CNC machining cost per part: a real-world estimate for buyers

I have been quoting CNC parts since 1998. I can tell you within a few percent what a part will cost before I open the CAD file, and no, I am not psychic. After 27 years and probably 40,000 RFQs, the patterns are obvious. This guide gives you those patterns so you do not have to learn them the hard way, one bad order at a time.

Let me be direct: most online cost calculators are wrong by 30% to 200%. They are built around average machine rates that no real shop charges, with cycle times that ignore tool changes, fixturing, and the inevitable faffing about. I will show you what we actually see on the floor, and where the calculator leaves the money on the table.

What actually drives the price

Four things, in this order: machine time, setup, material, and inspection. Everything else is noise until you control these.

Machine time is the per-hour cost of the CNC plus the spindle cutting. For a 3-axis vertical machining center in our shop, that runs USD 35 to 60 per hour. A 5-axis simultaneous machine is more like 90 to 150. Swiss lathe work for tiny precision parts can hit 120. These are real numbers, not list prices. List prices are what a sales rep tells you when they are not actually going to do the work.

Setup is where most first-time buyers get hurt. A job shop has to plan the work, fixture the stock, write or modify the program, set tools, and prove out the first piece. That is 1 to 4 hours of skilled labor even on a simple part. For a 5-axis part with a fourth-axis rotary, it is 6 to 10 hours. If you order 10 pieces, you are paying for one setup and dividing by 10. If you order 1,000, the setup per part becomes small. This is why prototype quotes feel brutal and production quotes feel generous. Same part, same shop, 10x different unit price.

Material is the easiest number to verify. Take the part volume, add a machining allowance of 20% to 40% for stock, multiply by the alloy's density, and you have the raw weight. Multiply by the per-kilogram price. For 6061 aluminum that is around USD 4 to 6 per kg, for stainless steel 316L about 8 to 12, and for titanium Ti-6Al-4V 35 to 60. The spread is wide because it depends on bar stock versus plate, and whether you need certified material for medical or aerospace work.

Inspection is the one nobody budgets. A full dimensional report on 10 critical features is roughly 30 to 60 minutes of CMM time plus the report. If you need 100% inspection, the cost multiplies. For medical components with FDA documentation, add another layer. We have had medical jobs where inspection and documentation were 25% of the total cost. The buyer did not see that coming.

Real price ranges I see every week

These are the numbers that come across my desk, not industry marketing claims. Treat them as a sanity check on your own quotes.

A simple CNC machined bracket in 6061 aluminum, 50 mm by 50 mm, three tapped holes, no tight tolerances, in volumes of 100 to 500 pieces, will land at USD 4 to 8 per part. Add a powder coat finish and you are at 5 to 10. Drop to 10 pieces and the same part is 25 to 40 because setup dominates.

A stainless steel housing with tighter tolerances, say plus or minus 0.05 mm on critical features, in 304 or 316L, runs 12 to 25 per part at 200 pieces, and 8 to 15 at 1,000. The drop is not magic. It is just setup getting diluted.

A complex 5-axis part in titanium for aerospace work, with AS9100 paperwork and full traceability, can easily hit 150 to 400 per part at low volume, and 60 to 150 at 5,000 pieces. The material alone is 25% of the cost. The certification is another 15%. The actual machining might only be 30% of what you pay.

For prototyping, the rule I trust is simple. One piece costs 10x what 100 pieces costs, and 100 pieces costs 3x what 1,000 costs. The first prototype is the most expensive thing you will ever make. Get used to it.

The hidden costs buyers forget

Tooling is where the calculator fails you completely. A job shop does not own every tool. End mills, drills, taps, reamers, and special form tools are either in inventory or they get ordered. Standard tools are baked into the machine rate. Special tools are billed separately, and they are not cheap. A custom form tool for a weird radius can be USD 200 to 600. A thread mill for an unusual pitch is 80 to 200.

Surface finish is another surprise. As-machined finish is included in the machine time. Bead blast, tumble, or hand deburr is usually a few dollars per part. Anodizing type II is 3 to 6 per part for small components. Anodizing type III hard coat is 6 to 12. Powder coat is 4 to 8. Electroless nickel, which buyers love because it looks great, can run 15 to 30 per part depending on the part size. If your drawing calls for a finish, expect that to add 20% to 50% to the bare-machined price. Read more about this in our finishing and inspection workflow.

Packaging and documentation are not free. A clean part in a labeled bag with a COC and a basic inspection report is included. A part in custom dunnage, vacuum-sealed, with a full PPAP and IMDS submission, is 5 to 20 per part on top. We have done automotive jobs where the PPAP package cost USD 8,000 to 12,000 before the first production part shipped. That is real money. The buyer who only looked at per-part pricing was shocked.

Certificate of conformance for material is usually free from a good distributor, but for medical or aerospace, the test reports (tensile, hardness, chemistry) cost 100 to 400 per heat. If your part is small and you are using tiny pieces of expensive bar, that cost gets spread over fewer parts and the per-part number jumps.

How to read a CNC quote without getting fooled

Any quote that does not break down machine time, material, setup, and finishing is a black box. Walk away or ask for the breakdown. A shop that refuses to itemize is either hiding margin or does not know their own costs. Both are bad signs.

Look at the lead time versus the price. If a quote is 30% under everyone else and the lead time is half, the shop is desperate. They might still deliver, but you have no negotiating position if something goes wrong. The middle of the pack, on time, with clear communication, is almost always the right choice. We have lost jobs to cheaper quotes and then received the same customer back six months later when the cheaper shop missed the third delivery date.

Watch for substitution. Aluminum 6061-T6 is not the same as 6061-T4 or 6063. Stainless 304 is not 316L. Titanium grade 2 is not grade 5 (Ti-6Al-4V). A 20% cheaper quote often comes from a 20% cheaper material. It might meet print, but only because the print was loose. If your print does not call out the alloy temper, the shop will pick the cheapest one that passes.

Check the tolerance claims. A quote that says plus or minus 0.01 mm on a 100 mm feature at low cost is suspect. Holding that across a production run requires a tight machine, a calibrated CMM, and slow, careful cuts. If the price does not reflect that, the tolerance will not be held on batch 200. See our machining tolerance guide for what is realistic at different price points.

Ask about yield assumption. Most shops quote assuming 95% to 98% yield. If your part has features that historically yield 85%, the real cost is higher than the quote. Ask the shop what they expect. If they cannot answer, they have not made the part before.

Five ways to cut CNC machining cost without breaking the part

One, relax tolerances that do not matter. A plus or minus 0.025 mm callout on a cosmetic feature doubles the machining time. If the part is going inside a housing and never touches a mating surface, that tolerance is costing you money for no reason. We see drawings where half the tolerances are tighter than they need to be. The cost of a 10-minute review with your engineer is usually less than the cost of the first production run.

Two, design for the machine. A pocket with internal radii matching a standard end mill cuts faster than one with sharp corners. Holes at standard sizes use standard drills. Avoid thin walls under 0.5 mm, deep pockets more than 4x the tool diameter, and small features that require small tools. Small tools break, and a broken tool mid-run is a hidden cost in scrap, machine time, and operator panic.

Three, order the right quantity for the break-even. If setup is 200 USD and the per-part run cost is 5 USD, your all-in cost for 10 parts is 250 USD (25 each). For 100 parts it is 700 USD (7 each). For 500 it is 2,700 USD (5.40 each). The unit cost keeps dropping, but the total spend goes up. Order what you actually need, not what gives the best per-part number on paper.

Four, consolidate material grades. If you have five parts and four different alloys, the material cost is high because you are buying small quantities at distributor minimums. Standardize where you can. We have helped customers drop 12% off their annual spend just by changing two non-critical parts from titanium to stainless.

Five, use the right process for the volume. For low volume, metal injection molding makes no sense. For high volume of a complex small part, it can be 70% cheaper than CNC. Powder metallurgy is another option for parts that can tolerate the porosity. Investment casting beats CNC on parts with complex internal geometry once you cross a few hundred pieces. The wrong process is the most expensive decision you can make, even when each quote is correct in isolation.

When a low quote is fine (and when it is not)

A low quote is fine when the shop is buying capacity. We have off-peak weeks. We would rather run a job at 80% of normal margin than have the machine sit idle. The customer gets a fair price, we keep the team paid. Everyone wins.

A low quote is not fine when it is a loss leader. The shop plans to lose money on the first order, then raise the price on order two once you are locked in. We see this all the time. The fix is a written price agreement for the first three orders, or at least a clause that locks raw material surcharges to a public index like LME or a published distributor price. If the shop will not agree to that, they are betting against you.

One last thing I want to leave you with. Cost is not the same as value. A part that costs 2 USD but arrives 8 weeks late kills your production line. A part that costs 4 USD and arrives on time, with paperwork, with a Cpk of 1.67, with a supplier who picks up the phone on a Saturday when you have a line stop, is cheaper in the end. The cheapest quote on a spreadsheet is rarely the cheapest part on the floor.

Spend an hour getting to know the shop before you spend a dollar on the parts.