Key takeaways
- Volume is the first filter, because it decides whether tooling cost can be recovered at all.
- Geometry is the second filter. Volume tells you the economics; geometry tells you what is physically possible.
- Material is the third filter. Not every process runs every material, and the property you need most narrows the list further.
- Every process sits somewhere between high tooling cost with low per-part cost, and no tooling cost with a higher per-part cost.
Three filters decide which plastic manufacturing method fits your part, applied in order: how many units you need in year one, what shape the part has to be, and what material properties it needs. Get the order right and one or two processes usually survive the filter. Get it backward, and you end up quoting injection molding for a part you will only ever build five hundred of, or 3D printing a part that needs to survive a drop test at scale.
This guide covers the five processes worth knowing for a plastic part: injection molding, 3D printing, urethane casting, CNC machining, and vacuum forming (thermoforming). It gives you the filter logic and a selection table you can apply to your own part. It does not cover enclosure design rules in detail; that is a separate decision once you know which process you are designing for.
The five processes worth knowing
Each of these processes trades tooling cost against per-part cost differently, which is why the “right” choice depends entirely on volume and geometry, not on which process is newest or most talked about.
- Injection molding. Molten plastic is injected into a hardened steel or aluminum tool under pressure, then cooled and ejected. High up-front tooling investment, very low cost per part at volume.
- 3D printing (additive manufacturing). A part is built layer by layer from a digital file, with no tooling at all. Fused deposition modeling (FDM), stereolithography (SLA), and selective laser sintering (SLS) are the three variants most founders encounter.
- Urethane (polyurethane) casting. Liquid urethane resin is poured into a silicone mold, typically made from a single 3D-printed or machined master pattern, and cured. A bridge between prototyping and full tooling.
- CNC machining. A block of plastic stock is cut away with rotating tools to leave the finished part. No tooling, but material is subtracted rather than formed, which limits some geometries.
- Vacuum forming and thermoforming. A heated plastic sheet is drawn over a single-sided mold using vacuum pressure (or, in the pressure-forming variant, positive air pressure from behind the sheet for finer detail). Suited to large, thin-walled parts.
Volume is the first filter
Volume decides whether tooling cost can be recovered at all. Every process above has a volume range where it is the economical choice, and outside that range the same part can cost several times more per unit.
Under 100 units, 100 to 5,000, and above
| Annual volume | Process most likely to win | Why |
|---|---|---|
| Under 100 units | 3D printing or CNC machining | No tooling to amortize; per-part cost is higher but total spend is lower than cutting any mold |
| 100 to roughly 5,000 units | Urethane casting or aluminum (“soft”) tooled injection molding | A silicone urethane-casting mold or an aluminum injection mold costs a fraction of a production steel tool, and a silicone mold can produce roughly a few dozen to a few hundred parts before it degrades |
| Above roughly 5,000 to 10,000 units per year | Steel-tooled injection molding | Steel tooling costs more up front but drives per-part cost down far enough to recover that investment, and steel tools are rated for anywhere from roughly 100,000 to over 1,000,000 cycles depending on the tool class |
Mold builders commonly classify tools by expected cycle life, often labeled SPI Class 101 through 105, running from a high-cycle production class down to a low-cycle class meant only for prototype runs. This is more a common industry convention than a single published standard, so confirm which class your vendor is quoting before comparing tools on cycle life alone.
These thresholds move with part complexity and material, so treat them as a starting filter, not a fixed rule. A simple, small part can push the injection-molding break-even lower; a large or highly complex part can push it higher.
Geometry is the second filter
Volume tells you the economics; geometry tells you what is physically possible.
What each process physically cannot make
- Injection molding requires a part that can be ejected from a two-piece (or more complex, multi-action) tool, which means draft angles on vertical walls and careful handling of undercuts.
- 3D printing has almost no geometric restriction, including internal channels and undercuts that no other process on this list can produce without secondary operations, but layer lines create a rougher surface and directional strength than molded parts.
- Urethane casting inherits most of the geometric freedom of the printed or machined master pattern, but a silicone mold still needs to be pulled apart to release the part, so the same draft and undercut logic as injection molding applies, just with more tolerance for a manual pull.
- CNC machining is limited by tool access. A cutting tool has to physically reach every surface, so deep internal cavities, undercuts, and certain internal threads are difficult or impossible without multi-axis setups or multiple operations.
- Vacuum forming and thermoforming use a single-sided mold, so they cannot produce undercuts or features on the hidden side of the sheet without a split or mechanized tool, and wall thickness thins unevenly as the sheet is drawn deeper into the mold.
Material is the third filter
Not every process runs every material, and the property you need most (impact resistance, transparency, flame rating, flexibility) narrows the list further.
- ABS: low cost, good impact resistance, the default for enclosures and housings; runs in injection molding, 3D printing (FDM), and CNC machining.
- Polycarbonate (PC): higher cost, higher impact strength and heat resistance, used for protective shields and lenses; runs in injection molding and CNC machining, and is available in some SLA-adjacent resins for prototyping.
- Nylon (PA): wear resistant, used for gears and snap-fit features; common in injection molding, SLS 3D printing, and CNC machining.
- Polypropylene (PP): low cost, good fatigue resistance, the standard choice for living hinges; runs well in injection molding, less well in most 3D printing processes.
- TPU and other TPEs: flexible, used for grips, gaskets, and overmolds; injection moldable and increasingly common in FDM printing, but not a fit for CNC machining or vacuum forming.
If your part needs a specific flame rating, flammability classification for plastics used in devices and appliances is governed by a published testing standard (UL 94), with tiers from HB up through 5VA depending on how quickly a sample self-extinguishes and whether it drips burning material. Confirm the rating for your actual resin grade and wall thickness, since flame rating depends on the specific compound and thickness, not the base polymer alone. Material grades themselves are sorted using a formal classification system, ASTM D4000, which groups plastics by generic family and measurable properties rather than by trade name.
Cost structure: per-part cost versus up-front cost
Every process on this list sits somewhere on a spectrum between “high tooling cost, low per-part cost” (injection molding) and “no tooling cost, higher per-part cost” (3D printing, CNC machining). Urethane casting and aluminum-tooled molding sit in the middle.
Why the cheapest per part is often the wrong answer
A steel injection-molded part can look dramatically cheaper per unit than a CNC-machined or 3D-printed equivalent, but that comparison only holds once you have sold enough units to absorb the tool. A steel tool ordered for a volume that turns out to be only a few hundred units produces a worse outcome, on a per-unit basis, than simply machining or casting those units instead. A quick sanity check: divide the tooling quote by the units you are confident you will sell in the tool’s useful life, then compare that per-unit tooling cost against the per-unit savings molding offers over casting or machining. If tooling cost per unit still outweighs the savings, the lower-volume process is the better choice for now.
Surface finish and what customers actually notice
Dimensional tolerances for molded plastic parts are addressed by a dedicated international standard, ISO 20457 (currently the 2026 edition), covering injection molding and several related molding processes; it is a useful reference when a vendor’s quoted tolerance needs an independent benchmark, though the standard explicitly does not cover surface defects like sink marks or weld lines, which are judged separately.
Injection-mold surface finish is specified using a graded system that runs from a highly polished, mirror-like finish suitable for optical or cosmetic surfaces, through matte and stone finishes, down to textured bead-blast finishes that hide tooling marks and fingerprints on functional or lower-visibility surfaces. The finish grade is a real design decision, not an afterthought: a mirror finish shows every fingerprint and flow mark, while a textured finish hides both but reads as less premium on a consumer-facing surface.
What each process delivers out of the box
- Injection molding: finish is a direct copy of the tool surface, so it can range from mirror polish to heavy texture depending on how the tool was finished.
- 3D printing: FDM shows visible layer lines unless post-processed; SLA produces a smooth, near-injection-molded finish straight off the printer; SLS has a characteristic grainy, matte texture from the powder bed.
- Urethane casting: finish quality depends on the master pattern and the mold, so a well-finished master can produce a near-molded-quality surface.
- CNC machining: tool marks are visible unless the part is sanded or polished afterward, and internal corners always carry a radius left by the cutting tool.
- Vacuum forming: finish generally mirrors the sheet stock’s own surface, which is usually a manufactured gloss or texture rather than something the process itself controls.
A selection table you can apply to your own part
| Your situation | Process to start with |
|---|---|
| A handful of parts to check fit and function | 3D printing (FDM for form checks, SLA for cosmetic or fine-detail checks) |
| Dozens to a few hundred parts, need production-representative material | Urethane casting or CNC machining |
| A few hundred to a few thousand units, first production run | Aluminum-tooled injection molding |
| Several thousand units a year and confirmed demand | Steel-tooled injection molding |
| Large, thin-walled part (housing panel, tray, liner) | Vacuum forming or thermoforming |
| Small quantity of a part with tight tolerances and no undercuts | CNC machining |
How this decision interacts with your enclosure design
The process you choose constrains what your enclosure design can and cannot do, so this decision has to happen early, not after the industrial design is locked. A design drawn assuming injection-molded draft angles and wall thicknesses needs rework if the first run turns out to be CNC-machined or urethane-cast instead. If you are still deciding where to manufacture your product, settle the process question first: geography and vendor selection both follow from what the part needs to be made of and how.
Once you know which process fits your volume and geometry, the next question is who will run it, covered separately in how to vet a hardware development company and, for a narrower first cut on manufacturer type, how to find the right manufacturer for a hardware product. For low-volume verification work rather than production, a fab shop running make-to-print jobs is a different kind of vendor than a full development partner; see what a rapid prototyping shop actually delivers. Whichever process you land on, validate the resulting parts against real test methods rather than a CAD model’s assumptions; a prototype testing checklist is a reasonable place to start.
3D printing deserves a closer look on its own, since it now does far more than rough form checks. For a fuller picture of what has changed and what still has not, see how 3D printing is transforming product development.
Frequently asked questions
What is the cheapest way to manufacture a small number of plastic parts?
For a handful of parts, 3D printing is usually cheapest because there is no tooling cost at all. For a few dozen to a few hundred parts in a production-representative material, urethane casting or CNC machining typically costs less overall than cutting any injection mold.
How many parts can you get from a urethane casting mold?
It varies widely by part geometry and resin chemistry. Simple parts in a well-designed silicone mold can produce well over a hundred castings; complex geometries with undercuts or thin sections wear the mold faster and may only yield a few dozen. There is no single universal number, so ask your vendor for an estimate based on your specific part.
Can 3D printed parts be used in a final product?
Sometimes, for low-volume runs or parts where layer lines and slightly lower directional strength are acceptable. SLS nylon parts, in particular, are used in short-run production. For higher volumes or parts needing isotropic strength, injection molding remains the better fit.
What is the difference between vacuum forming and injection molding?
Vacuum forming draws a heated plastic sheet over a single-sided mold and works best for large, thin-walled parts at low tooling cost. Injection molding injects molten plastic into a two-sided tool under pressure and produces more complex, precise geometry at a much higher tooling cost, which only pays off at higher volumes.
Do I need to pick a manufacturing process before finishing my industrial design?
Yes, in general. Draft angles, wall thickness, rib design, and undercut handling all depend on which process will make the part, so locking the process early avoids redesigning the enclosure later.
What plastic should I use for an enclosure that needs a flame rating?
Start from the flammability class your product needs to meet (governed by the UL 94 test standard) and confirm which resin grade and minimum wall thickness are actually certified to that class, since flame rating depends on the specific compound and thickness tested, not the base polymer alone.
Where Inventornest fits
Inventornest handles product design and development in-house, and we retain responsibility for engineering coordination and design clarification when production is handled by our manufacturing partners. If you have a part and are not yet sure which process fits it, get a quote and bring your volume target, geometry, and material requirements. For an overview of how this fits into a full product development and manufacturing engagement, that page covers the range of services available end to end.
