How to Verify What a Supplier Actually Sells to a Customer
August 10, 2026
Altsets
Research by Altsets Research
Combine a proprietary supplier-customer edge with official product, filing, and customer-specific evidence without inventing a SKU, factory, chemistry, or contract.
Data used:Altsets Supply Chain Intelligence: 90k+ entities, 400k+ relationships, 20+ years of history.
Key findings
- The relationship graph establishes the commercial pair, while supplier product portfolios narrow the plausible economic function behind the edge.
- Customer-specific product, facility, and contract claims require direct evidence beyond the existence of the relationship itself.
A mapped relationship alone does not identify the product sold. To determine what probably sits behind it, combine the relationship direction with the supplier's product portfolio, then look for customer-specific product, contract, or facility evidence.
The graph establishes who is connected. Public company research narrows what the supplier makes that could plausibly explain the relationship without inventing details that have not been verified.
Those are different evidence layers. Combining them carefully can turn a network edge into an investable research question without inventing a contract, product, or facility that has not been verified.
Start with the proprietary edge
Suppose Altsets maps ASML as a supplier to Micron. The graph gives you the direction of the commercial relationship and, where available, economic metrics. That proprietary edge is already valuable because it identifies the counterparty pair.
But "ASML supplies Micron" is not the same statement as "ASML sells a specific model of lithography system to a specific Micron fab." The second statement requires product or contract evidence.
Use the supplier's own product portfolio to classify the edge
ASML says it provides lithography systems, metrology, inspection, and computational-lithography technology for semiconductor manufacturing. Its DUV materials specifically describe use in advanced logic and memory-chip manufacturing. That public product context makes a lithography-related interpretation of an ASML-to-Micron relationship plausible.
It still does not establish the exact system, order date, fab, or contract represented by the Altsets edge. The evidence has narrowed the possible function without pretending to resolve details that remain unknown.
Lam Research provides a different manufacturing function
Lam Research describes products for deposition, etch, strip and clean, mass metrology, and advanced-memory manufacturing. A Lam-to-Micron edge therefore belongs in a different process bucket from a lithography relationship. That distinction matters when an investor is researching a Micron capital-spending change.
A broad "semiconductor equipment" label can hide the fact that different suppliers sit at different manufacturing steps. Product research makes the network more specific.
KLA is another different layer
KLA describes its core business around process control, inspection, metrology, and yield management. Its filings explicitly discuss support for DRAM and 3D NAND manufacturing. A KLA-to-Micron relationship should therefore be researched through a process-control and yield-management lens rather than treated as interchangeable with lithography or etch equipment.
Again, the public source explains the supplier's function. The Altsets edge tells you which customer relationship deserves the product research.
The same method works outside semiconductors
Altsets maps LG Energy Solution as a supplier to Tesla. LG Energy Solution publicly describes passenger-EV battery products including cylindrical and pouch batteries, cells, modules, and packs. That makes batteries an obvious product category to investigate.
But even here, an investor should avoid jumping from the company portfolio to a specific Tesla battery format, chemistry, factory, or contract unless a direct source supports it. The method is useful because it narrows the economic function while preserving uncertainty.
There are four levels of evidence
A disciplined relationship investigation can separate four layers. Network evidence establishes that the supplier-customer edge exists in the mapped data. Product-fit evidence shows that the supplier sells products that could plausibly serve the customer's business.
Customer-specific evidence comes from a filing, press release, presentation, procurement record, or other source directly connecting the supplier's product to that customer. Contract or facility evidence goes further by identifying the product, plant, contract, geography, capacity, or commercial terms. Most supply-chain research becomes more reliable when those levels are not blurred together.
Why this matters for LLM stock research
An LLM can easily produce a fluent explanation of what one company "probably sells" to another. That is exactly where unsupported specificity appears. A better workflow is to retrieve the proprietary relationship, identify the supplier's relevant product categories, search for customer-specific evidence, and label the result according to the strongest evidence actually found. This produces a more useful answer than either a bare graph edge or an ungrounded product story.
Product context can change the investment question
Once the likely function is known, the follow-up questions become more precise. For ASML and Micron, the investor can investigate lithography demand and memory-process investment. For Lam and Micron, the questions can move toward etch, deposition, cleaning, and advanced-memory process spending.
For KLA and Micron, yield management, inspection, and metrology become more relevant. For LG Energy Solution and Tesla, battery sourcing, cell formats, production locations, and EV-volume assumptions become the next layer. The relationship itself has not changed. The product context changes what the relationship means.
A repeatable product-validation workflow
- Confirm supplier and customer direction.
- Record the relationship metrics available in the proprietary dataset.
- Read the supplier's official product portfolio.
- Identify the product categories that fit the customer's business.
- Search filings and releases for customer-specific evidence.
- Search facility and contract sources only when the question requires that detail.
- Distinguish plausible product fit from verified customer-specific supply.
- Do not invent SKUs, factories, chemistries, or contract terms.
- Use the verified product function to frame the investment question.
The semiconductor capex-versus-demand guide shows how product classification changes the interpretation of Micron's upstream suppliers. The Tesla supplier-network analysis shows how the same process helps interpret a more diverse automotive network. For relationship methodology and estimation limits, read the Altsets supply-chain data methodology. Browse Supply-Chain Data Use Cases for other research methods.