Ask a factory how much capacity it has and you may get a number of machines, or a number of shifts, or a figure in square feet. None of those is capacity. Capacity is the amount of conforming work a system can complete in a period, and in a mixed-work shop the binding constraint moves from week to week.
On Monday the constraint may be a five-axis machine. On Wednesday it may be the welder who can pass a particular certification. On Friday it may be the inspection bench, where every part has to be measured before it can be packed. A shop that quotes a single capacity figure is quoting an average that no individual order will ever experience.
This is why two factories of similar size can behave very differently on the same job. The difference is not the machine list. It is the balance between operations, and the amount of slack deliberately left at the point where variability accumulates. The operation behind that balance is set out on the company page.
In a shop that runs one product for a month, capacity is easy: machine hours minus downtime. In a shop that runs mixed work, every switch costs time that produces nothing. Fixtures are swapped, programs are proven, first articles are measured, and the machine sits idle through all of it.
The cost of a changeover is not only its duration. It is also the yield lost while the process settles, and the attention it takes from the person who would otherwise be watching two machines. A shop that runs ten short jobs in a week pays that cost ten times, and the tenth payment is no cheaper than the first.
Buyers feel this as price and as lead time, though neither is presented as a changeover charge. The honest position is that a mixed shop cannot offer the unit economics of a dedicated line, and should not pretend otherwise. What it can offer is flexibility, provided the buyer understands that flexibility is purchased with setup time.
Spread work across four facilities in Shanghai and the capacity question becomes a routing question. A part may be cut in one building, formed in another, welded in a third, and finished elsewhere. Each move is a queue, and each queue can be the constraint on a given day.
More than 50 employees across those sites gives real depth, but depth is not the same as surplus. Skilled welders and programmers are the scarce resource in most fabrication businesses, and a shop with plenty of floor space can still be short of the one person who can sign off a procedure.
The practical consequence for a buyer is that capacity should be discussed per operation, not per factory. Ask which step is loaded. A shop that answers that question precisely is telling you where it would add a shift if it could, which is also where your job is most likely to wait.
The physical footprint, 53,819 sq ft, matters less than the arrangement of it. Distance between operations adds handling time that never appears on a machine schedule but always appears in the calendar.
A capacity figure is normally calculated on the assumption that everything runs. Material arrives on time, programs are proven, the first article passes, nobody is off sick, and the machine does not need repair. Every one of those assumptions fails occasionally, and the failures cluster.
The result is that nameplate capacity is a ceiling, not an expectation. A shop that plans to 100 per cent of that ceiling has no room for recovery, and a single broken tool cascades through the week. A shop that plans to a lower figure looks less impressive on paper and delivers more reliably in practice.
Buyers should be suspicious of a factory that claims to be full and still offers short lead times. Those two statements cannot both be true unless something is being redefined. The more credible answer is a specific window, with the constraint named.
A dedicated line produces a distribution of completion times that is narrow. A mixed shop produces a wide one, because each job inherits the variability of everything scheduled around it. This is a structural property, not a discipline problem.
It follows that a mixed shop cannot promise a single date with high confidence. It can promise a window, and it can protect that window by holding buffer at the operations that vary most. Buffer is not waste. It is the mechanism that converts a wide internal distribution into a narrow external promise.
The buyer's side of the bargain is to give the planner something stable to plan against. A frozen quantity for a fixed period is worth more than a forecast that changes weekly, because the first can be loaded and the second cannot.
Capacity claims are testable without an audit, and the test takes one conversation.
A shop that answers all five without consulting a screen is describing a system it understands. A shop that answers none is describing a brochure.
Flexibility in a mixed shop comes from three sources: multi-skilled operators, tooling that can be changed quickly, and a planning habit of leaving one window open. The first two are capital and training decisions. The third is discipline, and it is the one most often abandoned under pressure.
An open window looks like waste to anyone measuring utilisation, which is why it disappears in busy months and why busy months are when delivery performance falls. A shop that protects that window through a busy period is making a deliberate choice to be less efficient and more reliable.
Capacity is the amount of conforming, inspected, packed work a system can complete in a period, at the mix it actually runs, with the changeovers that mix requires. Anything else is a nameplate figure.
Hold that definition and the conversation changes. Instead of asking how big a factory is, ask which operation is loaded and what a changeover costs. The answers will tell you more about delivery reliability than any floor area ever could. Buyers who want to see how the work is planned can look at the customisation notes, the parts under other parts, stainless exhaust systems and titanium components, and the dealership terms, then test the claim operation by operation.
