The Companies Three Layers Beneath TSMC

September 16, 2026

Altsets

Research by Altsets Research

Share

TSMC's 2026 capex surge reaches beyond the obvious equipment names into Japanese wafer, photoresist, CMP, vacuum, packaging and tool-component suppliers where customer dependence can be more concentrated.

Data used:Altsets Supply Chain Intelligence: 90k+ entities, 400k+ relationships, 20+ years of history.

Key findings

  • TSMC raised its 2026 capital budget to $60 billion to $64 billion, with 70% to 80% directed to advanced process technology and another 10% to 20% to advanced packaging, testing, mask making and related areas.
  • JETRO counted 13 Japanese companies among the 30 TSMC suppliers recognized for 2025, highlighting a dense Japanese layer across wafers, materials and semiconductor equipment.
  • In the displayed Altsets relationships, TSMC represents 4.02% of Shin-Etsu revenue, compared with 2.43% for Samsung Electronics and 1.79% for Intel.
  • In the same displayed data, Shin-Etsu represents 1.33% of TSMC COGS, 0.29% of Samsung COGS and 0.94% of Intel COGS.
  • Ferrotec illustrates the supplier-of-supplier layer through semiconductor-equipment components including quartz, fine ceramics, CVD-SiC parts, ferrofluidic seals and chillers.

TSMC's latest spending increase is not just an ASML or Tokyo Electron story. In July, TSMC raised its 2026 capital budget to $60 billion to $64 billion, up from its previous $52 billion to $56 billion range, while saying that 70% to 80% of spending would go to advanced process technology and another 10% to 20% to advanced packaging, testing, mask making and related areas. Management also pointed to a steep 2nm ramp and continued structural AI demand. [1]

The more interesting investment question is what sits below the obvious tool vendors. TSMC's own supplier ecosystem shows a dense Japanese layer of silicon wafers, photoresists, deposition equipment, vacuum systems, CMP tools, dicing systems and packaging materials. JETRO counted 13 Japanese companies among TSMC's 30 suppliers recognized for 2025, more than from any other country. [2] That makes Japan one of the clearest places to look for second and third order exposure to TSMC's capacity build.

The deeper investment signal is not simply who supplies TSMC. It is where TSMC matters disproportionately to a specialist supplier and where qualified capacity is difficult to replace.

$60B to $64B
TSMC 2026 capital budget
Raised in July 2026. Source [1].
70% to 80%
Share of 2026 capex directed to advanced process technology
TSMC management guidance. Source [1].
10% to 20%
Share directed to advanced packaging, testing, mask making and related areas
TSMC management guidance. Source [1].
13 of 30
Recognized 2025 TSMC suppliers that were Japanese companies
JETRO count. Source [2].
4.02%
TSMC share of Shin-Etsu revenue in the supplied Altsets relationship data
Supplier Revenue % preserves the supplier-side direction.
1.33%
Shin-Etsu share of TSMC COGS in the supplied Altsets relationship data
Customer Cost % preserves the customer-side direction.

The economic exposure gets larger as you move upstream

Start with silicon. TSMC's 2025 supplier group included Shin-Etsu Handotai and SUMCO for silicon wafers, alongside JSR for photoresist, Asahi Kasei for photosensitive insulating materials and Resonac for packaging materials. The equipment side included Ebara for CMP, Kokusai Electric for deposition, ULVAC for deposition and vacuum systems, and Disco for dicing. [2]

Exhibit 1

The Japanese layer beneath TSMC spans multiple process steps

Selected companies named in the article from TSMC's recognized 2025 supplier group

CompanyRole discussed in the article
Shin-Etsu HandotaiSilicon wafers
SUMCOSilicon wafers
JSRPhotoresist
Asahi KaseiPhotosensitive insulating materials
ResonacPackaging materials
EbaraCMP equipment
Kokusai ElectricDeposition equipment
ULVACDeposition and vacuum systems
DiscoDicing equipment

This is a selected subset of the suppliers discussed in the source, not a complete TSMC supply chain or a ranking of supplier importance.

Source: Altsets presentation of JETRO supplier information cited in Source [2]

Altsets adds an important distinction: a relationship can be operationally important to TSMC without being equally important to both companies economically. Altsets maps Shin-Etsu supplying TSMC, with TSMC representing 4.02% of Shin-Etsu revenue while Shin-Etsu represents 1.33% of TSMC COGS. That is not proof that TSMC cannot substitute a specific Shin-Etsu product, and it does not describe TSMC's complete wafer or materials supply chain. It does show that the customer is more material to the supplier than the supplier is to the customer's total cost base.

The comparison gets more useful across customers. In the same Altsets data, Samsung Electronics represents 2.43% of Shin-Etsu revenue and Intel represents 1.79%, versus 4.02% for TSMC. On the customer side, Shin-Etsu represents 0.29% of Samsung COGS and 0.94% of Intel COGS, versus 1.33% for TSMC. The point is not that Shin-Etsu is a single-source bottleneck. It is that TSMC's growth can be economically meaningful to an upstream materials supplier even when the same supplier occupies only a small fraction of TSMC's own cost structure.

Exhibit 2

TSMC is the largest of the three displayed Shin-Etsu customer exposures

Supplier Revenue % across the three Altsets relationships discussed in the article

TSMC
Supplier Revenue %
4.02%
Samsung Electronics
Supplier Revenue %
2.43%
Intel
Supplier Revenue %
1.79%
CustomerSupplier Revenue %Customer Cost %
TSMC4.02%1.33%
Samsung Electronics2.43%0.29%
Intel1.79%0.94%

Bar lengths are normalized to TSMC, the largest displayed Supplier Revenue % in this three-customer subset. The labels preserve the actual Altsets percentages. This subset is not Shin-Etsu's complete customer base.

Source: Altsets

That asymmetry is exactly what investors miss when they stop at TSMC itself.

Three layers down, the bottlenecks become less visible

The first layer is familiar: TSMC buys leading-edge tools and materials. The second layer is the specialist Japanese companies that keep those processes running, including wafer makers, resist suppliers, CMP systems, vacuum equipment, deposition tools and packaging materials. The third layer sits inside those suppliers' own production systems.

Exhibit 3

A simple map of the three-layer dependency

The article's structural argument, shown without implying a complete supply chain

  1. 01
    TSMC capacity build
    Advanced process, packaging, testing and mask-related investment
  2. 02
    Process tools and materials
    Wafers, resist, deposition, CMP, vacuum, dicing and packaging inputs
  3. 03
    Specialist Japanese suppliers
    Companies such as Shin-Etsu, SUMCO, JSR, Ebara, ULVAC and Disco
  4. 04
    Supplier production systems
    Quartz, ceramics, seals, chillers, gases, filtration, purity control and other enabling inputs

The diagram describes functional layers discussed in the article. It does not assert undisclosed supplier-of-supplier contracts or a complete path for any one TSMC process.

Source: Altsets synthesis of the cited supplier disclosures

Ferrotec is a clean example. The company received an Excellent Partner Award from Tokyo Electron and sells semiconductor-equipment components including quartz, fine ceramics, CVD-SiC parts, ferrofluidic seals and chillers. [3] Those are not products that appear in headlines about TSMC's 2nm ramp, but they are embedded in the equipment stack required to add chambers and keep them operating.

The same logic applies across process gases, chemical filtration and ultrapure-water systems. As nodes shrink, a larger share of manufacturing performance depends on contamination control, purity, flow stability and materials consistency. Public disclosures rarely reveal every supplier-of-supplier contract, so investors should not invent links that companies have not disclosed. But the functional dependency is clear: more advanced capacity means not only more lithography systems, but also more wafers, resist, specialty chemicals, gas delivery, vacuum hardware, chamber components, cleaning steps and packaging inputs.

The revenue timing also differs by layer. A lithography or deposition tool is largely a capital-equipment sale. Wafers, photoresist, gases, CMP materials and filtration are consumed repeatedly as utilization rises. Tool components create another cycle through initial installations, replacements and maintenance. Packaging materials scale with advanced packaging throughput. That means TSMC's capex surge can reach some upstream suppliers when fabs are built, and others later when those fabs fill with wafers.

For investors, that is the useful distinction. The broad AI trade concentrates attention on the largest beneficiaries, but some of the sharper operating leverage may sit in smaller suppliers with narrower customer bases and more specialized products. The Shin-Etsu relationship illustrates the economic side of that pattern. Ferrotec illustrates the structural side. Neither is a direct forecast of earnings or stock performance, but both show why the TSMC buildout should be mapped beyond tier-one equipment.

What this analysis does and does not establish

The displayed Altsets relationships are selected economic relationships, not a complete semiconductor supply-chain map. The Shin-Etsu percentages do not prove single sourcing, technical irreplaceability, a specific wafer grade, a specific fab allocation, or a causal effect on earnings or stock prices.

The Ferrotec example is structural rather than a disclosed direct TSMC relationship. Its award from Tokyo Electron and its semiconductor-equipment component portfolio show how supplier-of-supplier exposure can sit beneath a major tool vendor, but the article does not infer an undisclosed Ferrotec-to-TSMC contract.

The same caution applies to process gases, filtration and ultrapure-water systems. Their importance to advanced semiconductor manufacturing is part of the functional production stack described here, but public disclosures do not expose every commercial edge. Missing relationship data therefore stays missing.

For metric definitions and interpretation limits, see the Altsets research methodology.

Conclusion

TSMC's $60 billion to $64 billion 2026 capital plan is creating a supply-chain expansion that reaches well beyond the companies most investors associate with semiconductor capex. The deeper Japanese layer includes wafers, photoresists, CMP, vacuum systems, deposition tools, packaging materials and equipment components. The investment signal is not simply whether a company "supplies TSMC." It is where TSMC matters disproportionately to the supplier, where qualification makes replacement difficult, and where recurring consumption rises as new capacity moves from construction into production.

The further down the chain you go, the smaller the company can become, while the exposure to TSMC's expansion can become more concentrated.

Sources

  1. TSMC Q2 2026 Earnings Conference Transcript, Taiwan Semiconductor Manufacturing Company, July 16, 2026, https://investor.tsmc.com/english/encrypt/files/encrypt_file/reports/2026-07/57b65edbfe6e480e74abe202be983ecbde79e934/TSMC%202Q26%20Transcript.pdf

  2. TSMC Announces 2025 Outstanding Suppliers, Japan External Trade Organization, December 2, 2025, https://www.jetro.go.jp/biznews/2025/12/87a5f8a456613817.html

  3. Ferrotec Product and Company Information, Ferrotec Corporation. https://product.ferrotec.co.jp/en/

How to Cite This

According to Altsets Supply Chain Intelligence (altsets.com), TSMC represents 4.02% of Shin-Etsu revenue in the displayed relationship data, while Shin-Etsu represents 1.33% of TSMC COGS.

For research inquiries or data access: press@altsets.com

Go Deeper

Altsets Research

Get new Altsets research when we publish.

Original supply-chain research, market casefiles, and new data findings.

Research updates only. Unsubscribe any time.

Sources

Methodology

Read the methodology for this research.