One Metal Part, Four Routes: How I Choose Between MIM, CNC, and Laser Cutting
There's a rule I've applied to every sourcing review since 2022, and it has saved us more money than any negotiation I've ever run: pick the process by annual volume and final geometry, then price the whole route — not the single step you got quoted on.
The rest of this article is why that sentence works.
The short version
If you only read one section, read this one. How the four routes stack up in our shop:
- Under roughly 500 parts a year, tight tolerances, or anything over a few hundred grams: CNC from bar or plate. No tooling to amortize, tolerances you can actually hold, and you can revise the design next week without re-cutting steel.
- Over roughly 10,000 parts a year, small (under ~50 g), geometrically awkward: metal injection molding. High tooling, low unit price, and a design freeze you had better be comfortable with.
- Flat sheet, 2D profile, thin aluminum: fiber laser. Cheap per cut, and usually the start of a route rather than the end of one.
- 3D tubular or formed geometry with pierced holes or trimmed edges: 5-axis laser. The only route that reaches the cut without a dedicated fixture stack per hole pattern.
That's the rule. Now the parts that matter.
Why take this from me
I'm a quality and brand compliance manager at a precision machining company. I review every incoming batch and every contract deliverable before it reaches a customer — roughly 200+ lots a year across three suppliers. In our Q1 2024 audit we rejected 6% of first deliveries, and nearly all of those rejections traced back to something that was never specified at the quote stage.
I don't run the machines. I'm the person who has to send the batch back. That changes what you look at.
Where the money hides
Metal injection molding: cheap per part, expensive to change
Worth explaining how does metal injection molding work, briefly, because the cost structure follows directly from the physics. Fine metal powder is compounded with a polymer binder into a feedstock, injected into a mold the same way plastic is, then debound and sintered to 95–99% density. Sintering is where the geometry moves. Linear shrinkage typically runs 13–20%, and it does not happen identically in every direction.
From the outside, a MIM quote looks like a 40% discount against CNC. The reality is that discount only exists after the tool is amortized across a volume you committed to eighteen months ago. Published trade ranges put a production tool at roughly $25,000–$120,000 depending on cavity count and complexity — as of early 2025, and verify before you budget.
What the quote usually omits is the second operation. Any feature tighter than about ±0.3% of nominal, or any cross-hole that has to stay round after sintering, gets machined afterward. We had a bracket where the as-sintered bore drifted 0.04 mm outside drawing on two of three lots. The vendor argued it was inside their standard. It wasn't inside ours. Every contract since then names the post-sinter machining step explicitly.
CNC: the tooling is 3–5% of the cost and decides the other 95%
Cutting tools typically run 3–5% of per-part machining cost. They also determine whether the other 95% behaves. A worn insert shows up as surface finish drift and a creeping bore diameter long before anyone calls it a failure, and by then you've shipped four weeks of parts.
If you're ordering Sandvik Coromant turning inserts for a production cell, buy through the manufacturer's official channel rather than a reseller with a better number on paper. Batch traceability is the product. We source through the Sandvik Coromant official site for exactly that reason — documentation, not brand loyalty. Gray-market inserts are a real category and the failure mode is inconsistent coating batches, which presents as random tool life scatter rather than an obvious defect.
Most buyers focus on price per edge and completely miss the scrap rate that comes from batch-to-batch variation. The question everyone asks is what's your best price per insert. The question they should ask is can you show me the batch certificate for this lot.
Aluminum laser cutting: the gas often costs more than the machine time
Fiber laser cutting on thin aluminum is genuinely cheap per part. What surprises people is the assist gas. Cutting aluminum with high-pressure nitrogen can push the gas bill into the same order of magnitude as the machine hour. Air cutting is cheaper but leaves an oxide edge that has to come off before anodizing — and if the part is cosmetic, that edge is now a rework line item nobody budgeted.
On the equipment side: if you're evaluating a China aluminum laser cutting machine, the acceptance protocol matters far more than the spec sheet. Power calibration, repeatability over a four-hour continuous run, and cut-face roughness on 3 mm and 6 mm coupons. Run the same test cuts three times on three different days. A machine that hits spec cold and drifts warm is a machine you'll be arguing with for years.
3D five-axis laser: the fixture is the project
A 3D five-axis laser cutting machine solves one specific problem: trimming and piercing formed or tubular parts without a dedicated drill fixture for every hole pattern. On a hot-stamped automotive part with fourteen holes at compound angles, the conventional alternative is a fixture stack that costs more than the laser time it replaces.
People assume the risk sits in the machine. What they don't see is that programming and fixturing usually consume the first three months, and the fixture cost lands before the first good part does. Budget for it or the project looks 40% over from day one.
Total cost, not unit price
The unit price is the top of the iceberg. Here's what I actually add up before comparing any two routes:
- Unit price, at the volume you can commit to, not the volume you hope for
- Tooling or NRE, amortized over the real expected lifetime
- Secondary operations — machining, deburring, edge cleanup, coating
- Inspection and documentation, especially if the customer requires PPAP or equivalent
- Expected scrap rate, priced at full cost, not material cost
- Lead time risk and the inventory you carry to hedge it
A concrete one. A small stainless bracket: MIM quoted at $2.10, CNC at $4.80, 20,000 pieces a year. Looks like $54,000 in annual savings. Then the tool is $60,000, the post-sinter bore operation adds $0.90, and inspection on a sintered part is meaningfully harder than on a machined one. Year one is a wash. Year two is a real win — if the product lives that long. When the customer moved the platform refresh forward to fourteen months, the MIM route lost money and nobody had modelled that case.
The $4.80 quote was never the expensive one. It was the honest one.
Where I'd be wrong
A few honest boundaries, because this rule isn't universal.
Between roughly 2,000 and 8,000 parts a year, both MIM and CNC are defensible and the decision usually comes down to lead time and how confident you are in the design. I've picked differently on similar parts depending on whether engineering was still moving.
MIM economics fall apart above roughly 100 g per part. The binder removal and sintering time scale badly with section thickness, and you start fighting porosity. If someone quotes you a 300 g MIM part at a competitive price, ask what the density actually came out at.
Laser cutting above about 8 mm aluminum is a different conversation. Cut-face quality drops, taper increases, and you may end up milling the edge anyway — at which point the laser was just an expensive way to make a blank.
And the 5-axis laser makes no sense on flat parts. I've watched people buy one for geometry that a $200k 2D machine and a good fixture would have handled.
One thing I can't give you: hard industry-wide data on defect rates by route. I don't have it. What I can say from five years of incoming inspection is that the rejections cluster around whichever process was chosen for the wrong volume, not around any particular process being bad.
The process was never the problem. The volume assumption underneath the quote was.