Semiconductor factories in Europe: how to distinguish announcements from operating capacity

When people talk about a semiconductor factory, they often picture a building where silicon wafers go in and finished chips come out. In reality, manufacturing consists of a chain of industrial and engineering operations. At a wafer fabrication plant, or fab, microscopic structures are created through successive stages of deposition, exposure, etching, cleaning and inspection. Wafers are then usually cut, and the individual chips are packaged and tested; those operations may take place at separate facilities. The word “factory” therefore does not, on its own, identify which parts of the chain a project covers.

The distinction between making wafers and carrying out the later stages also affects how an investment should be described. A project may focus on one part of the chain and depend on other facilities to complete the product. A claim about a “chip factory” should therefore be read alongside a description of its tasks: the general label does not show where each operation takes place or which operations fall outside the project.

The type of product intended for manufacture matters, too. A plant specialising in chips for automotive or industrial use should not automatically be described as a facility for advanced processors. Processes, equipment, materials, customers and quality criteria can vary considerably between products. Even two plants using wafers of the same diameter do not necessarily use the same technology or produce interchangeable components. Manufacturing capacity is informative only when accompanied by a description of the process, the type of chip and the scope of the operations. This avoids letting an isolated figure suggest that all facilities supply the same kinds of components or can meet the same industrial needs; this article does not claim that plants making different products are equivalent.

Industrial announcements often mix together different statuses: a company may announce an intention to invest; a government may approve aid; the company may secure financing, begin construction and, much later, install equipment and start production. Each event proves something different. Approval of a grant establishes a regulatory decision about public support, but does not, by itself, show that the full amount has been disbursed, that the plant is complete or that it is making chips. Keeping these stages separate makes it possible to report progress without turning a forecast or administrative decision into evidence of factory activity.

A practical scale for describing the situation helps avoid jumping to conclusions:

• Announced: there is a public statement about the project. This does not establish that financing conditions or permits have been fulfilled.
• Approved: a competent authority has approved a specific aid measure or regulatory aspect. The approval is limited to what the document says.
• Financed: there is evidence that financing has been committed or made available; this should not be inferred solely from a planned investment figure.
• Under construction: documentary confirmation shows that work has begun. This is not the same as having equipment installed or being in production.
• Being commissioned: facilities and processes are being tested and integrated; stable commercial production does not necessarily exist.
• In production: the company or an authority confirms that manufacturing is actually taking place. To describe capacity as sustained, further dated information on volume and operations is needed.

These labels are an editorial tool, not a universal certification. Milestones may overlap, stop or change; construction can continue while financing or schedules change. It is advisable to say who reported each milestone and when, and not to turn a corporate plan into an accomplished fact. If a source says that a plant “will be operational” in a given year, the accurate wording is that this is the announced target, not that production is guaranteed by that date. A label should reflect the document available at the time of writing, not an assumption based on the preceding stage: an announced investment does not justify calling a project financed, just as visible construction does not establish that production has begun.

Verification starts by defining the question. To find out whether aid was approved, the most direct source is the decision or communication from the authority that approved it. Corporate announcements and company registers can help establish which company will carry out a project, although a press release is not a substitute for a financial document. To confirm construction, planning files, permits, contracts or local official reports may be useful. To substantiate a claim that production exists, a specific, dated company statement or a public source describing the start of manufacturing is needed; an opening ceremony alone does not prove it.

The choice of source therefore depends on the claim being made. An approval document is relevant to describing a regulatory decision, but it does not necessarily establish the physical state of the facilities. By contrast, claims about construction or production require evidence that addresses those activities, as well as a check of the period covered by the evidence. This distinction helps prevent a reliable source for one fact from being used to support a different fact that it does not address.

The European Commission provides information about state aid measures and projects it examines under European Union rules. In August 2024, it announced that it was approving €5 billion in German aid to support the construction and operation of an ESMC fabrication facility in Dresden. The same information identifies ESMC as a joint venture between TSMC, Bosch, Infineon and NXP, and links the project to chips for automotive and industrial applications. This establishes approval of the measure described, not that the factory was already producing when the information was published. Attribution matters: it allows reporting what the Commission decided and described without extending the conclusion to an operational status that the communication does not, by itself, confirm.

An investment figure does not automatically describe how much has been spent. It may represent the project’s projected total cost, investment over several years, a construction budget, or a combination of public and private contributions. The maximum amount of approved aid is not necessarily money already transferred, either. Accurate reporting preserves the nature of the figure and attributes it to its source: “the Commission approved aid of up to…” is different from “the company has already received…”. If the source does not explain the period or concept represented by the amount, it is best not to fill in that information by inference.

Announced capacity raises a similar issue. It may refer to the plant’s design capacity once complete, a future project phase, or a technical metric that does not correspond to sellable finished units. Comparing two projects requires, at a minimum, matching units, periods, industrial scope and reference dates. If one company reports wafers per month and another chips per year, adding or comparing the figures without knowing chip size and process yield would be misleading. A figure stated as nominal capacity does not necessarily reveal how much is produced in a particular period; that requires a dated operational statement with clearly specified units.

A fact-checking record can include the following fields without turning them into a ranking: the entity making the claim, date, unit, project stage, funding source, intended technology or application, and document supporting the milestone. If a field is missing, it should be marked as unspecified. The absence of public information does not prove that a project does not exist; it limits what can be said about it. In particular, future targets and nominal capacity should not be added together as though they were production already available in Europe. Recording both what is known and what is unspecified makes comparisons more transparent and prevents the apparent precision of a figure from obscuring differences in scope.

The ESMC project in Dresden shows why regulatory support and industrial progress must be kept separate. On 20 August 2024, the European Commission announced that it had approved a €5 billion German state-aid measure to support the fabrication facility. It also described ESMC as a joint venture between TSMC, Bosch, Infineon and NXP, and identified automotive and industrial applications as intended uses. This source supports the claims that the measure was approved and that the stated purpose was as described; it does not establish that the full amount was disbursed or that the plant was then producing. The date of the communication also defines the scope of the claim: it is evidence of what had been approved and described at that time, not an automatic update on later stages.

Another example is Onsemi’s project in the Czech Republic. In 2025, the Commission announced the approval of €450 million in Czech state aid linked to the company’s project. This documents a European state-aid control decision and approved public support; it does not, by itself, certify completed construction or commercial manufacturing. The Commission’s page is an official source for its decision. To make a claim about the project’s current physical status, later documents from the company and Czech authorities would need to be checked, with clear dates. The same caution that applies to ESMC applies here: the figure is specific, but so is its scope.

The two cases illustrate a rule for reading announcements, not an assessment of success: a state-aid decision is a verifiable milestone, but it does not summarise the entire project lifecycle. Nor does it allow conclusions about how much European production has already increased, which chips are available to buyers or whether schedule targets have been met. Those conclusions require operational and production data that do not follow from the approval. This report therefore does not classify these initiatives as operating factories on the basis of financing announcements alone. Presenting the examples this way does not diminish the significance of documented decisions; it defines precisely what they support and which questions require additional sources.

The European Union adopted the Chips Act in 2023 as part of a framework to strengthen its semiconductor ecosystem. The regulation and associated policy describe objectives and institutional tools; they are not an inventory of plants already in production. Industrial policy can enable projects and support research, capacity or resilience, but assessing results requires examining specific facilities and distinguishing their different stages. The policy framework and evidence of production answer different questions: the former helps explain objectives and tools, while the latter requires information about actual manufacturing activity.

A rigorous account of manufacturing in Europe must therefore separate three questions: which projects have been announced or approved, which have reached the construction stage, and which have demonstrated production. It should also specify what is manufactured, at which facility and in what units capacity is reported. Mixing projects at different stages can create an impression of immediate expansion that the available documents do not support. Recognising this limitation does not mean concluding that projects will fail; it means that their status must be updated using specific evidence. A useful account need not erase uncertainty: it can record different stages and identify what information is missing to move from one claim to another.

For anyone reading the next factory announcement, a few initial checks are enough: find the original document; note its date and who is speaking; distinguish approved aid from investment actually made; look for evidence of construction or commissioning; and require an explicit statement before calling future capacity “production”. The official sources cited in this article document regulatory decisions and project objectives, but do not, on their own, provide a consistent measurement of current production. That is the limit of the conclusion: announcements and approvals explain what is intended to be built and what support is permitted; they do not replace evidence that chips are already coming off the line.