TSMC Alone Nearly Matches Samsung and Intel Combined in Shin-Etsu Exposure
September 14, 2026
Altsets
Research by Altsets Research
TSMC represents 4.02% of Shin-Etsu revenue in the supplied data, almost equal to Samsung and Intel combined at 4.22%. The three displayed semiconductor customers sum to 8.24% of Shin-Etsu revenue.
Data used:Altsets Supply Chain Intelligence: 90k+ entities, 400k+ relationships, 20+ years of history.
Key findings
- TSMC is the largest single displayed customer at 4.02% of Shin-Etsu revenue, but Samsung and Intel together total 4.22%, slightly more than TSMC alone.
- The three relationships sum to 8.24% of Shin-Etsu revenue in the supplied data, showing meaningful customer diversification across manufacturers without proving diversification away from the semiconductor cycle.
Almost. In the supplied Altsets data, TSMC represents 4.02% of Shin-Etsu Chemical revenue, while Samsung Electronics and Intel together represent 4.22%. TSMC is therefore nearly as large a Shin-Etsu customer exposure as the other two displayed semiconductor customers combined. Across all three relationships, the displayed revenue exposure sums to 8.24% of Shin-Etsu revenue.
TSMC is the largest single customer in the trio without dominating the trio
Samsung represents 2.43% of Shin-Etsu revenue and Intel 1.79%. Added together, they reach 4.22%, only 0.20 percentage points above TSMC's 4.02%. TSMC alone accounts for about 48.8% of the 8.24 percentage points represented by these three displayed customer relationships.
That creates a more precise conclusion than saying Shin-Etsu is diversified across TSMC, Samsung, and Intel. The displayed set is not evenly split, because TSMC is clearly the largest individual relationship. It is also not a one-customer story, because Samsung and Intel together slightly exceed TSMC's contribution to the three-customer total.
This is what useful customer diversification can look like without becoming cycle diversification
Shin-Etsu's annual report describes a semiconductor-materials portfolio that includes semiconductor silicon, photoresists, photomask blanks, encapsulating materials, and synthetic quartz products. Serving several major chip manufacturers can reduce dependence on the fortunes of one customer and give the supplier several ways to participate in semiconductor production.
The customer mix does not eliminate industry-cycle risk. TSMC, Samsung, and Intel can all be affected by semiconductor demand, capital spending, technology transitions, and geopolitical constraints. The Altsets conclusion is therefore specific: Shin-Etsu's displayed customer exposure is spread across several major manufacturers, with TSMC almost equal to Samsung and Intel combined. That is customer diversification, not proof of independent economic cycles.
The comparison also changes how market-share shifts should be interpreted
If market share moves among TSMC, Samsung, and Intel, Shin-Etsu does not automatically lose the same amount of demand as the losing manufacturer. The supplied graph shows existing relationships to all three companies. Some downstream share movement can remain inside Shin-Etsu's customer network if the winning company also uses relevant Shin-Etsu products.
The relationship percentages do not prove that demand transfers one-for-one. Product mix, customer-specific sourcing, pricing, technology, and supplier share can differ across the three relationships. What the data establishes is the possibility of partial insulation: the supplier has meaningful commercial paths to more than one competitor, and the largest single displayed path does not exceed the other two combined.
The displayed trio is more balanced than a one-customer story
The 4.02% TSMC exposure is large enough to deserve individual attention, but it does not overwhelm the other two displayed customers. Samsung and Intel together slightly exceed TSMC, which means the visible customer structure is better described as one large relationship plus two meaningful secondary relationships than as a single-customer dependency.
That distinction matters when one manufacturer has a weak quarter. A disappointing TSMC result can still matter materially to Shin-Etsu, but the supplied data gives the investor a reason to ask whether Samsung or Intel demand is moving in the opposite direction before turning one customer event into a company-wide conclusion. The three relationships create several demand paths, and their relative weights tell the investor which path deserves the most attention without pretending the paths are economically independent.
Historical changes in the mix could matter more than the current total
The current 8.24% combined figure is a snapshot of the three displayed customer relationships. A more powerful historical question is whether that total is becoming more concentrated in TSMC, spreading more evenly across the three manufacturers, or changing because one customer is losing importance while another gains it.
Those changes can alter the quality of the supplier's growth even if total semiconductor revenue remains strong. A rising TSMC share could increase dependence on one customer while preserving the same broad industry exposure. A more balanced split could reduce company-specific customer risk without reducing semiconductor-cycle risk. Point-in-time relationship history is therefore useful not only for measuring today's concentration but for determining whether the customer structure itself is becoming more or less resilient.
The conclusion is neither "fully diversified" nor "dependent on TSMC"
The stronger conclusion is more interesting. TSMC is Shin-Etsu's largest displayed semiconductor customer exposure at 4.02%, but Samsung and Intel together are slightly larger at 4.22%, and all three sum to 8.24% of Shin-Etsu revenue in the supplied data. The company has meaningful exposure to several major manufacturers without escaping the semiconductor cycle they share.
The picks-and-shovels guide explains why supplying several competitors can reduce winner-selection risk. The same-industry diversification guide explains why several customer names do not automatically create several independent economic cycles.
For relationship definitions and evidence limits, read the Altsets methodology.