Two countries.
Two different manufacturing machines.

China counted 4.05 million manufacturing legal entities at the end of 2023 in a national census, not an estimate. The United States had 284,452 manufacturing establishments with paid employees in the same year. That is roughly 14 to 1 — and even China's 512,000 largest manufacturers alone outnumber every US establishment.

The number is not a verdict on quality. The US is still one of the two largest manufacturing economies by value added while running about one-fourteenth the plants: fewer, larger, more automated. What the ratio tells you is how many different kinds of maker you can reach in a week.

Anything on a short fuse that uses a shape which already exists. A stock item, decorated, runs domestically from 24 to 144 pieces against 500 to 1,000 in Asia, with no tool and no ocean. Reorders come back in 3 to 5 weeks. If origin is contractual, or the quantity is modest and the date is inside four weeks, domestic wins outright and we will say so.

The domestic promotional base is also genuinely deep in decoration — pad print, screen, embroidery, heat transfer. That depth applies to blanks, which is exactly what it is built for.

New tooling, decoration depth, packaging, and retail-quality finish at merchandise quantities. A bespoke shape, in merchandise volume, on a retail date is routine there and exceptional here.

The reason is density, not wages. A custom product needs a mold shop, a decorator, a printer, a packaging converter and an assembler to iterate together. In the Pearl River Delta all five are inside an hour, so a correction loop closes in days. Across US states the same loop runs in freight cycles, and every cycle is a week of the launch calendar.

Tooling economics decide it. Where thousands of mold makers compete, a bespoke steel tool is affordable at merchandise quantities. Where the toolmaking trade is small and highly paid, the tool cannot be justified and the program quietly becomes a stock shape with a logo on it.

Component depth compounds it. A shaped bottle is six trades — mold, closure, liner, coating, decoration, carton. Missing even one domestically turns a single purchase order into a multi-vendor assembly project with nobody accountable for the sample.

For a mid-complexity injection-molded item, a single-cavity steel tool runs $25,000.00 to $60,000.00 in the United States against $8,000.00 to $20,000.00 in China. Same geometry. The gap is toolroom labor and competition, not steel.

The calendar gap matters more. Domestic tooling lead time is 8 to 14 weeks against 25 to 40 days, and the first article sample lands 2 to 4 weeks after the tool domestically against 7 to 10 days. Domestic toolrooms are booked; Chinese shops compete for the slot.

Read the first two rows together and the whole decision is visible: domestic lines are efficient at volume and expensive at variety, so a custom tool needs ten times the quantity to make sense at home. On stock goods the domestic decorator wins by an order of magnitude.

Decoration is a trade, not a step. Pad print, screen, heat transfer, in-mold label, hot stamp, laser, soft-touch coat and electroplate all behave differently on the same part. The domestic base applies them to a blank; the Chinese advantage is that the decorator sits next to the mold, so the finish is designed with the geometry rather than applied to it afterward.

On color specifically: a Pantone reference behaves one way on a coated blank and another way in pigmented resin, where wall thickness, gate location and the resin's own base tone all move the read. That is why a color is approved on an actual molded part and never off a screen — and why the base that can run three correction loops in a fortnight produces the better match.

The bigger limit is which part of the product a color can even apply to. Domestically you generally cannot specify a custom PMS color for an entire product — the body comes in the supplier’s stock colors and only the logo is matched to your Pantone. Tooled production in Asia colors the resin itself, so the object is your color, not a stock shell wearing your mark.

Tooling is a fixed cost, so the only honest comparison is landed cost per piece with the tool amortized into the run. Using midpoints — China at a $14,000.00 tool, $2.80 piece and $0.55 freight and duty; the United States at a $42,500.00 tool and $5.75 piece with no freight:

Two things fall out. The tool is the entire argument at small quantities and almost irrelevant at large ones — at 5,000 pieces it is 46% of the Chinese landed cost and 60% of the domestic one; at 100,000 it is 4% and 7%. And reshoring a custom program is not a 10% decision: in this model it roughly doubles landed cost. A tariff would have to be extraordinary to close a 1.8x gap, and it is the piece price, not the tariff, doing most of the work.

Door to door, a full custom program runs 14 to 22 weeks domestically against 12 to 16 weeks from Asia. The ocean leg is offset by faster tooling and faster sampling.

A custom program has four clocks and only one of them is shipping: tooling, sampling and correction, production, and transit. Moving domestically shortens the fourth and lengthens the first two, which is why programs reshored “to save time” usually lose four to six weeks net. The ocean is predictable. A toolroom queue is not.

A steel mold is an asset, not a fee. It belongs to the program, it stays dedicated to it, and it can be moved between qualified factories — which is what makes a second source possible at all. Single-origin risk is answered with a second qualified factory, not a second continent.

Tariff exposure is real, and it is the reason we model both origins; note that it applies to the piece price, not to the tool. Freight volatility is usually answered with cube engineering — a redesigned carton often recovers more than a rate negotiation. Quality is answered with third-party inspection at the same AQL standard, which we run either way.