Injection Molding vs CNC Machining: When Does Each Win?

Last updated: 2026-09-01 · Reading time: about 10 min

CNC machining carves each part individually from stock; injection molding builds a tool once and then repeats the part at a low marginal cost per piece. CNC wins small: prototypes, bridge quantities, metals and designs that are still moving.
Injection molding wins scale: once the volume justifies the tool and the design is frozen, the per-part cost drops far below anything a cutting process can follow. The professional question is not which process is better — it is where your program’s volume and design stability put the crossover, and whether your plan respects it. See also related article.

Key facts

<>

  • CNC carries its cost in every part — machine time, material waste and handling repeat with each piece — so its per-part cost is roughly flat across volume.
  • Injection molding carries its cost in the tool: expensive to build, quick to repeat, which makes the per-part cost fall steeply as volume grows.
  • The crossover is a real calculation, not a rule of thumb: part size, cycle time, material and machine rates move it by multiples between programs.
  • Design freedom differs: CNC reaches internal geometries molding cannot, and molding delivers texture, undercuts and repeatability that machining would price out.
  • Many mature programs run both in sequence — CNC for prototypes and bridge builds, molding for production — with the design freeze as the hinge between them.

    • Gate vestige — The visible mark left on the part surface where the gate was trimmed. Edge gate: 0.5–2.0mm vestige. Pin-point gate: 0.2–0.5mm. Valve gate: 0.05–0.1mm (flush). Vestige size affects cosmetic appearance and may require secondary trimming. See also related guide.

    What is each process actually for?

    The two processes answer different economic questions, and the confusion between them is usually a confusion of questions. CNC machining answers: “How do I get this part, in this material, now, without committing to anything?” It needs no tool, accepts design changes between batches, and works the full menu of plastics and metals — the natural home of prototypes, pilot builds, spare parts and low-volume production where the schedule matters more than the piece cost.
    Injection molding answers: “How do I get this exact part, repeatedly, at the lowest sustainable piece cost?” It invests in a tool once and then repeats faithfully for the life of the program — the natural home of production volumes where the same geometry ships by the thousands and the cost per part carries the business case.
    Neither process wins on cost in general; each wins inside its own territory, and the territory boundary is your volume, your design stability and your material. Programs that name that boundary out loud — before quoting, before prototyping, before anything is cut — spend their first dollars on the process that answers the question they actually have.

    Where do the two processes actually differ?

    Dimension CNC machining Injection molding
    Cost structure Cost repeats per part; volume barely moves it. Cost sits in the tool; per-part falls with volume.
    Design changes Re-cut the program; a change costs one edit between batches. Change the steel; changes are a modification project.
    Materials Plastics and metals, stock shapes. Thermoplastics (and some thermosets/LSR), pellet form.
    Geometry freedom Internal features, deep pockets, single parts. Undercuts, texture, living hinges, family parts — with draft rules.
    Lead-time shape Days to weeks, repeatable. Weeks for the tool, then seconds per part.
    Repeatability Good; batch-to-batch setup matters. Excellent once the process is locked and evidenced.

    Two rows in that table quietly decide most programs.
    The design-change row: a part that is still moving belongs in CNC territory regardless of volume, because every molding iteration is steel work. The lead-time row: molding’s “weeks” is the tool build, and it happens exactly once — which is why mature programs count molding lead time in parts-per-day after launch, not in weeks before it.
    Both rows cost nothing to verify early and a great deal to discover late — which is why this table belongs in the first DFM meeting, not after the first crisis.

    How do you find the crossover volume for your part?

    The crossover is the volume where molding’s total cost — tooling plus running cost — crosses below CNC’s flat per-part line, and for any real part it is a calculation, not a legend. Three numbers drive it: the tooling investment (which scales with size, cavitation and complexity), the per-part running cost (cycle time, resin, machine rate), and the CNC per-part price for the same geometry and material.
    Sketch the two lines and the intersection falls where it falls — and it moves enormously between a tiny clip and a meter-wide panel, which is why borrowed crossover numbers are worthless. Run the arithmetic once per program and the crossover stops being an opinion — it becomes a line on a page that procurement can read, and a supplier either honors it or explains it.
    The discipline that keeps the calculation honest is the one the rest of this site preaches: the tooling quote is itemized, the cycle-time estimate comes from wall thickness and cooling geometry, and the running cost lands as cost per good part over the program’s real volume — including the scrap and uptime realities, not the theoretical ones.
    Programs that skip the arithmetic and choose by habit pay for it in one of two directions: molding too early, and every design change becomes steel work; or machining too long, and the piece price quietly eats the program’s margin.

    What does the tool investment actually buy?

    Crossover volume
    The annual quantity at which molding’s total cost per part — tooling amortized across the program plus the running cost — drops below the machined part’s flat price. Below it, machining wins; above it, molding wins by a widening margin.

    It is worth naming what the tool buys, because the price tag reads differently depending on what you think you are purchasing. The tool buys repetition: the same geometry, the same surface, the same dimensions, thousands of times, with variation shrinking rather than accumulating. It buys the surface — texture, gloss and brand faces live in the steel, not in a secondary finishing step.
    It buys geometry that machining prices out: undercuts through side actions, living hinges, thin uniform walls, family parts that ship from one shot. And it buys clock speed — once the process is validated, parts arrive by the dozens per hour, which no cutting path matches. What the tool does not buy is flexibility, and that is the other side of the ledger.
    Every design revision after the steel is cut is a modification project with its own quote, its own schedule and its own validation run. So read the tooling line on a quotation as a purchase, not a cost: you are buying the right to freeze the design, and everything on the part you are not yet sure about should be settled in CNC or in the DFM conversation before it is settled in steel.

    What does the hybrid path look like?

  • Prototype in CNC — prove the design in the production-intent material where possible, and change anything freely while the changes are free.
  • Bridge the launch with CNC — when the market cannot wait for the tool, machine the first shipments and let the build run in parallel.
  • Freeze the design at DFM — the hinge decision; molding begins where the drawing stops moving, and the DFM record is the receipt.
  • Build and validate the tool — T1 through sign-off with the full evidence package, so the crossover arrives on a validated process.
  • Retire the CNC bridge deliberately — define the volume or date where production moves fully to molding, and write it down; bridges that never retire are programs leaking margin.

    • Which decisions are locked before either process starts?

    Whatever the process, three decisions are worth making before the first purchase order. The material decision: prototyping in a substitute resin validates the design but not the material behavior, so the production resin and its datasheet discipline belong at the start. The tolerance decision: CNC and molding hold tolerances differently, and a drawing written around machined precision will meet the shrinkage conversation at DFM — better to ask what the part truly needs first.
    And the evidence decision: which measurements, at which stage, prove the part is right — the same T1-style evidence logic that governs molding applies to machined prototypes, only faster. Programs that settle these three questions early switch processes on purpose; programs that skip them switch by accident, usually twice.

    “CNC buys you the right to keep thinking. Molding buys you the right to stop. Programs fail when they confuse which right they need.”
    — Arthur, Tooling Engineer at aCatMould

    Frequently asked questions

    Is a CNC prototype representative of the molded part?

    For geometry, closely — for material behavior, only if machined in the production resin, and even then the microstructure differs: no weld lines, no flow orientation, different residual stress. Prototype the design in CNC, validate the part in molding; the T1 report is where the real evidence lives.

    Can injection molding be economical at low volumes?

    Sometimes — aluminum prototype tools and single-cavity builds lower the entry, and stable designs with modest tooling can pay out at volumes that surprise people. But it is a calculation, not a habit; run the crossover math for your actual part before assuming either way.

    Why is my supplier pushing molding when CNC works fine today?

    Because today is not the question — the program’s full volume is. If the forecast is real and the design is stable, molding’s per-part line crosses below CNC’s and keeps falling. If either condition is missing, the push deserves scrutiny. Ask for the crossover calculation in writing; a credible partner shows the math.

    Which process holds tighter tolerances?

    Machining holds tighter tolerances on individual features; molding holds excellent repeatability once the tool is validated, within the limits of shrinkage behavior. Compare the drawing against both process realities at DFM — tolerance strategy is process strategy.

    Can CNC parts serve as tooling validation?

    Not directly — but machined parts in the production resin can de-risk assembly fit and cosmetic standards while the tool is built, which shrinks the list of surprises the first T1 has to carry. They answer different questions; run both, and record which question each answered.

    How do I keep the two processes honest in one program?

    One drawing, one evidence standard, and a written crossover plan: which volumes run machined, which run molded, and what triggers the move. The programs that document the handoff retire their bridges cleanly; the ones that improvise run two supply chains by accident.

    Does an aluminum prototype tool change the crossover?

    It moves, it does not erase. An aluminum tool lowers the tooling line, so molding becomes competitive at lower volumes — but aluminum carries its own limits in cycle life, injection pressure and textured surfaces, and it is a bridge decision unless the program’s total volume is genuinely small. Price both tooling options and run the crossover math twice.

    Who should calculate the crossover — us or the supplier?

    Both, and the two answers should be compared. You hold the forecast and the program history; the tooling side holds cycle time, cavitation and machine-rate reality. Comparing the two calculations is where hidden assumptions surface — and the written comparison is the deliverable, from a named tooling engineer, before the deposit.

    Key takeaways

    1. CNC costs repeat per part; molding costs sit in the tool — the crossover is volume arithmetic, and it moves by multiples between parts.
    2. Design stability is the hinge: moving drawings belong to CNC, frozen drawings unlock molding’s economics.
    3. The hybrid path is normal — prototype and bridge in CNC, freeze at DFM, validate the tool, retire the bridge deliberately.
    4. Material, tolerance and evidence decisions come before either process starts; settling them early makes process switches deliberate.
    5. Ask for the crossover calculation in writing — itemized tooling, honest cycle time, cost per good part — and compare partners on the math, not the habit.

    If your program is standing at the CNC-versus-molding fork, send us the 3D model, the volume forecast and the material target before the deposit. You will get the crossover calculation in writing with its assumptions named, the hybrid launch plan if the schedule needs one, and the tooling quote that follows the math — from a named tooling engineer, before any steel is ordered.
    If the arithmetic says stay in CNC for now, the answer will say so in writing; a crossover calculation has no preferred direction, and evidence against molding today is still evidence you can act on.

    Sources
    Plastics Industry Association — industry resources on plastics processing practice
    ISO 20457:2018 — Plastics moulded parts: Tolerances and acceptance conditions, International Organization for Standardization
    MatWeb — material property database for machined and molded resin grades
    ASTM D638 — Standard Test Method for Tensile Properties of Plastics, ASTM International

    Sources: Plastics Technology · Plastics Technology Online · Prospector — Resin Datasheets · MatWeb — Material Property Database