Diversifying Within Semiconductors by Customer Ecosystem
September 14, 2026
Altsets
Research by Altsets Research
Two semiconductor holdings can repeat the same major customer while another same-sector company introduces different direct demand paths. Customer-network diversification is therefore a separate layer from ticker count and sector classification.
Data used:Altsets Supply Chain Intelligence: 90k+ entities, 400k+ relationships, 20+ years of history.
Key findings
- The supplied Micron and SK Hynix relationships both repeat Nvidia demand, with Nvidia representing 17.62% and 27.88% of supplier revenue respectively.
- The supplied Shin-Etsu relationships instead point toward TSMC, Samsung, and Intel, adding different immediate customer paths without eliminating the broader semiconductor-cycle exposure shared across the portfolio.
Two semiconductor stocks can belong to the same sector and still add different customer ecosystems, while two different memory stocks can repeat the same downstream dependency. Within-sector diversification therefore improves when the investor compares major customers and bottlenecks instead of assuming every additional semiconductor ticker creates a new economic bet.
Micron and SK Hynix repeat Nvidia demand
The supplied Altsets data shows Nvidia representing 17.62% of Micron revenue and 27.88% of SK Hynix revenue. An investor who owns both companies gains exposure to two issuers, two management teams, and two sets of manufacturing assets, but the portfolio also repeats one major downstream customer.
That does not make the pair irrational. It can be a deliberate way to express a strong view on AI-memory demand. It simply means the investor should describe the pair accurately: the portfolio owns two memory suppliers with a shared Nvidia demand dependency.
If Nvidia demand weakens, the two positions have an economic reason to enter the same event analysis even if their stock returns are not normally perfectly correlated.
Shin-Etsu introduces a different immediate customer set
The supplied Shin-Etsu Chemical data shows TSMC at 4.02% of supplier revenue, Samsung Electronics at 2.43%, and Intel at 1.79%. Those customers place Shin-Etsu inside the semiconductor ecosystem through a different set of immediate demand paths than Micron and SK Hynix.
Adding Shin-Etsu to a Micron position would therefore introduce different direct customers into the portfolio. That is more meaningful from a dependency perspective than simply adding another company whose largest displayed customer is again Nvidia.
The conclusion still needs discipline. TSMC, Samsung, Intel, Nvidia, Micron, and SK Hynix all participate in the semiconductor cycle. Different immediate customers reduce one layer of customer-specific concentration without eliminating broader industry-cycle exposure.
Same-sector diversification should be evaluated in layers
The first layer is security diversification: are the holdings different companies? The second is direct customer diversification: do the companies rely on different major buyers? The third is supplier and bottleneck diversification: do the holdings depend on the same upstream equipment, foundries, materials, or infrastructure? The fourth is cycle diversification: are the underlying demand drivers genuinely independent?
A semiconductor portfolio can improve at one layer while remaining concentrated at another. Micron and SK Hynix are different securities but repeat Nvidia demand. Shin-Etsu adds different immediate customers but remains exposed to semiconductor manufacturing activity.
That layered view is more useful than asking whether the portfolio owns enough tickers.
The percentages help distinguish repeated demand from merely shared themes
Theme labels are broad. "AI," "memory," and "semiconductors" can include companies with very different economic roles. Relationship percentages make the comparison specific.
Nvidia is a much larger displayed supplier-revenue exposure for SK Hynix than any one of the three displayed Shin-Etsu customers. That suggests a different kind of event sensitivity. The investor can then decide whether the goal is to concentrate around Nvidia demand or spread exposure across several manufacturers.
The percentages are not portfolio weights and should not be added across companies. They are relationship-level evidence about where each business gets demand.
The conclusion is that sector diversification can begin inside the customer graph
An investor does not need to leave semiconductors to improve one form of diversification. Choosing companies with different important customers can reduce repeated company-specific demand exposure even while the portfolio remains concentrated in the same broad industry cycle. Micron and SK Hynix repeat Nvidia dependence in the supplied data, while Shin-Etsu's displayed customer set points instead toward TSMC, Samsung, and Intel.
The same-industry customer-diversification guide explains why customer diversity does not eliminate cycle risk. The hidden economic clusters guide explains how companies that look different by classification can still belong to one economic network.
For relationship definitions and evidence limits, read the Altsets methodology.