I'm the office administrator for a 26-person product-development company. I manage roughly $180,000 a year in external purchasing across 14 vendors, and I report to both operations and finance. When I took over in 2020, our prototyping workflow was split: the design team wanted 3D printed parts, the manufacturing lead wanted machined ones, and nobody agreed on a process.
This is not a vendor shootout, and it's not a “3D printing will replace CNC” post. It's a practical comparison I run every time a request lands on my desk: should we use our in-house Ultimaker 3D printer S5, or should I order CNC milling parts from an outside shop? I'll compare them on four dimensions—cost, lead time, quality, and process control—or rather, the four dimensions that caused the most arguments here.
1. Cost: per-part price is not the full story
Most buyers focus on per-unit pricing and completely miss setup fees, revision costs, shipping, and the labor spent managing the order. That's true for both in-house 3D printing and outside machining.
With the Ultimaker S5, the visible cost is filament and electricity. A PLA part might be a few dollars in material. The invisible cost is my time: failed first layers, wrong Cura profiles, someone choosing a material that doesn't bridge well. Our first six months probably cost $3,000 in material and wasted time. The printer itself is capital, so it only pays off if the design team actually uses it consistently.
When I order CNC milling parts, the quote is higher, but good metal CNC machining suppliers include setup, material certification, tolerances, and surface finish in their quote. That makes the total cost easier to predict. No surprise spool of exotic filament that turns out to be the wrong diameter.
The question everyone asks is, “What's your best price?” The question they should ask is, “What's included in that price?” A $200 quote with no mention of deburring or shipping becomes $400 later. A $250 quote with everything itemized is usually the better deal.
Conclusion? For one-off plastic prototypes, the S5 wins. For metal parts or small batches over about 10 units, outsourcing was cheaper than I expected—and definitely cheaper than metal-filled filament that needed a hardened nozzle and still looked terrible.
2. Lead time: Fast on paper is not fast in real life
It feels obvious that an in-house 3D printer is faster than a machine shop. The real bottleneck is the queue and the risk of a failed print. Last year, I assumed the S5 could run a part overnight. I sliced it, started the print, and found out the next morning that it had stopped at layer 40 because the filament spool was tangled. A 5-minute check before starting would have saved a full day.
That failure is why I created my 12-point checklist. It has saved us an estimated $8,000 in avoided rework and rush fees. 5 minutes of verification beats 5 days of correction.
The bigger surprise: a good CNC machining supplier can beat your in-house turnaround, especially for metal parts or when the printer queue is full. We once sent a small aluminum bracket to a shop and got it back in 7 days. If I remember correctly, our internal queue was two days behind before we even started, and the first print would have needed a rework anyway.
There's also the “office interference” problem. Someone printed a dragon for a birthday party on our only S5. (surprise, surprise). That pushed a real part by a day. The Ultimaker Digital Factory login helps because I can check the actual print status from my desk instead of walking over and trusting a post-it note, but it doesn't untangle filament or make the build plate bigger.
My rule now: if the part is small, plastic, and low-risk, print it. If it's metal, tolerance-critical, or needed by a hard date, get a quote. The numbers said in-house was faster. My gut said “you're betting the deadline on a machine you don't babysit.” My gut was right.
3. Quality and materials: Match the process to the part
FDM has limits. Layer lines are not just cosmetic; they make parts weaker in one direction. I've had printed parts crack along layer lines when the design team used them as if they were injection molded. For structural parts, get them machined.
A good metal CNC machining supplier gives you tolerances in thousandths of an inch, a smooth surface, and material traceability. That's not an insult to the S5—it's the right tool for a different job. The old “everything in-house with a desktop 3D printer” thinking comes from the early RepRap era, when any working machine felt like magic. That's changed. As of January 2025, Ultimaker's official specs list the S5's build volume as 330 x 240 x 300 mm, and it prints thermoplastic parts. It is not “every manufacturing process.”
And sometimes the right process is not machining at all. Someone in my office asked last month, “Can you cut paper with a laser cutter?” Yes, you can. Paper cuts beautifully, but it's a different procurement lane. If you're sending a job out, don't send paper-cutting work to a metal CNC shop. Laser cutting paper needs lower power and proper ventilation, and the paper weight matters. Standard conversions put 20 lb bond at about 75 gsm and 80 lb cover at about 216 gsm. Thin paper burns rather than cuts if the power is too high.
Conclusion: choose based on material and tolerance. Metal, tight tolerances, or a need for certification means CNC. Thermoplastic prototypes and internal fixtures mean the S5. Paper means laser cutting—and make sure they know the exact paper weight.
4. Workflow and control: Who owns the process?
This dimension doesn't show up on a quote, but it's often the deciding factor. With an in-house printer, I have the Ultimaker Digital Factory login and can see if a job finished, which designer submitted it, and how long it took. I don't have to email a supplier and wait for a status update.
But control comes with responsibility. When the print fails, there is no supplier to hold accountable. It's our file, our material, our profile. That is uncomfortable from a procurement perspective. Outsourcing, on the other hand, puts a contract between you and the result. You can specify revision limits, inspection reports, and a rework policy. That clarity is worth real money.
In our 2024 vendor consolidation project, I cut our supplier list from 22 to 14. The S5 helped by absorbing quick-turn plastic requests; the metal CNC machining suppliers we kept were the ones that answered my checklist accurately. Before I order CNC milling parts from any supplier, I confirm lead time, material cert, tolerance, freight, and revision policy. That prevents the “I assumed it was included” conversation after the invoice arrives.
Final recommendations: When to use each
If you're in the same position—managing parts for engineers, designers, or production leads—here is how I decide:
- Use the Ultimaker S5 in-house when the part is a thermoplastic prototype, fits the build volume, and the design team can still adjust the file. Log into Digital Factory first, check the queue, and verify the material profile.
- Order CNC milling parts when the part is metal, needs tight tolerances, requires material certification, or the volume is too large for a printer. Keep a short list of verified metal CNC machining suppliers before you have an emergency.
- Laser cutting? If someone asks, “Can you cut paper with a laser cutter?” Yes—use a laser service, not a CNC shop or a 3D printer. And remember the paper weight matters.
I've started every difficult order by asking, “What's the worst failure mode?” If the worst case is a 20-cent plastic part and a reprint, use the S5. If the worst case is a delayed production line or a failed audit, use a CNC supplier with a contract and an inspection report.
5 minutes of verification beats 5 days of correction.
That, more than any spec sheet, is the lesson from managing an Ultimaker S5 and a CNC supplier list at the same time.
