3D Printing On Demand for Local Business: Complete Guide (2026)

Autodesk Fusion for personal use is explicitly for non-commercial personal projects and home-based non-commercial fabrication. The moment you invoice for a part you modelled in it, you are outside the licence. The commercial subscription is listed by Autodesk at 85 US dollars per month or 680 per year.

Onshape Free is also non-commercial, and it has a second problem that is far worse when you work for businesses: documents on the free plan are public, and Onshape terms grant anyone who opens a public document a broad licence to use, copy, modify and distribute the intellectual property in it. If you model a customer part in a free Onshape account, you are publishing their design. That will end a commercial relationship and could expose you to a claim.

Your realistic options:

• FreeCAD. Free, open source, genuinely capable of parametric part modelling, and commercially usable. The learning curve is steeper and the workflow less forgiving.
• Fusion commercial. Roughly 680 a year. Fine as a cost of doing business once you are billing.
• Onshape paid, Solid Edge, Alibre, Plasticity, Shapr3D. Various subscription levels. Compare them on annual cost and on whether your customers will ever send you native files you need to open.

Budget for CAD from day one. It is a real operating cost, and you will spend more time in it than at the printer.

3D scanning

Do not buy a scanner first. Most replacement-part work goes faster and comes out more accurate with callipers, a photograph with a ruler in frame, and a bit of thinking. Scanners earn their place on organic or freeform shapes: a moulded handle, a curved trim panel, a worn casting.

If you do buy one later, handheld units aimed at reverse engineering are commonly listed from a few hundred to around 1,500 in US dollars depending on stated accuracy. The scan is only the beginning. Turning a mesh into clean parametric CAD is skilled work that takes longer than the scan itself.

Materials: what to stock and what it costs you

Filament price tracking sites in 2026 put PETG at an average of roughly 18 US dollars per kilogram, with single spools available from around 8 dollars per kilogram and bulk assortments lower. Engineering materials cost considerably more.

A sensible starting shelf:

• PLA. Cheap, dimensionally stable, easy. Fine for jigs used at room temperature, mock-ups and fit checks. It softens in a hot car or near machinery, so keep it away from anywhere warm.
• PETG. Your default functional material for most shop work. Tough, chemically reasonable, less brittle than PLA, and better with warmth.
• ASA. Where PETG goes for outdoor or UV exposure. Needs the enclosure. Smells.
• Polycarbonate or PC blends. Higher temperature and impact resistance, needs a hot chamber and dry filament.
• Nylon and nylon carbon fibre. Wear resistance, low friction, sliding parts, gears, bushings. Hygroscopic and demanding; keep it dry and print it from a dry box.
• TPU. Flexible pads, bumpers, soft jaws, grippers, gaskets that do not need to be pressure rated.

Two practical notes. First, abrasive filled filaments (carbon fibre, glass fibre) destroy brass nozzles in tens of hours. Fit a hardened steel or ruby nozzle before you print them. Second, keep a written record of which spool, which brand and which batch went into each customer job. When a part fails in six months, this is how you find out whether it was your process or your supplier.

What it costs you to start

You have now seen every machine, tool and spool the work asks for. Putting one figure on that pile is what decides whether you begin with a single printer or a small bank of them.

Here are three honest tiers, in US dollars. Convert roughly for your own market.

Minimum viable, around 900 to 1,400. One enclosed printer bought on offer, four or five spools across PLA, PETG and one engineering material, a decent calliper, a hardened nozzle, heat-set inserts and an insertion tip, a soldering iron, deburring tools, a filament dryer, and business registration. CAD on FreeCAD. No scanner, no insurance yet, working out of your spare room.

Sensible, around 2,500 to 4,000. Two enclosed printers so one machine failure cannot stop your business, a wider material stock, a proper dry box setup, a commercial CAD subscription, general liability insurance, a modest set of sample parts to show customers, and a small stock of fasteners and inserts.

Committed, around 5,000 to 8,000. Three or four machines, one of them large format or high temperature, a resin machine with wash and cure and extraction if you have a customer for it, a handheld scanner, dedicated extraction to outside air, a limited company or LLC with an accountant, and product liability cover.

Which tier could you fund today without touching money your family is counting on? Start there, and let the work pay for the next one.

Your ongoing costs are lower than most side businesses, though never zero: filament, electricity (a mid-size FDM machine typically draws well under 200 watts averaged over a print, so this is real but small), nozzles and build plates, CAD subscription, insurance, and replacing a printer roughly every few thousand hours of hard use.

How to price a part

So you know roughly what the kit costs you. What you put on the invoice decides whether any of it ever pays for itself.

This is where most new operators sink their own business by quoting like a hobbyist.

The hobbyist formula is material plus a bit of time. It gives you a price of four dollars for a part that took you ninety minutes of CAD, and it teaches your customer that your work is worth four dollars. Is that the lesson you want them to learn?

Price a B2B job from four components, and never skip the first.

1. Engineering time. The time you spend understanding the problem, modelling or fixing the file, checking a drawing, and deciding the orientation and infill. Bill it at a professional rate. In the US, BLS puts the median annual wage for machinists at 56,150 dollars as of May 2024, which is roughly 27 dollars an hour as an employee cost, and a shop rate sits well above an employee cost. Commonly quoted US machine shop rates in 2026 sit somewhere between 75 and 250 dollars per hour depending on machine and complexity. Your design time is worth 60 to 120 an hour in most markets. Charge for it as a line item, or fold it into a setup fee, but charge for it.

2. Machine time. Take the printer cost, divide by a conservative life of 4,000 to 6,000 hours, add electricity, add a maintenance allowance for nozzles, plates and belts, and add a failure allowance. A 700 dollar printer over 5,000 hours is 14 cents an hour of depreciation, which is trivially small and misleading. Your real constraint is the number of machine hours you have, and every hour you sell cheaply is an hour you cannot sell dearly. Set a machine rate you are happy to be busy at. Two to five dollars per machine hour is a realistic floor for straightforward work; treat it as a floor and keep climbing from there.

3. Material. Grams from the slicer, multiplied by your actual cost per gram, multiplied by a waste factor of at least 1.2 to cover purge, supports, brims and the print that failed at hour nine.

4. Post-processing and handling. Support removal, deburring, inserts, tapping, assembly, inspection, packaging, delivery. This is the cost that silently eats your margin, because it happens at your bench in fifteen-minute pieces you never write down.

Then apply the two rules that turn a hobby into a business:

Set a minimum job charge. Something in the region of 60 to 150 in local currency. It filters out the enquiries that cost you two hours of messages for a nine dollar part, and it reflects the truth that every job carries a fixed admin cost whatever its size.

Charge for the first article, then charge less for repeats. The first part costs you design time. The tenth part is a slicer file and a machine hour. Quote a setup fee plus a per-part price. Customers in manufacturing understand this instantly, because it is exactly how tooling is priced everywhere else.

Finally, price against your customer's actual alternative and leave your own cost out of that comparison. If their alternative is a machine shop with a 400 minimum and a three week lead time, or an OEM spare at 280 that arrives in six weeks, a 120 part delivered on Thursday is cheap. If their alternative is a five dollar part from a catalogue, walk away.

For a benchmark, online manufacturing marketplaces such as Xometry and Craftcloud will give you an instant quote on any file with no minimum order. Run your customer's part through one before you quote. If you cannot beat their price meaningfully on lead time, on unit price, or on the fact that you will actually pick up the phone, you have no business in that job.

Those early jobs at the $500 end of this guide's income range are relief money, and you will know exactly where each one came from. A few replacement parts for a packaging line, after filament and failed prints, could cover the car insurance and the electricity while your trade builds. What would it feel like to pay one of those bills from a bracket you drew yourself? Quote on lead time first, because that is where you beat an online quote.

Then picture the day a shop calls you before it calls anyone else. A regular customer who trusts your lead times is steady money: the boiler service you stopped postponing, the trip to see family you finally book. It all starts with a price that respects the hours behind it.

Registration, licensing and tax

That is the money side settled on paper. Now your business has to exist in a form that a client's accounts department and your tax office will both accept.

There is no additive-manufacturing licence anywhere in the four main English-speaking markets. What you need is ordinary small business compliance, and it differs by country.

United States. There is no federal licence for this activity. The US Small Business Administration is explicit that registration, licensing and permitting requirements are set locally, so you must check your state, county and city. In practice you will typically need: a business structure (sole proprietorship or an LLC), an EIN if you form an entity or hire, a state sales tax permit because you are selling tangible goods and most states tax them, and possibly a local general business licence and a home occupation permit if you work from home. Zoning is the one people forget: some residential zoning codes restrict manufacturing or customer visits at home, and running four printers in your garage can be read as light manufacturing.

United Kingdom. Register as a sole trader with HMRC for Self Assessment, or incorporate at Companies House. Under the trading allowance, if your gross self-employed income stays under 1,000 pounds in a tax year you generally need not register for Self Assessment on that income alone, but above that you must. VAT registration becomes compulsory above the registration threshold; check the current figure on gov.uk instead of trusting a number in an article. Check your tenancy or mortgage terms and local council rules for running a business from home.

Canada. Register the business name provincially or incorporate, get a Business Number from the CRA, and register for GST/HST once you exceed the small supplier threshold. Provincial sales tax rules vary.

Australia. You need an ABN, and GST registration once turnover passes the registration threshold. Council rules on home-based businesses vary by local government area.

In all four, the recurring themes are the same: you are selling goods, so sales tax or VAT or GST applies; you are working from home, so zoning and lease terms apply; and you are handling other people's designs, so contracts apply.

Liability, safety and the lines you must not cross

This section decides whether you are still trading in year three. Please read it slowly.

A printed thermoplastic part counts as an engineered component only when someone has engineered it. Fused filament parts are anisotropic: they are substantially weaker across layer lines than along them, their strength depends on print settings you may change between batches, and they creep under sustained load and degrade with heat and UV. That is fine for a drilling template. It is unacceptable for anything that holds a load above a person, restrains a machine, carries pressure, or forms part of a safety system.

Say no to, at minimum:

• Lifting, rigging, fall arrest or anything overhead
• Structural or load-bearing components
• Pressure-containing parts
• Brake, steering or suspension components
• Machine guards and interlocks unless the customer's own engineer has specified and signed off the design
• Firearms components, which are separately and seriously regulated in every jurisdiction here
• Anything implanted, or any medical device, which brings FDA, UKCA/CE, Health Canada or TGA regulation into play
• Food contact and drinking water contact, unless you have specifically sourced a compliant material and can deal with the fact that layer lines harbour bacteria and standard printers are not food-safe environments
• Aviation and marine safety-critical parts

Put this in writing. Write a short terms of supply document stating that parts are supplied to the customer's design and specification, that the customer is responsible for deciding fitness for purpose, that no safety-critical use is authorised, and that materials and process are as agreed. That document decides who pays when a part fails, so treat it as protection for you and your family.

Insurance comes next. General liability is your baseline; US insurers publish averages in the region of 45 to 80 US dollars per month for small businesses, though your rate depends entirely on industry classification and location. You also want products liability, commonly quoted in the several hundred to low thousands per year band for small manufacturers. Tell your broker, honestly, that you are manufacturing components for commercial customers. A policy bought on the basis that you sell craft items will not respond when an industrial customer claims. If a part you made failed tomorrow, would your current cover actually pay?

There is a regulatory change worth knowing about if you sell into the EU. Directive (EU) 2024/2853, which member states must transpose by 9 December 2026, revises the product liability regime and explicitly treats digital manufacturing files, meaning the files that drive a 3D printer, as products in their own right. If you sell design files as well as parts, or you supply parts into the EU, take that to an adviser instead of guessing.

Printing somebody else's part

You will be asked to copy an OEM spare. Sometimes that is legitimate and sometimes it is not, and the honest answer is that it depends on the jurisdiction and on the part.

The general shape of it: a part may be covered by patents, by registered or unregistered design rights, and its markings by trademarks. Copying a purely functional part whose patents have expired is usually the least risky case. Copying a distinctive shape while it is still protected, or reproducing a logo or brand name moulded into the original, is where you get into trouble. Law firms writing on this consistently note that additive manufacturing has outrun the settled case law and that liability may attach to whoever made the physical article, which in your workshop means you.

Rules that keep you out of most of it:

• Never reproduce a manufacturer's logo, brand name or part number on a printed part. Remove them from the model.
• Favour internal tooling and jigs, which the customer owns and uses in their own facility, over parts that will be resold onward.
• For repair parts, get the customer to confirm in writing that they own the equipment and that the part is for their own repair use.
• Decline work where the customer is obviously planning to resell copies of a branded part.
• If a customer sends you a file, your terms should require them to warrant they have the right to have it made.

Fumes, fire and where you put the machines

The US National Institute for Occupational Safety and Health published a guide, Approaches to Safe 3D Printing, covering makerspaces, schools and non-industrial settings, after research showing that fused filament printers emit ultrafine particles and volatile organic compounds. NIOSH guidance points toward enclosed printers, local exhaust ventilation, and where practical, exhausting to outdoor air.

In plain terms, here is what you should do:

• Run enclosed machines with the door closed.
• Never put printers in a bedroom, or in any room where you or anyone else in your home sleeps.
• Ventilate, ideally with a duct to outside. A carbon filter reduces odour, and only a duct actually removes the fumes.
• Treat ABS, ASA and polycarbonate as the materials that most need this.
• Never leave a printer running unattended in a room without a working smoke alarm. Heated beds and hot ends fail; thermal runaway protection helps but guarantees nothing.
• Keep a fire extinguisher suited to electrical fires nearby, and do not stack spools on top of the machine.

If you scale to several machines in a garage or a unit, check that your electrical supply and circuit protection can cope, and check whether your insurer needs to be told.

Getting your first client

Everything up to here you could do alone in a garage. This is the point where somebody else has to say yes.

Your first client is the hardest, and they come from doing something uncomfortable: showing up in person with objects in your hand.

Build a sample kit before you sell anything. Six to ten parts that show capability. Save the creative pieces for later. A jig with a heat-set insert and a threaded fastener. A part in PETG and the same part in nylon so they can feel the difference. Something with a tight sliding fit. A living hinge. A part with a printed thread and one with a tapped insert. A deliberately overhanging part to show what you can do without supports. Nobody in this market is impressed by a Benchy or a dragon. They are impressed by a part that fits.

Make a list of fifty local businesses. Use the business directory, an industrial estate map, and a search for engineering, fabrication, signage, plastics, packaging, agricultural repair, marine, dental lab, prototype. Aim at businesses with between three and fifty staff. Larger firms have procurement processes that will eat months of your time.

Turn up. Ask for the production manager, the maintenance lead or the workshop foreman. Skip the owner, and definitely skip marketing. Say something close to: I print plastic parts locally, I do one-offs and small batches with a two to four day turnaround, is there anything on your bench right now that is broken, taped up, or that you have been waiting on a supplier for. Then be quiet. In a workshop of any age, there is always something.

Do the first job cheaply and fast, once. Charge something, because free work is never valued and it sets a precedent. Cheap and same-week gets you a reference part sitting on their bench with your name on it. Then price properly from the second job.

Ask for the referral out loud. Who else do you know who has this problem? Manufacturing communities are small, and people talk.

Only then, build your online presence. A one-page site with your city in the heading, photographs of real parts, the materials you stock, your turnaround, your minimum charge and a phone number. Register on Google Business Profile so you appear for searches like 3D printing service near me, which is a low-volume but extremely high-intent query. Consider listing on manufacturing marketplaces to fill idle machine time, accepting that the margin will be thinner and the customer belongs to the marketplace.

Skip building a social media following. It will bring you consumer enquiries, which is the work you are trying to avoid.

What actually goes wrong

You quote from the slicer. The slicer tells you grams and hours. It says nothing about the ninety minutes of CAD, the two failed prints, the support removal, the four heat-set inserts, or the drive across town. Operators who quote from the slicer are busy and broke.

You take the wrong customer. One consumer job with a vague brief and an emotional attachment to the outcome will eat more of your week than five industrial jobs. A minimum charge and a written scope fix this.

You let a customer treat you as their free design department. They send a photograph and a sentence. You produce three iterations. They pay for one part. Charge for design as a separate, quoted line before you start modelling.

Tolerances. Your customer says the hole is ten millimetres. Printed holes shrink. You print ten, the pin does not fit, and you look incompetent. Characterise your machine once, properly: print a test block with holes and pins from six to twenty millimetres, measure everything, write down the offsets per material, and apply them every time.

Wet filament. Nylon and PC left on an open shelf give you weak, stringy, ugly parts. For those materials, dry boxes are mandatory.

Overnight failures. A print that fails at hour eleven of a fourteen hour job costs you the material and, worse, the delivery date. Never quote a lead time that assumes a first-time success on a long print. Add a buffer.

Being a single point of failure. One printer means one broken extruder equals a missed deadline equals a lost account. You buy the second machine for reliability first and capacity second.

Cash flow with business customers. Commercial buyers pay on terms, often thirty days and sometimes later. Get a purchase order number, invoice the day you deliver, and for new accounts take payment up front until they have earned terms. Could your household wait thirty days for the money you have already spent on filament?

Doing something you should have refused. This is the one that ends businesses. Read the liability section again.

What you can realistically earn

Nobody should publish confident income claims for this, and the figures circulating online are almost always either the revenue of a multi-machine shop dressed up as a side hustle, or fiction. What follows is reasoning from cost structure. It draws on nobody's reported earnings.

Two things limit what you can do: machine hours and your own hours. A single printer running realistically, allowing for failures, maintenance and the fact that you sleep, might deliver 250 to 400 billable machine hours a month if you are genuinely busy, which almost nobody is in year one. Your own time is the harder limit, because every job carries design, quoting, post-processing and delivery.

A realistic shape for your first year, part time, one to two machines:

• Months one to three: mostly cost. A handful of small jobs, possibly nothing. Revenue plausibly under 500 a month.
• Months four to nine: two or three repeat accounts, jobs mostly in the 60 to 400 band. Revenue plausibly 500 to 2,000 a month, with materials running roughly 10 to 20% of revenue and the rest going to your time and overheads.
• Year two, if the referrals compound: a solo operator with three or four machines and a handful of regular accounts is plausibly in the 2,000 to 6,000 a month revenue range, with gross margin after materials, consumables, insurance and software commonly in the 55 to 75% band.

So once you are established, your realistic monthly take-home sits roughly in the low hundreds at the start and the low thousands when it works. Treat anything above that as a full-time business with staff, a business with an unusual niche, or a claim you should not believe.

The genuinely attractive thing here is the floor. Your costs are low, your inventory risk is near zero because you print to order, and the work does not follow the seasons. It behaves like a small trade business, much more than it behaves like an online store.

Near the $3,000 end, held for a few steady months, the trade starts to reach your family. Some of what you keep could cover your parents' winter bills so your dad can pass on the extra Saturday shift, all paid for by parts that kept other people's machines running. Keep adding customers before you offer help your family will count on, so it never depends on two accounts.

So what does another season of thinking about it cost? The workshops near you keep losing hours to broken parts, and sooner or later someone else solves it for them. Your printer budget will buy the same next year, and the customers will be harder to win. Make the list tonight, and walk into one shop this week with a part you printed to show them.

Who should skip this

• If you want passive income, skip it. This is a service business with a machine in it. The machine does not sell, quote, design, deburr or deliver.
• If you are unwilling to learn CAD, skip it. You can start by printing files customers send you, but the money is in the jobs where the customer has no file, only a broken part and a problem.
• If you dislike cold approaches, skip it. Your first ten customers come from walking into workshops. No other route works as quickly.
• If you live somewhere without local manufacturing, skip it. If your nearest industrial estate is an hour away and the local economy is entirely retail and hospitality, the demand is missing. Check before you buy a printer: count the businesses on your fifty-name list. If you cannot find fifty, think again.
• If you cannot ventilate a dedicated space, skip it. A printer in a bedroom will never be a business.
• If you hope to compete on price with online marketplaces, skip it. You cannot win there. Your advantages are speed, proximity, and being a person who answers the phone.
• If you find it hard to say no, skip it. The work you refuse is what protects you.

Where this is heading

Desktop machine capability has improved sharply and prices have stayed flat, which means the barrier to entry keeps falling and more people will try this. That cuts both ways: it is easier for you to start, and it will be easier for the next person too. Your defensible position lies outside the printer. It is your relationships, what you have learned about your customers' machines and parts, the tolerance data for your own equipment, and your reputation for turning things around when someone's line is down.

At the same time, your customers are increasingly able to buy a printer themselves. Some will, and you will lose them. That is no reason to avoid the business, and it is a good reason to keep adding customers instead of depending on two, and to move up toward the work that needs engineering judgement as well as machine time. Five years from now, what will your customers still need you for that a printer on their own bench cannot give them?

That falling barrier is exactly why timing matters. Every year it gets easier for someone else to buy the same printer and knock on the same doors. Your relationships and your tolerance data take months to build, and they are the part nobody can buy. Start building them before the next wave of hobbyists turns professional.