The Chemical Companies Making AI Chips Possible

September 16, 2026

Altsets

Research by Altsets Research

Share

The AI chip boom is increasing demand for recurring semiconductor chemicals and packaging materials, creating a less obvious exposure through suppliers such as Shin-Etsu Chemical and Entegris as TSMC, Micron and Nvidia scale advanced compute.

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

Key findings

  • TSMC said advanced nodes accounted for 77% of second-quarter 2026 wafer revenue while it ramped 2 nanometer production.
  • SEMI reported record global semiconductor materials revenue of $73.2 billion in 2025.
  • Wafer-fabrication materials reached $45.8 billion and packaging materials reached $27.4 billion in 2025.
  • Altsets estimates TSMC represents 4.02% of Shin-Etsu revenue while Shin-Etsu represents 1.33% of TSMC COGS.
  • Altsets estimates Nvidia represents 17.62% of Micron revenue while Micron represents 14.00% of Nvidia COGS.
  • The materials opportunity spans both front-end wafer processing and advanced packaging.

TSMC's latest capacity ramp makes the investment case for semiconductor chemicals unusually clear. In July, the foundry said AI demand remained "extremely robust," 2 nanometer production was ramping steeply, and advanced nodes accounted for 77% of second-quarter wafer revenue. TSMC also said it was building 13 leading-edge and advanced-packaging fabs in Taiwan over the next several years.[1] The obvious beneficiaries are the equipment makers. The less obvious ones are the chemical companies whose photoresists, deposition materials, cleaning chemistries, polishing slurries, specialty gases and packaging resins are consumed every time those fabs process another wafer.

Altsets data shows why this matters before getting to any individual chemistry. Shin-Etsu Chemical supplies TSMC, with TSMC representing 4.02% of Shin-Etsu revenue and Shin-Etsu representing 1.33% of TSMC COGS. Shin-Etsu also supplies Micron, where Micron represents 1.83% of Shin-Etsu revenue and Shin-Etsu represents 1.34% of Micron COGS. Micron, in turn, supplies Nvidia, which represents 17.62% of Micron revenue, while Micron represents 14.00% of Nvidia COGS. These relationships do not prove that a specific Shin-Etsu chemical ends up in Nvidia-bound memory. They do show the economic path investors often miss: a diversified materials supplier can sit two layers upstream from an AI accelerator vendor while participating in both logic and memory expansion.

The materials opportunity is recurring rather than one-time. More advanced wafers and more complex packages create repeated demand for qualified chemicals, gases, slurries, and resins across logic and memory manufacturing.

77%
TSMC wafer revenue from advanced nodes
Second quarter 2026 company disclosure
13
Leading-edge and advanced-packaging fabs planned in Taiwan
TSMC, over the next several years
$73.2B
Global semiconductor materials revenue
Record 2025 level reported by SEMI
$45.8B
Wafer-fabrication materials revenue
2025 global semiconductor materials market
$27.4B
Packaging materials revenue
2025, up 9.3% according to SEMI
$883.2M
Entegris second-quarter 2026 sales
Up from $792.4 million a year earlier

The chemistry scales with process complexity

The semiconductor materials market is already reflecting that shift. SEMI said global semiconductor materials revenue reached a record $73.2 billion in 2025. Wafer-fabrication materials grew to $45.8 billion, while packaging materials rose 9.3% to $27.4 billion. Lithography materials and wet chemicals posted strong double-digit growth as advanced processes required tighter patterning and more material-intensive manufacturing.[2]

Exhibit 1

Semiconductor materials reached a record market size in 2025

Wafer-fabrication materials remain the largest category, while packaging materials also expanded

Category2025 revenueArticle context
Global semiconductor materials$73.2 billionRecord market revenue
Wafer-fabrication materials$45.8 billionFront-end materials used during wafer processing
Packaging materials$27.4 billionUp 9.3% as packaging complexity increased

The figures are market revenue reported by SEMI. They describe material spending categories, not company-specific exposure.

Source: SEMI

That is the key mechanism. A leading-edge chip is not made by buying one lithography system and running silicon through it. Each wafer repeatedly moves through patterning, deposition, etch, cleaning and planarization. Photoresists define the patterns. Ultra-high-purity acids and formulated cleans remove residues and particles without damaging increasingly fragile structures. Chemical vapor deposition and atomic layer deposition precursors build thin films. Specialty gases feed deposition, etch and ion-implant steps. CMP slurry removes material with extremely tight control so the next layer can be built on a flat surface.

Exhibit 2

The front-end chemical stack repeats across many process steps

Selected material categories named in the article

Material categoryManufacturing roleWhy qualification matters
PhotoresistsDefine lithographic patternsPattern fidelity must remain consistent at increasingly tight dimensions
High-purity acids and formulated cleansRemove residues and particlesImpurities can damage structures or reduce yield
CVD and ALD precursorsBuild thin filmsFilm composition and process repeatability are tightly controlled
Specialty gasesFeed deposition, etch, and ion-implant processesPurity and delivery consistency are part of the process window
CMP slurryRemove material and restore planarityRemoval rate and surface quality must be tightly controlled

This is a simplified process map, not a complete wafer-fabrication recipe. Material use and formulations vary by device, node, layer, and manufacturer.

Source: Altsets research using SEMI and company disclosures

Then the process starts to resemble a second chemical supply chain at packaging. AI accelerators increasingly sit beside stacks of high-bandwidth memory and depend on advanced interconnects. Underfills, molding compounds, encapsulants, die-attach materials and thermal materials become part of maintaining mechanical integrity, protecting fine connections and managing heat. AI demand can therefore increase materials intensity in two places at once: at the front end as nodes become more complex, and at the back end as packages integrate more dies and memory.

Exhibit 3

AI increases materials intensity at both ends of semiconductor manufacturing

More complex wafers and more complex packages create separate recurring material demands

  1. 01
    Advanced wafer fabrication
    Patterning, deposition, etch, cleaning, planarization, and repeated process cycles.
  2. 02
    Recurring front-end materials
    Photoresists, high-purity cleans, deposition precursors, specialty gases, and CMP consumables.
  3. 03
    Advanced packaging
    More dies, HBM stacks, finer interconnects, and greater thermal and mechanical requirements.
  4. 04
    Recurring packaging materials
    Underfills, molding compounds, encapsulants, die-attach materials, and thermal materials.

The diagram describes the process structure discussed in the article. It does not imply that every listed material is used in every AI chip or package.

Source: Altsets research

The public-company map is broader than most semiconductor screens suggest. Shin-Etsu Chemical and Tokyo Ohka Kogyo participate in advanced lithography materials. Entegris sells deposition materials, CMP slurries and pads, specialty gases, formulated etch and clean materials, plus filtration and contamination-control systems. Resonac and Shin-Etsu participate in semiconductor packaging materials, while Linde and Air Liquide sit in the electronic-gases layer. Specialist materials companies such as Fujimi add further CMP exposure.

Entegris is particularly useful as a read-through because multiple parts of the chemical stack sit inside one company. Its second-quarter 2026 sales reached $883.2 million, up from $792.4 million a year earlier. The company explicitly ties its Materials Solutions business to CVD and ALD materials, CMP consumables, ion-implantation specialty gases and etch and clean formulations, while its Advanced Purity Solutions business focuses on keeping critical liquid chemistries and gases free of contamination.[3] That is not peripheral infrastructure. At advanced nodes, contamination control is part of the process window.

Exhibit 4

The public-company exposure spans several material categories

Selected companies named in the article

CompanySelected exposure named in the articleRole in the manufacturing stack
Shin-Etsu ChemicalLithography materials and packaging materialsBroad semiconductor materials supplier across multiple manufacturing stages
Tokyo Ohka KogyoAdvanced lithography materialsPhotoresist and related lithography exposure
EntegrisDeposition materials, CMP, specialty gases, etch and clean, filtrationMaterials plus contamination control
ResonacSemiconductor packaging materialsBack-end materials exposure
Linde and Air LiquideElectronic gasesSpecialty gas supply
FujimiCMP materialsSpecialist polishing consumables

The companies occupy different parts of the semiconductor materials stack and should not be treated as interchangeable exposures.

Source: Altsets research using company disclosures

Where the economic dependence sits

The Altsets relationships suggest a different exposure pattern from the concentrated customer stories common elsewhere in the AI supply chain. TSMC is meaningful to Shin-Etsu, but it is not remotely the majority of Shin-Etsu's revenue. Shin-Etsu also supplies Micron, Samsung Electronics and Intel. That matters because the materials supplier can participate across foundry, logic and memory cycles without requiring one customer to dominate its economics.

Exhibit 5

The economic path extends from materials into AI memory demand

Altsets relationship economics for Shin-Etsu, TSMC, Micron, and Nvidia

RelationshipSupplier Revenue %Customer Cost %Interpretation
Shin-Etsu to TSMC4.02%1.33%TSMC represents 4.02% of Shin-Etsu revenue. Shin-Etsu represents 1.33% of TSMC COGS.
Shin-Etsu to Micron1.83%1.34%Micron represents 1.83% of Shin-Etsu revenue. Shin-Etsu represents 1.34% of Micron COGS.
Micron to Nvidia17.62%14.00%Nvidia represents 17.62% of Micron revenue. Micron represents 14.00% of Nvidia COGS.

Supplier Revenue % and Customer Cost % have different denominators. The relationships do not prove that a particular Shin-Etsu material is used in Nvidia-bound products.

Source: Altsets

The downstream relationships can be much more concentrated. Nvidia represents 17.62% of Micron revenue, and it represents 27.88% of SK Hynix revenue in Altsets data. In other words, the memory makers can have large direct exposure to the AI accelerator cycle even while an upstream chemical supplier captures smaller pieces across several manufacturers.

The investment exposure is therefore different. Micron and SK Hynix provide a more concentrated read-through to AI memory demand. A diversified chemical supplier can provide a broader read-through to manufacturing intensity across multiple fabs, nodes and device types. Rather than depending only on which accelerator wins share, the materials supplier can benefit from the recurring process steps required to manufacture advanced logic, HBM and other leading-edge semiconductors.

That distinction also changes how investors should think about bottlenecks. A chemical does not need to represent a large percentage of a chipmaker's COGS to be operationally important. Semiconductor manufacturing depends on qualification, purity and process repeatability. A low-cost input can still be difficult to replace quickly if changing it requires requalification or creates yield risk. Altsets cost percentages measure economic weight, not technical substitutability, so they should not be read as a complete measure of manufacturing dependence.

Conclusion

The AI semiconductor buildout is becoming a materials story as much as an equipment story. TSMC's 2 nanometer ramp, HBM growth and advanced packaging expansion increase the number of process steps in which specialized chemicals are consumed. The less obvious public-company exposure sits with businesses supplying photoresists, high-purity cleans, deposition precursors, CMP consumables, specialty gases and packaging resins across several major chipmakers.

The useful investment distinction is between concentrated downstream exposure and diversified upstream process exposure. Nvidia can be economically important to a memory supplier, while TSMC, Micron, Samsung and Intel collectively create a broader demand base for an upstream materials company such as Shin-Etsu. As AI chips become harder to manufacture and harder to package, the chemical layer becomes more important even when no single materials supplier dominates a customer's cost structure.

Sources

  1. TSMC 2026 Q2 Quarterly Results and earnings conference transcript, Taiwan Semiconductor Manufacturing Company, July 16, 2026, https://investor.tsmc.com/english/quarterly-results/2026/q2

  2. "Global Semiconductor Materials Market Revenue Reaches Record $73.2 Billion in 2025, SEMI Reports," SEMI, May 12, 2026, https://www.semi.org/en/semi-press-release/global-semiconductor-materials-market-revenue-reaches-record-73.2-billion-dollars-in-2025-semi-reports

  3. "Entegris Reports Results for Second Quarter of 2026," Entegris, August 4, 2026, https://investor.entegris.com/news/news-details/2026/Entegris-Reports-Results-for-Second-Quarter-of-2026/default.aspx

How to Cite This

According to Altsets Supply Chain Intelligence (altsets.com), TSMC represents 4.02% of Shin-Etsu Chemical revenue while Shin-Etsu represents 1.33% of TSMC COGS, illustrating how a diversified semiconductor materials supplier can sit upstream of multiple advanced manufacturing cycles.

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.

Methodology

Read the methodology for this research.