Every new plastic part hits the same fork in the road: print it, or cut a mold. The answer isn't a matter of taste. It comes down to how many you need, what the part has to survive, and how much time and money sit between the CAD file and the first good piece.

Here is how the two differ, and where each one runs out of road.

What injection molding does

Thermoplastic pellets get melted and pushed under pressure into a machined steel or aluminum mold. The plastic cools, the mold opens, the part drops out. Cycle times range from a few seconds to around a minute, and the same tool will repeat that hundreds of thousands of times.

The materials are the ones you would spec on a production part: ABS, PC, PA (nylon), POM, PP, PMMA, TPU, plus glass-filled and other reinforced grades.

Typical work includes automotive interior and under-hood components, electronics enclosures, medical device housings, connectors, robotics parts, and industrial equipment.

The strength of the process is consistency. Once the tool is proven and the process window is set, part 1 and part 50,000 measure the same. That is the whole pitch.

The catch is the tool. Design, cutting steel, fitting, and first-shot trials cost real money and take weeks. Spending that on a geometry you might redraw next month is a bad trade.

What 3D printing does

Additive manufacturing builds a part layer by layer straight from the CAD file. No mold, no fixture, no minimum order.

Common routes: FDM with PLA, ABS, nylon and TPU; resin systems such as SLA; powder bed processes such as SLS and MJF with nylons.

It earns its keep on prototypes, fit checks, ergonomics studies, one-off jigs, and shapes a toolmaker would struggle with, including internal channels, lattice structures, and undercuts that would need a side action in steel.

Change the model tonight, print again tomorrow. That loop is the point, and it is why printing dominates the messy early phase while the design is still moving.

The two side by side

Treat the numbers as ballpark. Part size, resin and geometry move them.

1. Volume

Volume decides most of this.

Ten housings for a test rig: print them. You hold parts this week and lose nothing if the design changes.

Ten thousand of the same housing: the mold pays for itself somewhere in the first few hundred to a couple thousand pieces, depending on size, cavity count and resin. Past that, each unit is mostly material and machine time.

The crossover point moves. Small, simple parts cross over early. Large parts with thin walls and tight tolerances cross over later. Ask a supplier to run it at your real annual quantity instead of a guess.

2. Tooling cost

Printing has almost no setup cost. You send a file.

A mold carries design, steel, machining, fitting, trials, and usually a round or two of adjustment. Complexity drives the price: side actions, lifters, tight tolerances, multi-cavity layouts, and polished or textured cavities all add to it.

So comparing quotes on unit price alone misleads. Compare program cost: tooling, parts, post-processing, inspection, freight, and the cost of a design change six months from now.

3. Unit cost

Printed parts cost about the same each no matter how many you order. Machine time and material scale with quantity. Order a hundred instead of ten and you pay close to ten times as much, with maybe a small volume break.

Molded parts invert that. High fixed cost, low marginal cost. The curves cross, and the crossing point is the decision.

Give suppliers an annual quantity when you request a quote. Without it they are guessing, and a quote built on a guess tells you nothing.

4. Dimensional accuracy

Molding holds tight tolerances well once the tool is qualified and the process is monitored. Shrinkage is predictable for a given resin, and a good mold is cut to compensate. That matters when parts have to mate with other parts, batch after batch.

Printing accuracy depends on the process, the material, orientation on the build plate, support strategy, and how much the part warps on cooling. Layer height sets a floor on fine features. Post-processing moves dimensions again.

Either way, mark the critical dimensions on the drawing and agree on how they will be measured before production starts. Most arguments about out-of-tolerance parts are really arguments about a spec nobody wrote down.

5. Surface finish

A molded part takes its surface from the cavity. Polish the steel and you get gloss. Etch it and you get a matte texture. Every part in the run matches.

Printed parts carry the signature of the process: layer lines, support marks, stair-stepping on curves. Sanding, vapor smoothing, priming and painting all help, and each adds handling time per part.

If the part sits where a customer can see it and you need thousands that look identical, molding wins on finish. On an internal bracket, nobody cares.

6. Materials and performance

Molding opens the full catalog of production thermoplastics, including grades built for heat, chemicals, UV exposure, wear and dimensional stability. Properties are close to isotropic and the resin supplier's datasheet applies.

Printing materials keep improving, but a printed part is weaker across layer lines than along them. Infill, orientation and print parameters all change the outcome. A printed nylon part and a molded nylon part are not the same part.

If the component will carry load, see heat, meet solvents, or sit in sunlight for years, check material data under those conditions. Do not pick a process on price and hope the material holds.

Choose 3D printing when

• You need fewer than a few dozen parts.
• The design is still changing.
• You need something in hand this week.
• The geometry would be miserable to tool.
• You are checking fit, assembly or ergonomics.
• Nobody wants to approve tooling money yet.

A robotics team printing five revisions of a housing to sort out cable routing and service access, before anyone signs off on steel, is using the process correctly.

Choose injection molding when

• Quantities run to thousands or more.
• Tolerances have to hold across batches.
• Cosmetic consistency matters.
• You need a production-grade resin.
• The design is frozen.
• Unit cost at scale drives the business case.

Most projects use both

The usual sequence:

Print prototypes, test them, fix the design, run a DFM review, build the tool, mold production parts.

Printing surfaces problems while they are still cheap. The DFM review catches the rest, meaning wall thickness, draft, sink, rib-to-wall ratios and gate placement. Then steel gets cut once, on a design worth cutting it for.

Going straight to tooling is how companies end up paying for mold modifications nobody budgeted.

How YIXIN PRECISION handles plastic parts

YIXIN PRECISION (YMP) runs CNC machining, injection molding, sheet metal fabrication, die casting, surface treatment and assembly. That range matters here, because the honest answer to "print or mold?" is sometimes neither. At 20 pieces with tight tolerances, a machined part can beat both.

On plastic work, our engineers look at resin, wall thickness, draft angles, ribs, bosses, shrinkage, tolerances, assembly requirements, expected volume and finish, then recommend a process. Often that means printing now and molding later.

We take prototypes and small batches too, which suits robotics, medical devices, automation equipment, electronics and precision instruments. Those are industries where a design goes through several validation rounds before it is final.

Dimensional control, process documentation, inspection and delivery performance are what decide whether a supplier is still usable two years in.

So which is better?

Wrong question. Printing handles uncertainty and low numbers. Molding handles volume, consistency and finish. Most plastic products around you went through both at different stages.

Before choosing, write down:

1. Annual quantity, and next year's
2. Target unit cost
3. Tooling budget
4. Tolerances that actually matter
5. Material and what it has to survive
6. Finish requirements
7. Mechanical and environmental loads
8. Expected product life
9. Lead-time limit
10. Whether volume will grow

Then send that list to a manufacturer before the design is frozen. Geometry locks in most of a part's cost, and geometry is cheapest to change while no steel has been c