CNC machining wins below roughly 500 parts per year because it needs no tooling and holds ±0.025 mm; aluminum die casting wins above roughly 2,000 parts per year because piece price drops 4 to 6 times once a $8,000-45,000 die is paid for. For a 400 g housing, machining costs $38 per part with no tooling while casting costs $6.50 per part plus $9,000 of tooling, so the two routes cost the same at about 290 pieces and casting saves 71% at 2,000 pieces. The decision is settled by material before cost in one case: 6061 and 7075 cannot be die cast, and die casting alloys such as A380 and ADC12 cannot be heat treated to T6.
A customer sent us a 400 g (14 oz) aluminum housing with 2,000 pieces per year of demand and a note attached: quote it as a machined part, we do not want to pay for tooling. We quoted it both ways. Machined, it came out at $38 per part with no tooling. Cast, it came out at $6.50 per part plus a $9,000 die. Across the first year, tooling included, the cast route cost $22,000 against $76,000.
That is not the answer every time. Below a few hundred parts a year, or with 6061 on the drawing, or with a design still moving weekly, machining is the correct choice and we say so. This article lays out how the two processes actually differ so the decision can be made on numbers rather than habit.
The short answer
| Situation | Choose |
|---|---|
| Under 500 parts per year | CNC machining |
| 500 to 2,000 parts per year | Run the cost model with your own numbers |
| Over 2,000 parts per year | Die casting, unless material or tolerance forbids it |
| Drawing specifies 6061, 7075 or 5052 | CNC machining, or change the material |
| Needs T6 heat treatment or structural welding | CNC machining |
| Walls under 2.5 mm, heavy ribbing, hollow sections | Die casting |
| Parts needed in under two weeks | CNC machining |
| Many features at ±0.025 mm | CNC machining, or cast then machine |
How each process actually works
Die casting
Aluminum is melted and held at 660 to 700 °C, a measured dose is ladled into a cold chamber shot sleeve, and a hydraulic plunger drives it through a gate into a hardened steel die at 40 to 100 MPa (6,000 to 15,000 psi). The cavity fills in 20 to 100 milliseconds. Pressure is held while the part solidifies, the die opens, ejector pins push the casting out, and the cycle repeats 30 to 90 seconds later. The full stage-by-stage sequence is covered in how aluminum die casting works.
The shape comes from steel that was cut once. Every part after the first is a replay.
CNC machining
A solid block of wrought aluminum is clamped in a 3-axis or 5-axis mill and material is removed until the part remains. Nothing about the part shape is pre-committed; the geometry lives in a program that can be edited in an afternoon. The cost of the part is dominated by how long the spindle runs and how many times the part has to be re-fixtured.
The shape comes from time, and you pay for that time on every single part.
Tooling: the fundamental difference
Everything else in this comparison follows from one asymmetry.
| Die casting | CNC machining | |
|---|---|---|
| Hard tooling cost | $8,000 to $45,000 | None |
| Tooling lead time | 25 to 40 days | None |
| Soft fixturing | Included in tooling | $0 to $1,200 |
| Cost of a geometry change | $800 to $4,000, or a new die | Reprogram, typically under $500 |
| Tool life | 80,000 to 150,000 shots | Not applicable |
A die casting die is a capital commitment made before the first good part exists. A machining program is an operating expense repeated on every part. That single difference generates the volume break-even, the lead-time gap and the design-change penalty all at once. Tooling cost drivers are broken down in more detail on our die casting tooling page.
Piece price and the cost curve
A worked example: 400 g aluminum housing
Take a ribbed enclosure, 240 × 160 × 55 mm, 400 g finished, in A380, with four tapped M5 holes and one flat sealing face at 0.1 mm flatness.
Machined from a 6061-T6 plate blank, the price is built from time:
| Cost element | Value |
|---|---|
| Billet blank, 2.1 kg gross | $6.80 |
| Roughing and finishing, 38 min spindle time | $22.60 |
| Second operation and fixturing | $4.90 |
| Deburr, inspection, packaging | $3.70 |
| Machined piece price | $38.00 |
| Tooling | $0 |
Note the material line. A 400 g machined part starts as a 2.1 kg blank, so 81% of the aluminum bought becomes chips. Chips sell back at a fraction of billet price.
Cast in A380, the price is built from mass and cycle:
| Cost element | Value |
|---|---|
| Metal, 0.42 kg net of internal runner recycling | $1.09 |
| Casting conversion, 55 s cycle, single cavity, 400 ton machine | $0.89 |
| Trim, deburr, shot blast | $0.55 |
| Machining the sealing face and tapping 4 holes | $2.40 |
| Inspection, packaging, 4% scrap allowance | $0.75 |
| Overhead and margin | $0.82 |
| Cast piece price | $6.50 |
| Tooling, single-cavity die | $9,000 |
Two things in that breakdown surprise people. Metal is a larger cast input than machine time, because runners and overflows go straight back into the melt and only the part weight plus melt loss is actually consumed. And the secondary machining after casting line at $2.40 is nearly three times the casting conversion cost, which is why removing a machined feature saves far more than shaving cycle time.
Total cost at volume
| Annual volume | CNC total | CNC per part | Cast total incl. tooling | Cast per part | Cheaper route |
|---|---|---|---|---|---|
| 100 | $3,800 | $38.00 | $9,650 | $96.50 | CNC by 60% |
| 500 | $19,000 | $38.00 | $12,250 | $24.50 | Cast by 36% |
| 2,000 | $76,000 | $38.00 | $22,000 | $11.00 | Cast by 71% |
| 10,000 | $380,000 | $38.00 | $74,000 | $7.40 | Cast by 81% |
| 50,000 | $1,900,000 | $38.00 | $334,000 | $6.68 | Cast by 82% |
The machined column is flat because machining cost is cycle time, and cycle time does not care how many parts came before. In practice a machining quote drops 10 to 15% at high volume through program optimisation and setup amortisation, never by the factor that tooling amortisation delivers.
Why the arithmetic break-even is not the decision threshold
The pure crossover is easy: $9,000 ÷ ($38.00 − $6.50) = 286 pieces. Yet we tell customers the practical threshold is 1,000 to 2,000 pieces per year. The gap is real, and here is why.
- At the crossover, tooling is still $31.50 per part. You have committed capital, frozen the geometry and waited 25 to 40 days for exactly zero savings. Savings only become meaningful once tooling burden drops under about $4 to $5 per part, which happens somewhere between 2,000 and 3,000 pieces.
- The crossover moves with machining complexity, not with casting. The same die at $9,000 against a simpler part that machines for $18 gives a crossover of 783 pieces. Against a 5-axis part that machines for $75 it gives 132 pieces. There is no universal break-even number, only a break-even for your part.
- Design change is asymmetric. Steel added to a die is cheap; steel removed is a weld repair or a new insert. One dimensional change after tooling can cost more than the entire first year of machined parts at 200 pieces.
- Lead time has a cost you may not have priced. Tooling at 25 to 40 days plus production at 15 to 25 days means 40 to 65 days to first production parts, against 5 to 10 days for machined parts.
- Low volumes carry setup overhead. A die casting cell takes two to four hours to set and heat-stabilise, so short runs pay a disproportionate share. We work from 500 parts per year up to 500,000 and above; below that, machining is simply the better tool.
Tolerance and surface finish
| Attribute | Die cast, as-cast | Die cast then machined | CNC from billet |
|---|---|---|---|
| Linear tolerance, first 25 mm | ±0.1 mm (±0.004 in) | ±0.02 mm (±0.0008 in) | ±0.025 mm (±0.001 in) |
| Added tolerance across parting line | ±0.1 to ±0.15 mm | Removed if machined | Not applicable |
| Flatness per 100 mm | 0.2 to 0.4 mm | 0.02 to 0.05 mm | 0.02 to 0.05 mm |
| Surface roughness | Ra 1.6 to 3.2 µm | Ra 0.8 µm | Ra 0.8 to 1.6 µm |
| Hole position | ±0.15 mm | ±0.03 mm | ±0.03 mm |
| Internal corner radius | 1 mm minimum | Cutter radius | Cutter radius |
A cast part is not disqualified by a tight tolerance. It is disqualified by many tight tolerances on faces that cannot be reached in one fixture, because each additional machining setup pushes the cast route toward the machined price. One bearing bore and one sealing face on a casting is normal practice. Fourteen features at ±0.03 mm across five faces is a machined part wearing a casting costume.
One surface finish warning. As-cast surfaces at Ra 1.6 to 3.2 µm are good, but they carry draft, parting line witness and ejector pin marks, and cast Al-Si alloys anodize to a mottled grey rather than the clean tones 6061 gives. Decorative anodizing is a machining requirement. Powder coat, wet paint and chromate on castings all behave well.
Material: you cannot die cast 6061
This is the point that ends more arguments than cost does, and it is regularly missed.
Die casting alloys need 7.5 to 12% silicon. Silicon is what gives the melt enough fluidity to fill a 1.5 mm wall in 30 milliseconds and what suppresses hot tearing as the casting shrinks against rigid steel. Wrought alloys are built on completely different chemistry: 6061 carries roughly 0.6% silicon with magnesium and chromium, and 7075 is a zinc-magnesium-copper alloy with silicon held below 0.4%. Try to die cast either one and you get short fills, hot tears at every internal corner and severe die soldering.
So the practical list is short. Cast: A380, ADC12, A360, A413. Machined: 6061, 7075, 5052, 2024, and effectively any wrought grade. The aluminum-silicon alloy family page covers why the chemistry splits this way, and A380 vs ADC12 works through the two grades most programs actually choose between.
If your drawing says 6061 because it was copied from a machined predecessor, that is worth revisiting. If it says 6061 because the part is anodized for appearance, welded into an assembly or heat treated to T6, the drawing is right and the part should be machined.
Mechanical properties
| Property | A380 die cast | ADC12 die cast | 6061-T6 billet | 7075-T6 billet |
|---|---|---|---|---|
| Ultimate tensile strength | 324 MPa (47 ksi) | 310 MPa (45 ksi) | 310 MPa (45 ksi) | 572 MPa (83 ksi) |
| Yield strength (0.2%) | 159 MPa (23 ksi) | 150 MPa (22 ksi) | 276 MPa (40 ksi) | 503 MPa (73 ksi) |
| Elongation in 50 mm | 3.5% | 3.5% | 12% | 11% |
| Brinell hardness | 80 HB | 75 HB | 95 HB | 150 HB |
| T6 heat treatable | No | No | Yes | Yes |
| Structural weldability | Poor | Poor | Good | Poor |
Read the yield row rather than the tensile row. A380 matches 6061-T6 on ultimate tensile strength, which makes casting look better than it is, but it yields at 159 MPa against 276 MPa and stretches 3.5% against 12%. Castings are less forgiving of overload and less tolerant of a stress concentration, and their entrapped gas porosity derates fatigue life further.
Castings are not disqualified by that. They win on stiffness per dollar, because you can put a 1.5 mm rib grid inside a die casting for nothing while the same grid machined from billet is hours of spindle time. Most housings, brackets and enclosures are stiffness-driven and deflection-limited, not yield-limited, and that is exactly the population where casting dominates.
Geometry: what each process makes easy
| Geometry | Die casting | CNC machining |
|---|---|---|
| Uniform 1.5-3 mm walls | Native | Chatter and distortion |
| Integrated rib grids | Free | Hours of spindle time |
| Hollow or organically curved volumes | Native | Often unreachable |
| Cast-in bosses and mounting features | Free | Each one is cycle time |
| True flat and square datum surfaces | Needs machining | Native |
| Sharp internal corners | Impossible, 1 mm fillet minimum | Cutter radius only |
| Features on five or six faces | Slides at $2,000-5,000 each | Extra setups |
| Deep narrow pockets | Core pin limits apply | Native |
| Undercuts | Slide required | Native |
| Thick in one place, thin in another | Shrinkage porosity | No penalty |
| High material removal ratio | Not applicable | 80% of the billet becomes chips |
| Four fabricated parts consolidated into one | Native | Rarely economic |
Lead time and design change flexibility
| Milestone | Die casting | CNC machining |
|---|---|---|
| Quote after receiving CAD | 24 to 48 hours | 24 to 48 hours |
| First part in hand | 30 to 45 days | 5 to 10 days |
| First production shipment | 40 to 65 days | 10 to 15 days |
| Repeat order lead time | 15 to 25 days | 10 to 20 days |
| Cost of one dimensional change | $800 to $4,000 | Under $500 |
| Cost of adding a feature that needs a slide | $2,000 to $5,000 | Reprogram only |
The design-change column is the real risk on a casting program. Add material to a part and the die has to lose steel, which is straightforward. Remove material from a part and the die has to gain steel, which means welding and re-cutting a hardened H13 insert, or making a new one. That asymmetry is why the honest advice is to machine prototypes, validate, freeze, and only then release tooling.
Where castings hit hard limits
Four limits are not process-parameter problems and cannot be tuned away.
- Porosity. Every high pressure die casting contains some entrapped gas. It is manageable to acceptable levels and often invisible, but it is present. Machining a pressure boundary into a casting can expose a pore and create a leak path.
- No T6. Solution treating a die casting at 500 to 540 °C expands the entrapped gas and raises blisters. Die castings ship as-cast, sometimes with a T5 stress relief, never T6.
- Weldability. The same trapped gas boils into the weld pool and produces porous, unreliable welds. Design for mechanical fastening or adhesive bonding instead.
- Thick-section soundness. Sections above about 6 mm develop shrinkage porosity at their thermal centre. Where a machined part just gets thicker, a casting has to be redesigned with ribs.
The failure modes and the fixes are catalogued in common aluminum die casting defects.
The hybrid route, which is what most programs actually build
The real answer for most volume programs is not one process. It is a near-net casting that carries the shape, followed by machining only the features that need precision. Our example housing spends $4.10 on casting, trimming and blasting, then $2.40 on the sealing face and four tapped holes, and lands at $6.50 while holding ±0.02 mm exactly where the drawing demands it and ±0.1 mm everywhere else.
Design for that hybrid deliberately: leave 0.5 to 1.0 mm of machining allowance on the faces you will cut, give the fixture cast datum pads to locate on, and cast every non-critical feature to size. The rules are in the aluminum die casting design guide.
Deciding in five questions
- Annual volume? Under 500, machine it. Over 2,000, cast it. In between, build the cost model.
- Does the material spec allow A380, ADC12, A360 or A413? If it demands 6061, 7075 or T6, machine it.
- How many features need better than ±0.05 mm, and how many setups do they take? Many features across many faces pushes toward machining.
- Is the design frozen? If it is still moving, machine until it stops.
- When do you need parts? Under three weeks means machining, at least for the first batch.
We run both processes, so there is no advantage in steering a 200-piece program into a die. Ask for a dual quote on a part in this grey zone and you will get the machined price, the cast price, the tooling cost and the crossover volume for your geometry rather than a generic rule.
Our aluminum die casting services page covers the casting side, from 20 g to 12 kg on 160 to 1,250 ton machines. Send 2D and 3D files through contact for both prices and a written DFM assessment inside 48 hours.
