Article

Semiconductor and materials companies: get paid for your records

By TrustPast

Updated November 7, 2026

This page is for companies that make chips, wafers, substrates and the materials that go into them: fabless design houses, foundries and specialty fabs, packaging and test houses, equipment makers’ applications teams, and producers of specialty chemicals, gases, films and advanced materials. A company like that with a few dozen engineers and a decade of runs behind it has a record that few other industries can match for discipline.

What most of them don’t realise is that the investigation is the asset, not the result. The yield number on a dashboard is only the result. The thing behind it, a failure analysis that went through four hypotheses before landing on a contaminated chemical lot, is reasoning written down while the work happened. A buyer pays to license records like these to train AI models. It isn’t buying your process. It wants the way your engineers worked through it.

We made the same argument for roofers and plumbers on what data buyers actually want from a trade business: the trail of how a job went from quote to invoice is what has value. In a fab, the trail runs from a design review to a shipped lot. Here is what that looks like, what stays out, and how it works with micro1.

The records a semiconductor or materials company keeps

These are our examples, not a buyer’s list. They are what a company of this kind produces as it runs, lot after lot and year after year.

The lot traveler. Every lot carries a record of where it went: each step, each tool, each hold, each rework, with the operator’s notes. When a lot was put on hold, the traveler says why and what released it. It is the closest thing a fab has to a work order, and a busy line generates thousands.

The non-conformance report. A measurement came out of spec. Someone wrote it up, decided whether to scrap, rework or use as is, and signed it. The disposition and the reasoning behind it are the record.

The failure analysis. A customer return or a yield drop goes to the lab. The request says what failed. The report says what was tried: cross-sections, electrical tests, a hypothesis about a tool, a second hypothesis about a supplier, the evidence that settled it. Many companies keep these for as long as they have run.

The corrective action. Once a cause is found, someone writes the fix: containment, root cause, permanent action, verification. The structured problem-solving reports a quality team files are among the most orderly records any company keeps.

The engineering change notice. No step in a qualified process changes without a request, a risk review, a qualification plan and an approval. The change board’s minutes say who worried about what. A process that has drifted through a hundred approved changes carries a hundred written reasons.

The experiment plan and its results. A new film, a new etch chemistry, a new anneal. The plan names the variables. The results say which mattered. The engineer’s conclusion, and the manager’s question about it, sit underneath.

The tool maintenance log. Preventive maintenance work orders, unscheduled downtime, the part that kept failing and the vendor thread about it. Each repair note says what the technician found.

The design review record. For a design house, the specification, the review checklists, the issues raised at each gate, the tape-out sign-off and the post-silicon debug notes when first silicon didn’t behave.

The batch record. For a materials producer, the formulation sheet, the batch log, the in-process checks, the out-of-spec investigation and the certificate the customer received. The investigation is where the knowledge sits.

Two record types carry most of this. The change notices, design reviews, experiment plans and corrective actions are project history: the plan, the changes and the review of what happened. The travelers, maintenance tickets and non-conformance reports are job files and work orders: the individual jobs, each with notes the status field never shows.

Together they show how a yield problem was diagnosed, how a process change was argued for and approved, how a new material was qualified. That is the how, not the who. The engineers’ names come out. Their reasoning stays.

See your estimate

What counts, in micro1’s own words

micro1’s data partnerships page says: “We are looking for data that reflects how work is actually performed within an organization.” It lists “standard operating procedures (SOPs), process documentation, internal knowledge bases, workflows, templates, operational frameworks, and other business process artifacts.” A fab runs on SOPs; few industries document a procedure more tightly.

The page goes on: “The most valuable datasets capture real operational expertise, decision-making processes, and repeatable workflows.” A change board is a decision-making process with minutes. A failure analysis is operational expertise in its most concentrated form.

micro1 lists industries whose operational data “can contribute”. It doesn’t name semiconductors or materials. Two of its headings fit. Operations & logistics: “Operational SOPs, logistics processes, fulfillment workflows, inventory management, and internal operations.” And Software & engineering: “Source code documentation, engineering workflows, bug tracking, product planning, and technical knowledge bases.” A fab’s operations and an engineering team’s workflows both sit under those words. We apply the headings; micro1 decides what it licenses.

micro1’s October announcement says the investment “will expand the range of industries and workflows represented” in its work. It doesn’t say which industries. It does say that “Real business operations involve incomplete information, competing priorities and exceptions that require judgment.” A lot on hold at the end of a quarter, with a customer waiting and an engineer unsure, is exactly that.

On eligibility, micro1 says it is “looking for established companies with 30+ full-time employees, mature business operations, and substantial historical operational data.” On volume: “There’s no fixed minimum number of files, documents, or terabytes.” It says it evaluates “the historical depth, breadth, quality, and relevance of the available data.” And it says “Data should be primarily in English,” which matters for a company whose sites span several countries.

What is never included

Personal data stays out. Employees’ names, badge records, training files, medical surveillance records from a chemical plant and anyone’s contact details are not what a buyer wants. Our page on what is never included shows where those details turn up: in signatures, in shift logs, in the approval line of a change notice.

Customer material stays out too. A foundry or a packaging house works on designs it doesn’t own. Mask data, a customer’s design files, its product specifications and anything the contract says belongs to the customer stay with the customer. Our page on client-owned work and NDAs covers how to read those terms. In this industry, they are usually strict.

Export controls come next. The Bureau of Industry and Security publishes the Export Administration Regulations, “15 CFR Parts 730-774”. Its deemed-exports page says: “A deemed export refers to the sharing or release of controlled technology or source code to a foreign person within the United States.” Some process technology and design data in this industry is controlled. Whether any of your records are, and what a licence would mean for them, is a question for counsel and your export compliance lead before anything is scoped. Ask counsel.

Then your own crown jewels. A recipe, a formulation or a proprietary process window may be the company’s most valuable secret. Nothing micro1 publishes asks for them; its page asks for how work is performed. Whether any of it goes into a scope is entirely your decision. As our own advice, a company can license its investigations and its change history without licensing the numbers that make its process work. Our page on health, payment and other regulated data covers other categories that need care of their own, including classified and export-controlled material.

How de-identification works

You don’t have to clean the records yourself. micro1 says: “Does my company need to anonymize its data before transferring it? No, micro1 has processes for de-identifying and transforming sensitive information.”

Before any participation, micro1 says the two sides confirm four things. De-identified before sharing, in micro1’s words: “Personal and identifying information is removed or transformed before your data is shared with third parties.” Data segregation: “Your data is kept separate from other companies’ data, with access restricted to authorized personnel.” Defined retention periods: “Raw data is deleted according to an agreed schedule.” And data is licensed “exclusively to labs and enterprises based in the U.S. and allied countries.” For a company that works under export rules, that last line is worth reading with counsel; it is a description, not a legal clearance.

micro1 says what happens after transfer: “Depending on the nature of the data, personally identifiable information may be removed and certain datasets may undergo synthetic rewrites that preserve workflow structure while reducing direct ties to original records.” For a failure analysis, the workflow structure is the hypothesis, the test, the result and the next hypothesis. The tool names, part numbers and customer references are what a rewrite can change.

micro1 also says “companies have the opportunity to review representative samples before use,” and “Upon request or at the end of an engagement, data is deleted according to the agreed retention and deletion terms.” A quality manager reading a sample will see quickly whether a rewritten report still gives away a process detail you didn’t mean to share. Our page on how de-identification works covers what to look for.

What decides the price

micro1 publishes three tiers. “Qualified Enterprise Data Partnerships” are $100K-500K (indicative, not an offer). “Large-scale operational datasets or ongoing participation across multiple teams and business functions” are $500K-3M (indicative, not an offer). “Highly unique, proprietary operational data” is $3M+ (indicative, not an offer). micro1 says payments are “typically ranging from $100K to over $3M” (indicative, not an offer).

micro1 says compensation “is based on the quality, uniqueness, and value of the data provided,” and lists “dataset size and volume, workflow complexity, data quality, domain expertise represented, and the uniqueness of the operational knowledge contained within the dataset.” It adds that “Each partnership is evaluated individually.”

Read against a fab or a materials plant, several of those factors run high. Domain expertise: a failure analyst’s report takes years of training to write. Workflow complexity: a change notice crosses process, equipment, quality, the customer and sometimes a supplier. Quality: a company that runs under a formal quality system has records that are structured, signed and complete by habit.

micro1’s payout page asks for employees, years of data and headquarters country, and says the result is an “Indicative estimate only. Final payout is determined based on data sources, volume, quality, access terms, and due diligence.” In our reading, what lowers the figure: a scope cut down heavily by customer contracts and export rules, or records locked in a system nobody can export from. What raises it: a long, unbroken run of change notices, investigations and corrective actions, each with its reasoning. Our page on what decides the price goes through each factor.

On timing, micro1 says it has “30 days to review and approve” data after transfer, and that “payment will be made within 30 days of invoicing.”

Which programme fits

micro1’s programme is for companies with 30 or more people, counted now or at peak, and for a company of any size outside the US. micro1 says it is “currently prioritizing US companies, followed by other Western markets.” Our estimate asks how many people work at the company, counting everyone on every shift, and names the programme that fits that headcount. The line is set out on the programme page for 30 or more people.

This industry closes sites and companies too. micro1’s company referral page lists a “Defunct Company Data Partnership” and says: “We are acquiring data from companies that are no longer operating.” Its requirements are “30+ employees”, “10+ Countries”, “English” and “All industries”, with company compensation of “$100K-$3M+ for approved data packages” (indicative, not an offer). A closed fab’s quality records, if they were archived when the lights went off, are what that route describes. Who may sign for a company that no longer exists, and whether export rules still apply to its records, are questions for counsel.

Semiconductor and materials companies share the clinical habit of documenting every case with medical practices and health systems, where a symptom leads to a diagnosis and a plan the way a yield drop leads to a root cause. They share the committee-and-approval trail with investment firms, where nothing moves without a memo and a vote. And they share strict change control with payments and fintech companies, where a ledger change goes through review the way a recipe change does.

TrustPast

Questions and answers

Does micro1 want our process recipes or design files?

micro1’s pages don’t ask for them. They say micro1 is “looking for data that reflects how work is actually performed within an organization”: procedures, workflows and decision-making. Which recipes or designs, if any, go into a scope is the company’s decision, and for anything export-controlled, ask counsel first.

Are export controls a concern?

They can be. The Bureau of Industry and Security says “A deemed export refers to the sharing or release of controlled technology or source code to a foreign person within the United States.” Ask counsel and your export compliance lead which records are controlled before any scope is drawn.

What about designs our customers own?

Anything a customer owns under contract stays theirs and stays out, as our own advice. A foundry’s or contract manufacturer’s own procedures, investigations and change history are what remain.

How many years of records count?

micro1 says “There’s no fixed minimum number of files, documents, or terabytes,” and that “Companies with several years of records across multiple business systems are generally a stronger fit.”

What does micro1 pay?

micro1 lists three tiers: $100K-500K, $500K-3M and $3M+ (indicative, not an offer). It says compensation is based on “the quality, uniqueness, and value of the data provided.”

Do we keep ownership?

micro1 says: “Companies retain ownership of their underlying data.”