Skip to content
Interface language: English
Oriens / Signal

China's solar industry: a cell, a module and a completed installation tell different market stories

Distinguish photovoltaic manufacturing stages, production capacity and installed capacity before researching a specific market.

Article language: English

Aeneia concept image: a woman adjusts a gear mechanism beneath two large pale discs.
In this article

A cell fits inside a module. A module goes into an installation. A factory may expand its capacity to make one of those products while statistics on new connections describe a different point in the chain. Put every number under the heading "Chinese photovoltaics" and the result may be an impressive chart but a poor answer to who would buy your service.

This matters for a company that does not intend to build power plants or sell modules. It might offer equipment for a particular production stage, packaging, information software or material for a manufacturer. Its market does not start with all installed capacity in China. It starts with the task of a particular participant in the chain.

From silicon material to a module

An expert commentary published on the MIIT website on 28 July 2026 distinguishes silicon material, wafers, cells and modules. This is a useful division of industrial stages. We use it as an industry description, not a statement of legal requirements or an assessment of a factory.

The US Department of Energy explains the manufacturing basics: in silicon technology, wafers are processed into cells and connected cells form modules. A complete system includes other components, such as mounting structures and power electronics. This describes the technological chain, not the specifications of a selected product.

Research must therefore name the subject of the conversation. A wafer manufacturer and a finished-module manufacturer may work in the same broad industry but need different materials, equipment and skills. "PV" means photovoltaics; it does not identify a stage, product or way of working together.

Consider a fictional Polish company offering batch identification in a factory. Its proposal should explain what object is identified and at which point in the work. There is no need to claim relevance to the whole industry. A precise point of use makes it easier to find a participant who can assess the proposal substantively.

Another example is a packaging producer. Do not derive wafer-packaging requirements from photographs of finished modules in transit. If proposing reusable B2B packaging (see the guide to reusable B2B packaging (in Polish)), establish its return route and preparation for reuse. An external observation helps frame a question about the stage and object to protect. Confirming a solution needs data and suitable technical assessment. Market research should not award itself a materials-engineering degree along the way.

Apply the division to the companies collected. Record the specific product and role described for each. If a participant describes making both cells and modules, keep both scopes instead of choosing one for the convenience of a spreadsheet. If the description is ambiguous, do not guess from a company name. A list with gaps to check later is more useful than a complete classification built on assumptions.

For a batch-identification provider this can change the first conversation. One discussion concerns information passed between stages inside a factory; another concerns recognition of a product reaching its recipient. These are hypotheses to check, not problems confirmed across China's industry by the research so far.

Installed capacity describes use, not module production

At a conference on 30 July 2026, China's NEA reported 71.77 GW of newly connected PV capacity in the first half-year and 1,272 GW of total capacity. The first figure is an increase over the period; the second is the stock at its end. Neither is module production.

Keep the unit and period beside every value. Capacity added in six months is not the same as capacity operating for years. A factory's production capacity describes what it can make, not automatically what it did make. A count of components produced does not itself say where or when they were used. Moving a column label can change the meaning of the analysis even when its figures are correct.

Start a comparison with the question, then choose the measure. A company seeking buyers for manufacturing equipment needs a particular plant type and process task. A supplier of materials for an installation contractor needs a different relationship. Overall installed capacity can provide background but does not give either company its number of potential customers.

The same principle applies to researching China without one aggregate number (see the guide to researching a Chinese market segment (in Polish)): data must fit the decision. A large market does not yet tell you whether a stage is open to your offer, who assesses it or what your company can actually deliver.

Check time boundaries before placing numbers together. "In 2026" may describe a completed half-year in one publication and a full-year plan in another. Do not present them as two measurements of the same event. Retain the author's description and publication date beside each value, so a completed period and an intention are distinguishable without a footnote hunt.

A team presentation can use three separate sentences: what the source says about manufacturing capacity, actual production and use. If data exists for only one, leave the others open. Do not fill a blank column with a figure from the neighbouring stage. A missing measure is a reason for further research, not a label-design problem.

Three different quantities can sit side by side

On the Trends in Photovoltaic Applications 2026 page, IEA PVPS describes global module production in 2025 as above annual installations but below available manufacturing capacity. This is a worldwide relationship, not a calculation of utilisation at a particular Chinese business.

The observation exposes a category error. Do not assign a global relationship to any company simply because it makes modules. Nor assume capacity equals sales or sales equal new installations. Each needs its own definition and source. A polished legend cannot repair a report that mixes them into one series.

Aeneia conceptual photography

Company scope creates a similar issue. A group presentation may describe several stages while a conversation concerns one company or factory. Establish which entity a statement concerns and whether it describes current activity, its own claim of capability or a future announcement. You need not settle everything from the first brochure, but should know which statement belongs to whom.

Partner verification (see the guide to verifying a potential partner (in Polish)) helps connect a name to an actual entity. It does not automatically confirm line output, module quality or the economics of cooperation. Those are separate assessments requiring appropriate expertise and material.

A comparison should retain the geographic scope too. An activity described in China, a group's global presentation and an offer for a particular buyer may have different boundaries. If you need a factory, seek information about that factory. If considering a trading entity, identify its actual role. Do not attribute everything under a shared brand to one company.

Record what the manufacturer's material does not explain. It may describe a product well without explaining how to work with a new supplier, or present a group without a relevant contact. These gaps differ from a lack of industry-scale statistics. Filling them needs closer organisational research, not another global number.

The offer should answer the right question

Return to the fictional batch-identification solution. A Polish supplier can prepare a short account of the object identified, the use of the data and the limits of its offer. A potential contact should be able to judge whether the task exists in their work. Adding "for photovoltaics" does not replace that description, even against a very blue panel.

Research questions can concern how an organisation describes its scope and the materials it offers partners. Do not assign it needs from an industry statistic. If evidence about a specific task is missing, record a hypothesis. A conversation or further material is needed to test it.

A listed 1688 price (see the article about listed prices on 1688 (in Polish)) relates to a specific offer and its conditions, not the whole industry. A cheap or expensive listing alone does not identify the stage or even the product being compared. First establish that the objects match; then discuss trade data with the right specialists.

If cooperation involves a private-label product, distinguish who designs, who manufactures and what is adapted. OEM and ODM (see the guide to OEM and ODM arrangements (in Polish)) do not by themselves confirm rights or documentation. In photovoltaics too, identify the actual contribution and subject of cooperation instead of inferring them from a model name.

The fictional Polish company might use two brief application descriptions instead of one general brochure. One shows a task involving a specific object; the other shows how a recipient uses the information. Each can end by asking whether that scope matches the contact's work and what data is needed to assess it. Do not promise technical suitability before it has been checked.

Separate an organisational outcome from a technical one. A referral to the right department helps the contact continue, but does not confirm suitability. A request for specifications means further assessment, not an order. Refusal of one application does not describe the whole solar chain. These distinctions help learning without turning each signal into confirmation of a huge-market thesis.

Test the map against three fictional offers: one describes module manufacture, another a structure for a specific use, and the third a service preparing an installation. The shared word "solar" does not make them interchangeable. Identify a potential buyer of batch identification by the actual task it performs, not the word on its website.

For the first participant, the Polish company should identify the product and information passed onward. With the second, it must first establish whether and where a similar task exists. The third may not need that offer at all. Different answers are not a contradiction in the market; they reflect conversations with companies at different stages.

Use the distinction when reading references too. Experience with a finished installation does not by itself establish experience at a particular manufacturing stage. Supplying a component does not explain an entire project. Record the contribution the participant actually claims, without extending it beyond the source.

Only then return to the size of the relevant segment. You know which product, task and type of buyer you seek. The global installation figure may remain background but no longer has to be the only argument. Further research requires data fitted to the narrower relationship, not multiplying the whole industry's size by an assumed company share.

Trade data is an additional layer

Trade statistics can help examine flows of a product group, but only within the correct scope. Comparing HS data (see the guide to comparing HS trade data (in Polish)) starts with what the selected category includes. Do not use one item as a proxy for every industry stage without a basis. Retain the product description, period and unit.

A consumer channel is not a universal route to every industrial relationship. Retail CBEC imports (see the article on retail CBEC imports (in Polish)) concern a particular transaction type. For an industrial component or manufacturing equipment, first identify the different object and recipient, then the appropriate trade model. Product admission and requirements remain a matter for specialists.

Aeneia Oriens research may organise the segment, entities, local materials and limits of the data. It can help select next questions and contacts. It is not a manufacturer recommendation, price forecast or technical assessment of an installation. A useful industry map leads to a specific task; it need not predict the future of energy as well.

Aeneia / Aeneia editorial

Editorial article. Source links are included in the text.

About Aeneia

Brands we have worked with

  • L’Oréal
  • Garnier
  • Netflix
  • Carrefour
  • Warner Bros. Discovery
  • Biedronka
  • HBO Max
  • Wolt
  • Żabka
  • COSRX
  • Embassy of China in Poland
  • Mixa
Aeneia / Initium

From knowledge to a decision

Tell us what you want to put in order. We will agree on where to start.

Contact Aeneia

We partner with and use solutions from:

  • Google
  • OpenAI
  • Microsoft
  • Stripe
  • Anthropic
  • Linear