0.53% of Nvidia revenue and 1.76% of Quanta's cost of goods sold. The Micron-to-Quanta edge is visible without comparable displayed metrics. Missing values remain unknown, so this cannot become a complete customer-weighted concentration calculation.
Nvidia's visible downstream set also includes Amazon, Microsoft, Super Micro Computer and Samsung Electronics. Nvidia to Super Micro Computer shows a 28.4B relationship size. Nvidia to Samsung shows 0.04% of Nvidia revenue and 0.10% of Samsung's cost of goods sold. These observations describe the selected map; they do not establish that any named customer is the largest dependency in a broader account.
Reproduce the boundary before adding a risk score
The holding set is exactly ASML, Micron and Nvidia. Include every visible first-degree supplier or customer edge touching those companies, retain supplier-to-customer direction, classify internal links separately and count a shared external entity only when it connects to at least two holdings. Keep unquantified edges in the structural result without assigning them zero economic weight.
For a real account, add dated holding weights only after this decomposition. Breadth measures how many positions share a counterparty; directional metrics describe estimated economic weight; replacement difficulty requires product, qualification, inventory and capacity research. No map count establishes exclusivity, exact volumes, facility locations or a future stock-price response.
This case therefore establishes two internal commercial links and one external customer overlap. The holdings-to-network comparison explains the ownership join. Use the shared-customer screen for weighted demand overlap, or the external-company watchlist to set monitoring priorities. Altsets methodology describes the estimates.
Sources
Related research
Methodology
Cite this research
Altsets Research. "Portfolio Supply Chain Concentration: Internal Links vs. Outside Overlap." Published August 25, 2026. https://www.altsets.com/research/portfolio-supply-chain-concentration