CNC machining cost in China: what you actually pay
Machine-hour rates, material, complexity multipliers, finishing, inspection and the hidden costs buyers forget, with DFM moves that lower the price.
Cost · CNC Machining CNC machining cost in China: what you actually pay The price of a machined part is not one number — it is the sum of machine time, setup, material, complexity, finish and inspection, each of which you can influence before you ask for a quote. This guide breaks the cost down, shows the typical brackets, and explains the DFM moves that genuinely lower the number. Request a quoteCost drivers Why the same part gets three different quotes For a sourcing engineer. A CNC quote is a stack of independent costs. Two shops quoting the same drawing can land far apart because one optimizes machine time and the other does not — and because the drawing itself carries hidden cost the buyer did not see. The way to control price is to understand each layer and remove the cost you do not need. The full machining picture. The six cost lines, line by line Every quote, however it is presented, reduces to six independent lines. You control four of them through the drawing and the order; the other two are set by the shop and the part itself. Material Stock cost plus scrap. A part cut from a near-net bar wastes less than one hogged out of a solid block, and standard grades cost less than exotics. This line is set the moment you choose the alloy and the blank. Machine time Spindle hours plus the non-cutting time around them — tool changes, rapid moves, probing. It is the line a good shop optimizes hardest, and the one your geometry either inflates or deflates. Setup and programming A fixed cost per job that amortizes across the batch. One setup on a thousand parts is noise; three setups on three parts is the whole bill. Finishing Anodizing, plating, blasting and polishing each carry their own handling and minimums, added after the part is machined. They are priced per part, so they never amortize away the way setup does. Inspection Scales with requirement. A go/no-go check is nearly free; a full FAI with a CMM report and material certificates is a line item, and a regulated part cannot skip it. Quantity The denominator, not a line of its own. Every fixed cost above divides by it, which is why the same part quoted at ten pieces and ten thousand pieces can differ by an order of magnitude. The machine-hour rate This is the largest single line for most parts. A shop bills the time the spindle is running plus the time it is not — programming, fixturing, setup. In China, 3-axis machining-center rates commonly sit in an indicative bracket of roughly $8–$20 per machine-hour, 4-axis around $15–$30, and 5-axis $30–$60, depending on region, equipment and overhead. The rate itself is lower than in the US or EU mainly because labor and facility overhead cost less — not because the machines are cheaper. A part that takes two hours on a 5-axis center therefore costs more per hour but often less total time than the same part run across three setups on a 3-axis. When 5-axis actually saves money. 3-, 4- and 5-axis: what the extra axes cost The hourly rate climbs with the number of axes, but total hours can fall faster than the rate climbs. That is the whole tradeoff, and the break-even is geometry-specific. 3-axis is the cheapest per hour. It machines one face per setup, so a part with features on three faces means three setups and three datum chains. 4-axis adds a rotary, so four sides of a prismatic part machine in one or two clamps. The rate is higher, the re-fixturing is lower. 5-axis reaches five faces in a single clamp and holds cross-face true position to the machine's native band instead of the sum of several setups. The rate is highest, but setup count and stack-up collapse. The decision rule is simple: multiply your setup count by the cost of a setup, and compare it to the premium of the higher-axis machine. When setups dominate, the expensive machine is the cheap machine. On our floor that is not theoretical — 50+ 3-axis, 20+ 4-axis and 20+ 5-axis mills sit side by side, so a part gets routed to whichever machine is cheapest for its geometry, not to whatever is free. The full 5-axis tradeoff. Material cost You pay for the stock the part is cut from, not just the weight that remains. A small bracket machined from a large bar carries the cost of the whole bar plus the scrap. Buying standard stock sizes the part roughly fits, or picking a near-net shape, trims this line. Exotic alloys (titanium, Inconel, some specialty stainless) carry a material premium that dwarfs the machining cost — which is why material choice is the first cost lever, before any cutting strategy. How material choice drives cost. The complexity multipliers Setups. Every time the part is reclamped or repositioned, you pay setup and risk tolerance stack-up. Designing for one or two setups is the biggest unofficial cost control. Deep pockets and thin walls. They force slow, conservative toolpaths. A feature that doubles machining time is a feature that doubles cost. Tolerance. Tight tolerances mean slower feeds, more measurement, sometimes a finish-grind. Loosening a tolerance from ±0.01 mm to ±0.05 mm can cut cost sharply with no functional loss. Special features. Threads, undercuts, asymmetric geometry and hard-to-reach areas each add tooling or setup. The tolerance-vs-cost curve Tolerance does not add cost in a smooth line; it adds it in steps, and each step lands when you cross the natural capability band of a process. Loose (±0.1 mm and up). Standard feeds, standard tools, no extra measurement. This is the floor price. Standard (±0.05 mm). Normal machining practice. Most drawings should sit here. Tight (±0.01 mm). Slower feeds, more probing, possibly a finish pass and a measurement loop. Cost steps up. Very tight (±0.005 mm and below). At the edge of a mill's capability; often needs a secondary grind to ±0.002 mm, plus CMM verification. Cost steps again, sharply. The expensive mistakes sit at both ends of this curve: tolerances drawn too tight for the function, and tolerances drawn so loose the part fails and gets reworked. The leverage move is to keep every dimension at the loosest value the function actually needs — and to remember that a ±0.005 mm callout on a cosmetic face is buying measurement you cannot see. Mills here hold ±0.005 mm as a standard band and grinding reaches ±0.002 mm; the capability exists, but you should only pay for it where the drawing demands it. GD&T, and when plus-minus is enough. Finishing and inspection For a quality lead. Surface finish is a separate cost line: bead blast and brushing are cheap; anodizing and plating add process and handling; polishing and tight cosmetic standards add labor. Inspection scales with requirement — a basic check is nearly free, a full FAI with CMM report and material certificates costs real time but is what regulated products need. The finish options and their cost. How quantity changes the price Volume is the strongest single lever on unit cost, but it works through specific mechanisms, and each has a limit. Setup amortization. A fixed setup divided by more parts. The first part carries the whole setup; part one thousand carries almost none of it. Material buying. Larger runs justify buying stock at quantity and nesting blanks to cut scrap. Process tuning. A run long enough to dial in feeds, speeds and fixturing spends less time per part than a one-off. The floor. Beyond a certain quantity, machining gives way to casting, forging or stamping — the CNC unit cost stops falling and the process itself should change. The counterpoint is risk. High volume locks the design and the tooling; a change after one thousand parts costs more than a change after one. That is why a shop with MOQ of 1 piece matters: you can qualify the process on real parts at prototype quantities, then scale only when the design is proven. Prototype-to-volume economics. The costs buyers forget Design iteration. Every revision after the first is a new job. A clean, reviewed model costs less than three rushed ones. Expedite. Jumping the queue costs more — and is often avoidable with lead time planned up front. Shipping and tariffs. The part price is not the landed price. For small, light parts shipping is minor; for heavy, dense parts it is not. Landed cost: from the factory floor to your dock The part price on the quote is not what the part costs on your dock. Between them sit freight, insurance, duties and customs — and each is decided by a choice you make before the parts ship. Incoterms set the boundary The Incoterm in the quote defines where the seller's cost ends and yours begins. EXW hands you everything from the factory gate; FOB a Chinese port puts freight and import on you but keeps the export leg with the seller; DDP prices the part to your door, duties included. For most hardware teams FOB is the transparent middle: you control the ocean leg and the import, and you see every line. DDP is convenient, but the duty and compliance decisions sit inside someone else's quote — so compare DDP against your own FOB-plus-duty estimate, not against thin air. Classification decides the duty Your tariff is set by the Harmonized System code, not by the words "CNC part." The six-digit HS is the global base; the US import code is the ten-digit HTS. Classification follows the part's function and material, not how it was made: a steel structural bracket generally lands in Chapter 73, an aluminum one in Chapter 76, and a transmission or machine part in Chapter 84. Get the code wrong and the duty can change by several times — ask the supplier for the exact HTS they propose, then verify it against the current tariff schedule before you commit. Two surcharges sit on top for US imports Beyond the base duty rate, two country-based charges apply to many machined parts. Section 301 adds a China-specific surcharge, commonly in a 7.5%–25% band depending on the list the code falls in. Section 232 adds 25% on steel and 10% on aluminum — and it applies to the material regardless of origin, so a steel or aluminum part carries it even when 301 does not. The two stack: base rate plus 301 (if the code is listed) plus 232 (if steel or aluminum). No free-trade shortcut to the US There is no US–China free-trade agreement, so there is no preferential certificate of origin that lowers the US duty on Chinese parts. Any claim that a certificate will remove the tariff should be treated as a red flag. What does reduce the landed number is honest classification and correct valuation — not a certificate. The DDP trap to avoid A forwarder offering a "tax-included to door" rate may be under-declaring the value to cut the duty. The saving is theirs; the liability is yours, because the US importer of record answers to customs regardless of who filled in the invoice. If you use DDP, require the actual duty payment record. Otherwise, hold the import yourself under FOB and file an honest entry. Freight: air versus ocean Air is priced by weight and speed; ocean by volume and patience. For small, light, high-value parts, air is a rounding error next to the part cost and buys you weeks. For heavy, dense parts, ocean dominates and air can exceed the machining cost itself. The rule of thumb: quote both, and let the lead time, not habit, pick the mode. How to lower the real number Relax tolerance where function allows. The single highest-leverage move. Design for one or two setups. Consolidate features to accessible faces. Pick standard stock. Avoid custom bar sizes and exotic alloys unless required. Batch. One setup across many parts amortizes setup cost per part. Consolidate parts. Two parts that become one remove a whole assembly cost. The DFM moves that pay off. The DFM rules that pay For a manufacturing engineer. Most of the cost of a machined part is decided before the first chip falls — it is baked into the model. These are the geometry decisions that move the price, in rough order of leverage. Fewer setups. Put features on faces a single clamp can reach. Every saved setup removes a datum chain and a cost line. Looser non-critical tolerances. The single highest-leverage edit in most drawings. Standard stock. Design to a size the mill already carries; near-net beats a solid block. Accessible tools. Avoid deep pockets, sharp internal corners and thin walls that force a tiny end mill to creep. Part consolidation. Two parts that become one remove an assembly cost and a tolerance stack, not just a machining line. Uniform sections. Even wall thickness and open geometry cut vibration, scrapped parts and rework. None of these require a cheaper shop. They require a drawing that was looked at before it was quoted. The complete DFM guide. Cost-reduction levers, ranked by leverage When a number needs to come down, buyers reach for the hourly rate first. It is usually the weakest lever. The strong ones, in order: Geometry. Setups, tolerance and stock — decided in CAD, before any negotiation. Volume. Batch parts to amortize setup, and consolidate SKUs into one drawing where possible. Material. Standard grades and near-net blanks, before any exotic alloy is justified. Process routing. Let the shop choose 3-, 4- or 5-axis and the finishing path from the geometry, not from a habit. Rate. The hourly rate negotiates last and moves least. If it is the only lever you pulled, the part was already expensive. China vs the US or EU: the real delta The saving is in labor and overhead, not material — metal costs about the same anywhere. For medium and high complexity, China machining is typically a meaningful fraction of US/EU pricing. The exception is very low volume of simple parts, where international shipping and a long logistics tail can erase the gap. For most hardware teams, the question is not “China or not” but “which Chinese partner controls the cost honestly.” Why Dongguan keeps cost down. A buyer's estimating checklist Send these eight things and the quote that comes back will be a real number, not a bracket. Miss one and the price moves after the fact. The STEP or IGES model plus a PDF drawing — not a screenshot. Material with the grade, not just "aluminum." Quantity now, and the volume you expect if the first run works. Tolerances only where they matter, with GD&T if you use it. Surface finish and any coating, anodizing or plating. Inspection requirement — visual, dimensional, or full FAI with CMM and certs. Target landed terms — the Incoterm and destination. Lead time you actually need, versus the one you would like. One more thing to ask of the shop, not the drawing: the itemized quote. A single number tells you nothing; a line for material, machine time, setup, finishing and inspection tells you exactly where the money goes — and exactly which line to attack. Send it here and we will itemize it. How a shop keeps the number honest A low hourly rate with loose process control is not a saving — it is a rework bill waiting to arrive. The honest signal is a shop that quotes a number it can hold and then proves it. On our Dongguan floor (Hengli Town, 6,800 m², operating since 2006 (Zhuohang)) that proof is statistical: we target Cpk ≥1.67 on critical characteristics and ≥1.33 on general ones. That is what "we can hold it" looks like in numbers, and what protects the part price from becoming a scrap-and-rework cost. How we prove it. On the floor the checks are concrete rather than aspirational. Every BOM line is cross-checked against the drawing and the purchase spec before release, so a 6061 callout cannot silently become 5052. Critical stock is bought through a cross-checked second source with a matching mill test certificate, and each incoming lot is confirmed by XRF / PMI before it reaches a machine. For aerospace programs we run FAI per AS9102 so the first article is dimensionally signed off before the run begins. Indicative planning brackets ElementTypical China bracket (planning only) 3-axis machine-hour~$8–$20 5-axis machine-hour~$30–$60 Setup / programming~$20–$80 per setup Anodize (type II/III)~$0.5–$2 per part by size FAI + CMM report~$50–$200 per first article These are indicative planning brackets across Chinese job shops in 2025–2026, not a quote. Actual price depends on geometry, volume, material and finish — send the drawing for a real number. Frequently asked questions How much does CNC machining cost per hour in China?Indicative brackets run roughly $8–$20 per machine-hour for 3-axis, $15–$30 for 4-axis and $30–$60 for 5-axis, depending on region, equipment and overhead. The rate is lower than in the US or EU mainly because labor and facility cost less, not because machines differ. Why is my CNC quote higher than I expected?Usually it is setup count, tight tolerance, deep pockets or thin walls forcing slow toolpaths, or an exotic alloy whose material cost dwarfs the machining. Most of these are visible in the drawing before you quote — which is where DFM lowers the price. Is CNC machining in China cheaper than the US?For medium and high complexity, typically yes — the saving is in labor and overhead, not material. The exception is very low volumes of simple parts, where shipping and logistics can erase the gap. How do I reduce my CNC machining cost?Relax tolerance where function allows, design for one or two setups, use standard stock, batch parts to amortize setup, and consolidate assemblies into fewer parts. These moves beat negotiating the hourly rate. Does tighter tolerance cost more?Yes. Tight tolerances force slower feeds, more measurement and sometimes a finish-grind pass. Loosening a non-critical tolerance from ±0.01 mm to ±0.05 mm often cuts cost with no functional penalty. Can you quote from a drawing only?Yes — send the STEP/IGES model and any tolerance or finish notes to [email protected] and we will return a priced quote with the cost drivers called out, so you can see what is driving the number. What information do you need for an accurate quote?The STEP or IGES model, a PDF drawing, material and grade, quantity, tolerances, surface finish, inspection requirement, Incoterm and lead time. With those eight items the quote is a real number; without them any number is a guess. What is my landed cost, and how do I compute it?Part price plus freight, insurance and duty. The duty follows the HTS classification — material and function, not how the part was made — plus Section 301 for China and Section 232 for steel and aluminum. The Incoterm defines where each cost starts and stops. What is your minimum order quantity?One piece. We quote from a single prototype through to mass production, so you can qualify the process on real parts before committing to volume. What are your lead times?Three tiers: roughly 3 days for simple prototypes, 7 days standard, and about 30 days for complex multi-process parts with finishing — depending on the 10–20 steps involved. How do you prove the tolerance you quote?With measurement, not adjectives. We run CMM first-article reports and SPC, with a Cpk target of ≥1.67 for critical characteristics and ≥1.33 for general ones, and we hold ±0.005 mm on milling and ±0.002 mm on grinding. Can you help me lower the cost of an existing part?Yes — send the current drawing and quote, and we will flag the DFM changes that reduce the number: setups, tolerance, stock and finishing are the usual levers. Sources & coverage note: Machine-hour brackets are indicative ranges from Dongguan shop-floor quotes and published service comparisons, not a single audited dataset; treat them as planning figures, not quotes. Quality-system claims reference ISO 9001:2015 (certificate 116024/A/0001/UK/En, URS, valid to 2027). Want a real number, not a bracket?Send the drawing to [email protected] — we will quote with the cost drivers itemized and flag the DFM changes that lower the price.Request a quote Related articlesThe Complete Guide to CNC Machining in ChinaThe Complete Guide to DFMDie Casting vs CNC MachiningThe Complete Guide to Material Selection