The question is not which machine is cheaper
Two processes with different shapes of cost cannot be ranked without a quantity attached. Take the numbers this page opens with. The router carries 45 minutes of setup and 6 minutes a part; the printer carries 15 minutes of setup and 95 minutes a part, at a machine rate of 30 cents an hour against the router at 12 dollars. At one part the printer wins easily — 14 dollars 55 against 28 dollars 40 — because the router has to pay for three quarters of an hour of skilled setup before it cuts anything. The two draw level between eighteen and nineteen parts, and by forty the router is 16 dollars 20 ahead on the job and pulling further away with every part added.
What makes the trade-off exist at all is that one process wins on setup and the other on unit cost. When the same process wins on both, there is no crossover and no decision to make — this page says so explicitly rather than inventing a quantity.
Attended time is the term people forget
A printer running for 95 minutes unattended costs 30 cents an hour of machine and almost nothing in labour. A router running for 6 minutes with an operator watching it costs 12 dollars an hour of machine and 30 dollars an hour of labour for the share that is attended. That is why the attended percentage on this page is a separate field rather than an assumption: it is frequently the largest single difference between two processes, and it is the one most likely to be set by habit rather than by measurement.
It is worth being honest about it in both directions. A router job where the operator hovers over every part is 100 percent attended, not 10. A print farm where somebody clears plates between jobs is not 0 percent either.
Scrap multiplies the parts you make, not the parts you sell
The model here takes the good parts you need, divides by one minus the scrap rate and rounds up. Forty parts at 8 percent scrap means starting 44. Those four extra parts consume material, machine time and consumables at the full rate, which is why an eight percent scrap rate costs more than eight percent — it costs eight percent of everything variable, on top of a fixed setup that has not moved. On short runs the rounding matters as well: needing five parts at 8 percent scrap means starting six, which is a twenty percent overhead rather than eight.
What the money does not cover
The two parts are not the same part. That sentence is doing a lot of work. A machined part and a printed part differ in tolerance, in surface, in how they behave under load, in which direction they are weak, and in what they can be made from at all. Where those differences matter, the cheaper process is not automatically the right one, and the calculation above is telling you the price of the difference rather than settling the argument.
Lead time is the other absence. Ninety-five minutes a part is not the same commitment as six, even where the totals match, and a run that ties up the only machine for a week has an opportunity cost that no rate on this page captures.
Where this sits
The machine-time side of a router job is on the CNC router job time calculator and the tooling that goes with it on the bit life and cost per part calculator. For the printer column, the 3D print time calculator gives the run minutes and the 3D print quote calculator builds a full price including margin. The laser column comes from the laser cutting time and cost calculator. The same setup-against-unit-cost trade in a decorating shop is on the screen print setup break even calculator, which amortises screen charges the same way. If the question is about a whole business rather than one job, the break even sales calculator works on monthly fixed costs instead of per-job setup.
Questions people ask
What should I put in the machine rate?
Your own figure, worked out from what the machine cost, how long you expect it to last in hours, and what you spend keeping it running. A twelve thousand dollar router expected to give six thousand useful hours is two dollars an hour of depreciation before you add power, tooling, maintenance and the share of rent the floor space represents. A printer that cost four hundred dollars and will give three thousand hours is thirteen cents an hour. There is no published rate for any machine because the number is mostly about your assumptions, not the hardware. What matters for a comparison is that you apply the same method to both columns.
Should tooling go in the machine rate or in consumables?
Either, as long as it goes in exactly once. Putting cutter wear into the machine rate is convenient when the tooling cost per hour is roughly constant. Putting it in the per-part consumables field is more accurate when parts differ a lot in how much cutting they involve, which is usually the case. Double counting it is the error to avoid, and it is easy to make when the machine rate was borrowed from somewhere and its contents are not known. If the router column here is meant to include cutters, the bit life calculator gives a per-part figure that drops straight into the consumables field.
Why does the crossover come out as a range rather than a number?
Because the underlying cost is not a smooth line. The scrap allowance rounds the number of parts started up to a whole part, so the total cost steps rather than slopes, and the two step patterns are different for the two processes. A closed-form crossover would have to ignore that rounding and would be slightly wrong near the boundary, particularly on short runs where a single extra started part is a large fraction of the job. Scanning quantity by quantity and reporting the pair where the answer changes is honest about that step, and on a long run the range is a single part wide anyway.
Can I use this to compare an in-house process against an outside quote?
Yes, and it is one of the more useful things to do with it. Set one process to the outside shop, put the whole quoted price per part into the material field, set the run time and rates to zero and put any setup or tooling charge into the setup field with a labour rate that reproduces it. That models a quote as pure per-part cost plus a fixed charge, which is what most of them are. What it will not capture is the value of not tying up your own machine, or the risk of a lead time you do not control, both of which usually argue for the outside option more than the arithmetic does.
The comparison changes completely when I nudge one rate. Is it that fragile?
Near the crossover, yes, and that is a real finding rather than a defect. When two processes cost nearly the same, the decision is genuinely being made by things smaller than your uncertainty in the inputs, which means cost is not the right basis for it and something else should decide: lead time, finish, who is available, which machine is already free. When one process is ahead by a wide margin the answer is robust to the inputs and you can trust it. Running the numbers twice at the ends of the range you believe each rate could be in is a quick way to tell which situation you are in.