The short answer
For a part weighing 50–200 g, with no special material, the break-even between machining and injection moulding usually falls somewhere between 200 and 500 pieces. Below that, machining is cheaper once you count the tool. Above roughly 1,000 pieces, moulding is almost always cheaper and stays cheaper.
Two things move that number more than volume does: how many times the design will change, and whether the material can be moulded at all. Both are discussed below, because in practice they decide more projects than the arithmetic does.
What is actually being compared
The two processes cost money in different places. Machining front-loads almost nothing and charges per hour. Moulding front-loads the tool and then charges very little per part.
| CNC machining | Injection moulding | |
|---|---|---|
| Up-front cost | Programming only, typically a few hundred dollars at most | Tool, typically USD 2,000–15,000 for a small to medium part |
| Cost per part | Hourly rate × cycle time; falls slowly with volume | Material + machine time + labour; falls sharply with volume and cavities |
| Tooling lead time | None | 25–40 working days, plus T1 samples |
| Typical first delivery | 15–20 working days | 45–60 working days including tooling |
| Design change | Edit the program, no cost | Modify or cut a new tool |
| Tolerance, as moulded | ±0.02–0.05 mm achievable | ±0.1–0.3 mm typical, plus warp |
| Material range | Any machinable stock shape | Thermoplastics that flow; not UHMWPE, not PTFE in the usual sense |
| Surface | Tool marks unless finished | Mould texture, repeated exactly |
The mistake buyers make is comparing only the piece price. The quoted moulded price of USD 1.80 against a machined price of USD 18 looks like a ten-to-one saving until the USD 5,000 tool is divided across 150 parts — which makes the moulded part USD 35 each.
The arithmetic, in one line
Break-even quantity = tool cost ÷ (machined piece price − moulded piece price)
Worked example, a 100 g POM bracket with modest features:
- Machined: USD 18 per part at 100 pieces, USD 14 at 500
- Moulded: USD 5,000 for a two-cavity tool, USD 1.80 per part
- Break-even: 5,000 ÷ (18 − 1.80) ≈ 300 parts
At 300 pieces the two routes cost the same. At 150 pieces machining saves about USD 2,500. At 3,000 pieces moulding saves about USD 43,000.
Now the same part with a design change at 250 pieces: the tool is modified for USD 800 and another two weeks, or the machined route simply absorbs the change in the next program. That is why the honest answer to “should I tool this?” is usually “not until the design has stopped moving”.
When machining wins even at volume
The material cannot be moulded. UHMWPE is compression moulded or ram extruded, then machined; it does not injection mould in the ordinary way. PTFE is pressed and sintered, then machined. PEEK can be injection moulded, but the grades used for wear parts — carbon fibre, graphite, bronze filled — are often only available as stock shapes, and the tooling for high-temperature engineering plastics costs several times what a POM tool does. If the drawing says PEEK CF30, machining is not a compromise, it is the only route.
The geometry is thin or deep. Moulding needs draft, uniform walls and a gate position that does not leave a weld line where the part must be strong. A part with a 0.8 mm wall, an undercut, or a bore that must be dead straight is easier to machine than to tool.
The volume is real but small. 200 parts a year for five years is still 200 parts a year. A tool amortised over annual demand, not over a purchase order, is a different calculation — and it usually loses.
The tolerance is tight in one place only. Machining holds ±0.02 mm on the two dimensions that matter without special measures. Moulding holds a general tolerance across the part, and tightening one dimension means tool work, process capability studies and often 100 % inspection.
When moulding wins earlier than people expect
The part is small and the tool is cheap. A 5 g cap in a four-cavity tool at USD 2,500 breaks even at a few thousand pieces, and at 50,000 pieces the piece price is a fraction of the machined one.
The part is a consumable. Anything that wears out and is reordered — caps, plugs, clips, covers — has a lifetime volume far above its first order.
The part must be identical, every time. Moulding reproduces the same surface, the same weight, the same colour. Machining reproduces the same drawing, with tool marks and a slightly different finish from batch to batch.
Assembly is designed in. Moulded snap fits, living hinges and integrated bosses remove parts and labour that machining would have to add back as separate items.
The comparison table buyers actually need
| Question | Leads to machining | Leads to moulding |
|---|---|---|
| Annual volume | Under 500 | Over 2,000 |
| Design status | Still changing | Frozen, or changed by tool modification |
| Material | UHMWPE, PTFE, filled PEEK, ceramic-filled | Unfilled or glass-filled POM, PA, PP, ABS, PC |
| Tolerance | ±0.05 mm or tighter on functional features | ±0.2 mm acceptable on most features |
| Wall thickness | Varies, or under 1 mm | Uniform, 1.5–3 mm |
| Surface | Machined finish acceptable | Texture or colour must repeat exactly |
| Time to first parts | 15–20 working days | 45–60 days including tooling |
| Risk if it does not fit | Rework the program | Cut or modify a tool |
A bridge that usually works
Most projects do not choose one route and stay there. The sequence that keeps risk low:
- Machine the prototype or first article. It validates fit and function in the real material, often in two weeks.
- Machine the pilot batch — typically 100–500 pieces — to supply early customers or field trials. At this stage the design still moves, and every change is free.
- Freeze the design, then tool. By now the part has been in service, the tolerances that matter are known, and the tool is cut once.
- Keep the machined route alive for variants. Low-volume colours, sizes or one-off replacements stay machined while the main variant is moulded.
The one rule that makes this work: use the same datum scheme and the same drawing for both routes. If the machined prototype is dimensioned differently from the moulded part, the first moulded samples will not assemble with the prototypes, and the comparison you were relying on is worthless.
Lead time, stated plainly
| Stage | Machining | Moulding |
|---|---|---|
| Quotation | 2–3 working days | 3–5 working days, including DFM review |
| Tooling | — | 25–40 working days |
| T1 samples | — | 1 week after tool trial |
| Sample approval | 5–7 days for first article | 1–2 rounds typical |
| Production | 15–20 working days | 15–25 working days |
| Total to first saleable parts | 3–4 weeks | 8–12 weeks |
The gap is roughly two months. For a product launch that is the real cost of tooling, and it belongs in the comparison next to the piece price.
How to decide this week
Answer four questions and the route is usually obvious:
- How many parts will you order in the next 12 months, and how many over the product’s life?
- Is the design frozen, or is there a known change coming?
- Does the drawing name a material that can be injection moulded?
- Is the tight tolerance on one feature or across the whole part?
If the answers are “under 500”, “not frozen”, “UHMWPE” and “one feature”, machine it — and revisit in a year. If they are “5,000 a year for five years”, “frozen”, “POM” and “general”, tool it now.
We do both, so the recommendation is not steered by which one we would rather sell. Send the drawing and the volume forecast and we will quote the two routes side by side, including the tool amortised over your first-year quantity.
Related reading
- CNC machining services — tolerances, materials and lead times for machined parts
- Injection moulding — tooling, DFM review and production volumes
- Designing plastic parts for CNC machining — the features that are cheap to machine and expensive to mould
- How to send a drawing for a quote — what to include so both quotes come back usable