How to Separate Semiconductor Capex Exposure From Product Demand
July 21, 2026
Altsets
Research by Altsets Research
Split a semiconductor network into upstream manufacturing suppliers and downstream customers so capex events and demand events are not treated as the same signal.
Data used:Altsets Supply Chain Intelligence: 90k+ entities, 400k+ relationships, 20+ years of history.
Key findings
- Micron's upstream equipment relationships point toward fab-spending and manufacturing-capacity research, while downstream customers point toward product and end-market demand.
- ASML, Lam Research, KLA, and Applied Materials occupy different semiconductor-manufacturing functions, so the same Micron event should not be propagated through every supplier identically.
Separate semiconductor capex exposure from product demand by reading upstream equipment and materials relationships as manufacturing-investment paths and downstream customer relationships as demand paths. A semiconductor company's supply-chain graph can carry both signals at the same time.
Upstream relationships can point toward fab spending and manufacturing capacity. Downstream relationships can point toward product demand and end-market activity. Micron is a useful example because the supplied network shows both sides.
On the upstream side, quantified relationships include ASML, Lam Research, KLA, and Applied Materials. On the downstream side, the graph includes Nvidia plus Taiwanese computer and server manufacturers identified by their exchange codes: Inventec (2356), Quanta Computer (2382), and ASUS (2357).
The important conclusion is not that every node reacts to the same Micron event. It is the opposite: the direction of the relationship tells you what kind of event should matter.
Upstream suppliers are often a capex question
ASML, Lam Research, KLA, and Applied Materials all sell technology used in semiconductor manufacturing. ASML describes its lithography systems as tools used to mass-produce chip patterns, with DUV systems used for advanced logic and memory manufacturing. Lam Research sells wafer-processing technology across deposition, etch, cleaning, and advanced memory applications.
KLA develops semiconductor process-control and yield-management systems, including inspection and metrology. Applied Materials sells semiconductor manufacturing systems used for deposition, etch, patterning, materials modification, and analysis. That product context changes how an investor should read an upstream Micron edge. A Micron production or capital-spending decision may be more relevant than a short-term unit shipment change.
The economic weights are not equal
The displayed supplier relationships also have different directional percentages. For the Micron relationships shown in Altsets:
| Supplier into Micron | Supplier revenue percentage | Micron cost percentage |
|---|---|---|
| ASML | 7.64% | 11.91% |
| Lam Research | 5.61% | 5.52% |
| KLA | 4.59% | 2.84% |
| Applied Materials | 2.96% | 3.84% |
Those percentages help prioritize which manufacturing suppliers deserve deeper work after a Micron capex event. They do not prove which supplier would experience the largest order change. Equipment orders can be lumpy, delayed, cancelled, pulled forward, or tied to different process steps.
The metric tells you where the commercial relationship is economically visible. Public company research tells you what the supplier actually does.
Downstream customers are a demand question
The downstream side of the Micron graph has a different interpretation. Micron is connected to Nvidia, Inventec, Quanta, and ASUS in the supplied network. Those companies sit much closer to computing, servers, and end-market hardware demand than the upstream semiconductor-equipment suppliers do.
Inventec describes an enterprise business that includes AI servers, general-purpose servers, storage, and networking systems. Quanta sells server, storage, and network-switch platforms for data-center customers. ASUS offers servers, AI infrastructure, storage, and GPU systems. That makes a Micron-to-customer edge useful for a different question: what demand environments are connected downstream from Micron?
Nvidia is the quantified downstream anchor
Among the displayed downstream relationships, the Micron-to-Nvidia edge is quantified at 17.62% of Micron revenue and 14.00% of Nvidia's cost base. That makes Nvidia a much stronger economic read-through anchor than a structural-only customer edge. Inventec, Quanta, and ASUS remain valuable because they reveal the type of downstream network Micron participates in.
But without comparable percentages in the supplied view, those edges should not be ranked against Nvidia numerically. This is a good example of why topology and magnitude should remain separate.
One Micron headline can point in two directions
Suppose Micron announces stronger memory demand but keeps capital spending flat. The downstream side may become more interesting than the upstream equipment side. Now suppose Micron raises fab investment or accelerates equipment purchases.
The equipment suppliers may become the more relevant research set. The same network supports both workflows. The event determines which half of the graph matters.
This helps avoid a common semiconductor research mistake
Investors often treat "semiconductor exposure" as one category. But an equipment supplier, a memory manufacturer, a GPU designer, and a server ODM can react to very different parts of the same cycle. Supply-chain direction helps separate manufacturing-capacity exposure, equipment-spending exposure, component-demand exposure, server and data-center demand, and downstream customer concentration. That is more useful than grouping every company under one semiconductor or AI label.
A repeatable capex-versus-demand workflow
- Start with the company where the event occurred.
- Split the network into upstream suppliers and downstream customers.
- Classify upstream suppliers by manufacturing function.
- Classify downstream customers by end market.
- Use supplier revenue percentage for customer-driven supplier exposure.
- Use customer cost percentage for supplier importance to the customer.
- Keep structural-only edges separate from quantified edges.
- Match the event to the side of the network it can actually affect.
- Research the relevant products before forming a company-specific thesis.
The Micron downstream-network analysis applies the customer side of this framework. The supplier-product validation guide explains how to add product evidence before interpreting an edge.
For metric definitions and limitations, read the Altsets supply-chain data methodology. Browse Supply-Chain Data Use Cases for other investing workflows.