How many parts a year justify buying a machine instead of outsourcing?

Ask how many print-hours a week you can keep the machine fed with, not how many parts a year you need. A desktop machine that runs two or three jobs a week pays for itself quickly, because the alternative is a stream of small outsourced orders where setup and shipping dominate the invoice. A machine that runs twice a month does not, and it usually prints worse each time because nobody on the team stays current with it. The honest screening question is whether one named person will own the machine as part of their job. If the answer is nobody in particular, outsource.

Is a service always more expensive per part?

Per part, usually yes — a provider prices in machining time, labour, overhead and margin that your own machine appears not to charge you. The comparison is misleading because the machine's costs are real and simply land somewhere else: engineer hours, failed builds, filament you dried badly, and the week a part waited for someone to have time. Compare the quote against fully loaded internal cost including the labour, and compare it against the delay too. On a one-off prototype the service is often cheaper outright; on a hundred simple brackets it rarely is.

What can a service print that a desktop machine cannot?

Three things, and they are the reason most in-house setups still outsource. Processes you cannot buy at desk scale — SLS and MJF nylon, metal powder-bed fusion, production-grade urethane casting. Size, when a part exceeds roughly 300mm and needs to arrive as one piece rather than four bonded sections. And certified, inspected, repeatable output, where somebody signs that the batch was measured and the material was traceable. If your part needs any of those, buying a machine does not remove the need for a provider; it just handles the other work.

Does owning a machine make prototyping faster?

For small, forgiving parts, dramatically — the value of in-house printing is the ten-minute iteration loop, not the unit cost. Being able to print a fit check at 4pm and reprint it at 6pm changes how a design team works. That advantage narrows fast as parts get larger, more demanding or more material-specific, because a difficult job on a desktop machine can eat two days of failed builds while a provider would have delivered it overnight. Keep the fast loop in-house and let the hard jobs go out; the two are not competing for the same work.

What ongoing costs do people forget when they buy a printer?

Time first — someone slices, levels, watches, removes supports, sands and reprints, and that labour usually exceeds every other line item combined. Then consumables and wear parts: nozzles, build sheets, belts, tubes, and filament that has to be dried and stored properly or it prints badly. Then space and ventilation, which is a real constraint for resin and for anything running ABS or ASA. Then failure rate, which is highest in the first few months on an unfamiliar machine. Budget the first year at well over the purchase price and the decision gets easier.

Can we do both without wasting money on either?

That is what most teams settle on, and the split is by consequence rather than by part count. Keep in-house whatever is fast, cheap, low-stakes and iterative: fit checks, jigs, mock-ups, shop aids, concept models. Send out whatever is structural, sealing, certified, large, metal, or needed in a consistent batch. The useful discipline is to log what you outsource for six months; if one category keeps recurring, that is a real capability case for a second machine, and if nothing recurs, the current arrangement is correct.

3D Printing Services vs Buying Your Own Machine

Nobody buys a machine to save money on the first part. They buy it because outsourcing felt slow, or because a quote for six brackets came back higher than a printer costs, and both of those feelings are honest. Choosing between 3D printing services and buying your own printer goes wrong when the comparison is drawn as price per part, because that is the one line where the machine always wins and the one that almost never decides the outcome. What decides it is how much print-hour demand you can actually keep fed, who owns the machine on a Tuesday afternoon, and whether the parts you need are inside what a desk-sized machine can make at all. This guide sets out both sides, where the line really falls, and what most teams end up doing instead of picking one. If you want quotes to compare against, start with the 3D printing provider directory.

The Comparison Everyone Runs, and Why It Misleads

The usual arithmetic is a quote for a part next to the cost of a machine plus a spool. It comes out in favour of the machine by a wide margin, and it is wrong for a specific reason: the service quote is fully loaded and the in-house number is not.

A provider's price contains machine time, labour, post-processing, quality checks, consumables, failed builds, overhead, packaging, shipping and margin. The in-house estimate usually contains filament. The other items do not disappear when the machine moves to your office — they move onto your payroll and out of the accounting, which is why in-house printing so often feels cheap and rarely gets measured.

Run the comparison this way instead:

Cost the human time honestly. Slicing, levelling, monitoring, support removal, finishing and reprinting. At a loaded engineering rate, an hour of it is worth more than a kilogram of filament.
Include the failure rate you will actually have, not the one the machine is capable of after six months of practice.
Price the delay on both sides. A provider's lead time is visible and quotable. An internal queue behind whoever knows the machine is invisible and frequently longer.
Separate the one-off from the repeat. These are different decisions and merging them produces a wrong answer to both.

How much 3D printing costs breaks down what actually drives a quote, which is the number you need for the outsourced side of the comparison.

Where Owning a Machine Genuinely Wins

The case for in-house is strong and it is not mainly financial.

The iteration loop. Print at 4pm, hold the part, change the model, print again at 6pm. Nothing a provider can offer competes with that on small, forgiving parts, and it changes how a design team works rather than just what it spends. This alone justifies a machine in most product development groups.

Small, low-stakes, high-frequency parts. Fit checks, mock-ups, tooling aids, assembly fixtures, shop templates, enclosure mock-ups, packaging dummies. Individually trivial, constantly needed, and painful to outsource because setup and shipping dominate a small order. 3D printing for jigs, fixtures and manufacturing aids covers the category that most repays owning a machine.

Confidentiality by default. Nothing leaves the building. There are ways to handle this with a provider — see NDAs and IP protection with 3D printing services — but in-house removes the question.

Volume of simple parts in common materials. Once a design is settled and you need forty of something in PLA or PETG, a machine running unattended overnight is hard to beat on cost.

Where a Service Wins, and Cannot Be Replaced

Three categories, and none of them is solved by buying a better desktop machine.

Processes you cannot buy at desk scale. Powder-bed nylon and metal are the clearest examples. SLS and MJF produce support-free nylon parts with usable isotropic strength and clean snap features; metal 3D printing produces parts no filament machine approximates. A desktop printer does not do a worse version of these — it does not do them.

Size. Past roughly 300mm a desktop build volume forces the part into sections to be bonded, with every joint a weakness and a visible line. Large-format 3D printing services deliver it as one piece.

Repeatability and evidence. A batch that is measured, a material that is traceable, and someone accountable for the result. If a customer, a regulator or your own quality system needs that, an office machine cannot supply it whatever it prints. Tolerances and accuracy in 3D printing covers what dimensional promises are realistic to ask for.

Peak load. Nine parts needed on Friday is an ordinary job for a provider and a bad week for one desktop machine. Same-day and rush 3D printing services exist precisely for the spike you cannot staff for.

The Costs the Ownership Side Forgets

Line item	What people budget	What it actually is
Labour	Ignored	The largest cost by a distance — slicing, babysitting, support removal, finishing
Failed prints	Ignored	Highest in the first months; material plus the time lost twice
Consumables and wear	Filament only	Nozzles, build sheets, belts, tubes, adhesives, resin and its disposal
Material handling	Buy a spool	Nylon, PETG and TPU need drying and sealed storage or they print badly
Space and extraction	A desk	Ventilation is mandatory for resin and advisable for ABS and ASA
Learning curve	A weekend	Months to reach reliable output on demanding materials
Someone who owns it	Assumed	Without a named owner the machine quietly stops being used

Why 3D prints warp and how to fix it and why nozzles clog are a fair sample of what the learning curve actually consists of. None of it is hard; all of it takes someone's hours.

Which Side Your Job Falls On

Your situation	Sensible answer
Frequent small iterations, PLA/PETG, non-critical	Buy a machine
A few parts a month, varied materials	Outsource — the machine will sit idle and go stale
Needs nylon, metal, or a certified process	Outsource; a machine does not address this
Part over ~300mm, wanted as one piece	Outsource to large-format
Dozens to low hundreds of a settled design	Low-volume services, or in-house if the material is simple
Thousands of a settled design	Neither — look at rapid tooling and injection moulding
No CAD file yet	Outsource; printing without a CAD file covers the route
Deadline is this week and nothing is set up	Outsource — a new machine will not be productive in time

The Answer Most Teams Actually Land On

Both, split by consequence rather than by part count.

The stable arrangement is one capable filament machine in-house for the fast, cheap, low-stakes work — fit checks, jigs, concept models, shop aids — and a provider relationship for anything structural, sealing, certified, large, metal or needed as a consistent batch. Each side then does what it is good at, and neither is asked to carry the other's work. Functional and end-use 3D printed parts covers the crossover point where a part stops being a mock-up and starts having consequences, which is usually the same point it should leave the building.

There is a simple way to test the split rather than argue about it: log what you outsource for six months. If one category keeps recurring, you have a real capability case for a second machine. If nothing recurs, the current arrangement is already right.

If You Outsource: What to Send

The quality of a quote is set almost entirely by what you send with it.

STEP geometry rather than STL alone, with a drawing marking critical dimensions and any mating or sealing face. STL vs STEP files covers why the distinction matters to the shop.
The duty, in the first line — what the part touches, what loads it, indoor or outdoor, temperature, chemical exposure.
Material by property, not by brand. "Stiff, dimensionally stable, lives outdoors" gets you a better recommendation than a trade name. How to choose a 3D printing material covers the selection logic.
Quantity now and quantity later. Five now and five hundred later is a different process answer.
Your real deadline, because it changes the process recommendation more than anything else on the list.

How to prepare files for a 3D printing quote is the full submission checklist, and how to choose a 3D printing service covers qualifying the shortlist once quotes come back.

Get Both Numbers Before You Decide

The mistake is deciding from one side of the comparison. Price the machine properly — purchase, plus a year of consumables, plus the hours of a named owner — and get two or three real quotes for the parts you actually need, then look at them together. Very often the answer is not the one the spreadsheet suggested, because the machine turns out to be for a different class of part than the quotes were for.

Browse 3D printing providers to get the outsourced number, or start near you — California providers, Texas providers, Illinois providers and Ohio providers between them cover much of the country's prototyping capacity. How the directory works explains how to search it, and online vs local 3D printing services covers the choice you face next.

Related Resources

How Much Does 3D Printing Cost? — the outsourced side of the comparison
Low-Volume 3D Printing Services — where batch work sensibly goes
Functional and End-Use 3D Printed Parts — when a part stops being a mock-up
How to Choose a 3D Printing Service — qualifying a provider
Online vs Local 3D Printing Services — the next decision after this one
Injection Moulding vs 3D Printing — where both answers stop being right
Browse 3D Printing Providers — send the job out