Union Semiconductor Hefei Co., Ltd.

It earns per-order processing fees for turning customer-supplied wafers into packaged, tested chips, with each fee built from the cost of materials consumed plus a negotiated service charge that varies by equipment type, processing time and market demand. Essentially all of its disclosed product revenue comes from this single service line, and that fee income has converted into a positive bottom line on a sustained basis rather than in occasional years only.

It scales by adding physical processing capacity in discrete blocks, sized by how many wafers it can run through in a given period, and then has to draw in enough customer orders to fill that new capacity before the equipment behind it earns back the depreciation already running against it. By its own account, capacity brought online recently was not immediately matched by a comparable rise in orders. Among the broad population of companies CompanyGraph classifies as running this same kind of fixed-capacity conversion system, it sits within a very large group, so the general shape of the business is widely shared rather than rare.

Its core input, the wafer itself, comes from its own customers rather than from an open market, which makes it dependent on those same customers for what it processes, not just for who buys the result. It also depends on a concentrated handful of suppliers, on production equipment and some materials imported mainly from one overseas country, and on specialist technical staff and specific process know-how to run its lines. Separately, CompanyGraph's mapping places it downstream of a wide band of other industries that ultimately feed into what it needs to operate.

A small number of chip-design customers account for most of its revenue, so its near-term order flow is concentrated in relatively few hands rather than spread broadly. Beyond that direct customer, it also supplies further downstream to the panel manufacturers those customers designate to receive the finished chips, and its chips ultimately reach everyday consumer electronics. CompanyGraph separately maps it as feeding a small number of other industries beyond its direct customers.

CompanyGraph's broader data shows that running a fixed-capacity conversion business of this general kind is common, shared by a very large population of companies, so operating one at all sets nothing apart. The company itself claims a narrower distinction: that it is among a small number of mainland Chinese firms running both smaller and larger wafer full-process lines side by side, and among a small global group combining bump fabrication, wafer testing and both major display-driver packaging formats in one operation. That narrower claim is the company's own account of its position; CompanyGraph has not independently confirmed whether rivals can replicate it.

By the company's own account, a customer cannot simply move its packaging and testing work to a new supplier at will: the customer's own chip designers must first put any prospective packaging partner through a lengthy qualification process before committing to a long-term relationship. For automotive customers specifically, a named quality certification is described as a condition of entry, which narrows the field of suppliers a customer could switch to in the first place. Once a customer has already qualified this company for a given process, repeating that qualification elsewhere is itself the friction that discourages a switch.

By its own account, what limits it is less the physical processing ceiling itself than the slower-moving constraints around that ceiling: keeping pace with technology leaders elsewhere in the industry, a shortage of experienced specialist staff in its home market, and the long qualification process each customer requires before shifting meaningful volume to it. It also states that newly built capacity took time to be matched by customer orders, so equipment can be depreciating before it is fully used. The general pattern CompanyGraph tests this against is one where a fixed processing line caps how much a company can convert in a given period; here the company's own disclosures point more toward technology and qualification lag than toward that physical ceiling itself.

By its own account, the risks it lists first are technological: falling behind on process technology, the gap between its own capability and global industry leaders, and losing the specialist engineers who run its lines, followed by intensifying competition. Behind those, it names concentration in a small number of customers that account for most of its sales, and a similarly concentrated small group of suppliers. It also names reliance on equipment and materials sourced mainly from one overseas market and exposure to the cyclical swings of the consumer electronics markets its chips ultimately serve.

It answers to securities regulators for its public listing and to a local environmental regulator for its discharge permit, and has already drawn a fine for exceeding discharge limits in the past. Because its equipment and some materials come mainly from Japan and its major customers are concentrated in Taiwan, it names cross-border trade and tariff restrictions as a pressure that could disrupt equipment, materials, or customer orders. Its overseas sales and purchases also expose it to swings in foreign exchange rates against its home currency.