Domestic vs Overseas CNC Machining: Real Cost Compared

The question "is it cheaper to machine parts in China or at home?" has no useful answer, because the number that decides it is not the one on the quotation. A part quoted at $18 domestically and $6 overseas is not a 3× saving. By the time it is on your shelf, inspected and available to build with, the gap is frequently 1.4× — and on some part families it inverts entirely.

This article gives you the cost model instead of the sales pitch. We build the full landed-cost stack line by line, show which line items each geography genuinely wins, give you a break-even calculation you can run on your own numbers in twenty minutes, and — the part most suppliers skip — set out plainly the situations where you should not buy from an overseas supplier at all, including from us. If you are still deciding whether to outsource machining rather than where, our piece on the benefits and challenges of outsourcing CNC machining covers that prior question; this one assumes you have already decided to buy rather than make.

The quick answer: what actually decides it

Three variables decide the geography question, and part complexity is not one of them.

1. Annual volume per part number. Every offshore order carries a fixed overhead that does not scale down: freight, customs entry, incoming inspection setup, and the engineering hours spent on drawing clarification and first-article review. Spread over 2,000 parts that overhead disappears into the noise. Spread over 25 parts it can double the effective unit cost. This is why prototype shops and production suppliers are different businesses even when they own the same machines.

2. Design stability. An offshore supply chain has a pipeline in it — typically 4–6 weeks of parts in production, in transit, or in customs at any moment. Every design change either scraps that pipeline or forces you to run out old revision stock. If your drawing is still moving, that pipeline is a liability, and the cost of change is where offshore programmes lose money. Freeze the design first, then move it offshore.

3. Tolerance of delay. Not average lead time — the cost of the tail. Domestic machining at 2 weeks and offshore at 5 weeks is a manageable difference if it is predictable. What breaks programmes is the 5 weeks that becomes 9 because a port was congested or a first article failed. Price that risk honestly; if a line stoppage costs you $40,000 a day, no unit-price saving justifies an unbuffered offshore supply.

Why unit price is the wrong comparison

The unit price on an offshore quotation is genuinely lower, and the reasons are structural rather than mysterious: lower labour cost per machine hour, denser supplier clustering that shortens the material and finishing supply chain, and machine utilisation strategies that make a shop willing to quote thinner margins on repeat work. None of that is controversial.

The error is treating that number as the comparison. On a domestic purchase, the quoted price is close to the whole cost — you add ground freight and receiving, and you are basically done. On an offshore purchase, the quoted price is one line in a stack of eight or nine, and several of the others are large, variable, and easy to forget at the point when you are choosing a supplier.

Here is the full stack. Fill it in for both options before you decide anything.

The nine lines of total landed cost

1. Unit price × annual quantity. The visible number. Insist on it being quoted at your real annual volume, not at a sample quantity, and ask what price break exists at 2× and 5× that volume — the shape of that curve tells you how the supplier is loading fixed costs.
2. Tooling, fixtures and programming. On pure CNC machining this is often modest — soft jaws, a fixture plate, the CAM programme. It becomes decisive the moment a part needs a die or mould. Our CNC machining service covers 3–5 axis milling and turning with no MOQ, so the tooling line stays small; a die-cast or moulded version of the same part carries tooling of 7–20 days lead and a 1,000–3,000 pcs MOQ, which changes the arithmetic completely. Amortise tooling over the parts you will honestly order, not the parts in the optimistic forecast.
3. Freight. Sea freight is cheap per kilo and slow; air freight is fast and priced on chargeable weight, which is the greater of actual weight and volumetric weight. Bulky light parts — sheet metal enclosures, thin-wall housings — get billed on their volume, and this is the single most common place a landed-cost model goes wrong. Get a real quote from a forwarder using your actual carton dimensions before you model anything.
4. Duties, tariffs and customs entry. See the method section below. Never model a rate you read anywhere other than the official schedule on the day you are buying.
5. Incoming inspection. Domestic parts from a known supplier often go to skip-lot or dock-to-stock quickly. A new offshore supplier realistically gets 100% inspection on the first two or three lots, then AQL sampling. Cost this in inspector hours, not in principle — it is real money in the first year and it declines steeply afterwards.
6. Inventory carrying cost. The pipeline plus safety stock is capital sitting still. Take average inventory value and multiply by your internal cost of capital plus storage and obsolescence — most industrial buyers land somewhere in the 18–28% annual range. An offshore programme routinely carries 2–3× the inventory of a domestic one for the same consumption rate. This line item is bigger than most people expect and is almost always omitted.
7. Rework, scrap and containment risk. Not the expected cost — the cost when it happens. A bad lot discovered domestically can be sorted and re-run in a week. The same lot discovered after a sea crossing means you either sort locally at your own labour rate, air-freight replacements, or wait five weeks. Model the cost of one bad lot per year and see whether the programme still works.
8. Travel and audit. A serious offshore relationship involves at least one visit, and often an annual one, or a paid third-party audit in its place. It is a modest number but it is not zero.
9. Your engineering and buying time. The most under-counted line in the stack. Time-zone-offset communication, drawing clarification, first-article review, revision control across a long pipeline. Budget it in hours at loaded rate. On a well-run repeat programme this settles to a few hours a month; in the first quarter of a new relationship it can be several hours a week.

How to handle duties without guessing

Duty rates change — sometimes several times in a year, and sometimes retroactively in effect through trade actions. Any specific percentage in an article like this one is wrong by the time you read it. So here is the method rather than a number, and you should treat the output as an estimate until a licensed customs broker confirms it. We are a manufacturer, not a customs, legal or tax adviser, and nothing here is advice on any of those.

1. Classify the part. Determine the correct commodity code for the finished article — machined parts frequently classify by their function in an end assembly rather than by material or process, and getting this wrong is the most expensive error in the whole exercise. The World Customs Organization maintains the international Harmonized System nomenclature that national schedules are built on.
2. Look the rate up in your own national schedule. US importers use the USITC Harmonized Tariff Schedule; UK importers use the UK Trade Tariff tool; EU importers use TARIC. Check on the day, for your code, for the country of origin.
3. Add every non-duty charge at the border. Import VAT or GST where applicable, merchandise processing and harbour fees, brokerage, and any trade-remedy measure attached to your code. These frequently exceed the base duty itself.
4. Fix your Incoterm before comparing quotes. An EXW quote and a DDP quote are not comparable numbers. Read the actual definitions from the ICC Incoterms rules and state the term in your RFQ so every supplier prices the same scope.
5. Get a binding ruling for high-value repeat programmes. If you will import the same part for years, a formal classification ruling from your customs authority removes the uncertainty permanently. Your broker will handle the filing.
6. Confirm export-control status on your side. If your part or its technical data falls under defence or dual-use control, the geography question is already answered — see the section below.

Where each side genuinely wins

The honest version of this table has wins on both sides. Anyone showing you one with a green tick down a single column is selling, not analysing.

Read down the "who wins" column and the pattern is clear: overseas wins on the price of making things and on capacity, domestic wins on almost everything to do with time, change and risk containment. Which set matters more is a property of your programme, not of the suppliers.

Running the comparison: a step-by-step method

This takes about twenty minutes per part family and it is the only version of this exercise worth doing.

1. Pick one real part family, not an average. Landed cost is dominated by weight, volume, value density and volume — averaging across a mixed BOM produces a number that describes nothing you actually buy. Do your highest-spend family first.
2. State the annual quantity and the order pattern. 2,400 pcs delivered in four releases is a different cost structure from 2,400 pcs in twenty-four. Write down both numbers.
3. Get both quotes on identical scope. Same drawing revision, same tolerance callouts, same finish specification, same Incoterm, same inspection deliverables. If you want CMM data, first-article inspection reports and material certificates, put them in the RFQ — they are chargeable scope and a quote without them is not comparable to one with them.
4. Price freight with real cartons. Ask the supplier for packed carton dimensions and gross weight per 100 parts, then get a forwarder quote for sea and air. Calculate volumetric weight yourself. This step alone kills more offshore business cases than any other.
5. Classify and look up duty using the method above. Add import taxes and brokerage. Flag the number as an estimate pending broker confirmation.
6. Cost your first-year inspection plan. Assume 100% inspection on the first two lots from a new supplier, then AQL sampling. Multiply hours by your loaded inspector rate.
7. Add inventory carrying cost. Average inventory value × your carrying rate. Offshore carries the transit pipeline; domestic usually does not.
8. Add one bad lot per year at full containment cost. Sorting labour, replacement air freight, or line downtime — whichever your process would actually incur. If the business case only works when nothing goes wrong, it does not work.
9. Add your own hours. Estimate management time per month at loaded rate for each option, higher for the first quarter offshore.
10. Divide by good parts received on time. Not parts shipped. This final division is what makes the comparison honest, and it is where a cheap supplier with a 4% reject rate stops looking cheap.
11. Re-run at 0.5× and 2× volume. If the answer flips inside that range, your decision is fragile and you should split the volume across both geographies rather than commit.

When a domestic supplier is the right choice

There are programmes we turn away, and there are programmes we would quote but where you should buy locally anyway. Being clear about them is more useful to you than winning an order that goes badly. Here is the honest list.

When the lead time is measured in days. If you need parts on a bench this week, no overseas supplier can help you, and the ones who promise they can are quoting ex-works and letting you discover transit later. Our sample lead time of 3–5 working days is time in the shop — transit sits on top of that. A local shop with a spare spindle will beat that every time for a genuine emergency. Keep a domestic shop qualified for exactly this reason, even on a programme that is otherwise fully offshore.

When the work is export-controlled or ITAR. Sendot does not take ITAR or export-controlled work, full stop. This is not a capability gap we are working on — it is a category we decline. If your part, its drawing, or its technical data falls under defence or dual-use control, your supplier must be domestic and appropriately registered, and you should not be circulating the drawing in an international RFQ at all while you determine status. Check with the controlling authority and your compliance function before anything leaves your building.

When parts are too large or too heavy to ship economically. Value density decides this. A part worth $40 that fills a quarter of a cubic metre is a part you should buy near where you use it. Weldments, large frames, big fabricated enclosures — run the freight number first and you will usually stop there. The same logic applies to anything requiring specialised crating or oversize handling at both ports.

When the design will iterate weekly with someone in the room. Early-stage development where an engineer wants to hand a machinist a part, discuss a radius, and get a revised one back the next morning is domestic work. That loop is worth more than the unit-price saving, and trying to run it across twelve time zones costs you weeks that never appear on any quotation. Move offshore once the design is frozen — not before.

When freight would exceed the part cost. Very low-value, high-volume, bulky parts — simple brackets, spacers, thin stampings — frequently fail this test outright. If a $0.80 part carries $1.10 of chargeable-weight freight, there is no negotiation that fixes it. Consolidating into full container loads helps, but only if your consumption rate genuinely supports containers.

When you need a certification we do not hold. Sendot holds ISO 9001. That is the only certification we hold. If your programme requires a supplier certified to ISO 13485, IATF 16949 or AS9100, we are not that supplier. We can supply the documentation an ISO 13485 supply chain requires — material certificates, dimensional records, traceability — and our certification status should be confirmed directly with us for your specific case, but do not assume a certification we have not claimed. If a supplier's certification is a condition of your quality system, verify the actual certificate and its scope with the issuing registrar. The ISO standard family is public; a real certificate names a registrar, a scope and an expiry date, and you should read all three.

When the total programme is small enough that management time dominates. If your annual spend on a part family is $6,000, saving 35% on unit price is $2,100 — roughly two or three days of a senior engineer's loaded time. Any friction at all consumes the entire benefit. Keep small spend local and put your sourcing effort where the money is.

None of these are qualified afterwards. If two or three of them describe your situation, buy domestically and stop reading cost models — the answer is already correct.

Common mistakes buyers make in this comparison

Comparing a prototype quote to a production price. Quoting 5 pieces offshore and comparing it to a domestic production price is the most frequent apples-to-oranges error. Offshore economics are a volume argument; test them at volume.

Ignoring chargeable weight. Modelling air freight on actual kilos when the carrier bills volumetric weight can understate freight by two or three times. Ask for carton dimensions in the RFQ.

Treating tolerance as free. A blanket ±0.05 mm across a drawing where only two features need it adds cost on both continents. We hold ±0.05 mm and Ra 0.2 µm where the function requires it — but tighten only what matters and the price falls everywhere.

Forgetting the pipeline in a cost-down calculation. A saving announced in January does not reach the P&L until the domestic-sourced stock is consumed and the offshore pipeline fills. That is often a full quarter. Model the transition, not just the steady state.

Assuming a low price implies a hidden problem, or that it doesn't. Neither assumption is analysis. A low quote is a question to ask, not a verdict. Our companion article on how to vet an overseas CNC supplier sets out what to actually verify — certificates, process capability, inspection equipment, and what a first-article report should contain.

Skipping first-article inspection to save time. The single most expensive shortcut in offshore machining. A first-article inspection report against a ballooned drawing, reviewed before the balance of the lot runs, is what prevents you discovering a misread datum after 2,000 parts have crossed an ocean.

Awarding 100% of a part family on the first order. Split it. Run a qualification lot, keep the incumbent warm, and shift volume as data accumulates. The cost of dual-sourcing for two quarters is trivial against the cost of a failed single-source transition.

What to send us if you want a real number

A landed-cost model is only as good as its inputs, and the input that costs nothing to improve is the quality of your RFQ. Send a STEP file as the geometry of record — we also accept IGES, X_T, DWG, PDF and STL, and DXF for 2D laser-cut parts — plus a drawing carrying the tolerances and finish that actually matter, your real annual quantity and release pattern, and your required inspection deliverables. State the Incoterm. With that in hand we quote within 12 hours, in English, and you can drop a comparable number straight into the model above rather than guessing at half of it.

Frequently asked questions

Total landed cost is a spreadsheet, not an opinion. Build it once for your highest-spend part family, and the geography decision usually makes itself — sometimes in our favour, sometimes not. If you want a comparable number to put in the unit-price line, send us a STEP file and your real annual quantity through our CNC machining service and we will come back within 12 hours.