What CNC Machine Should I Buy?

• 1
  Buy or outsource: the first question behind what CNC machine should I buy
• 2
  Match the work envelope and spindle to the parts you actually cut
• 3
  Tolerance, finish and inspection decide the machine class
• 4
  Batch size, labor and floor space rule out more machines than price does
• 5
  The costs that do not appear on the machine quote
• 6
  New, used, or partner capacity: what the numbers look like
• 7
  Machine class vs part and production profile

Most buyers skip a step. They compare spindle tapers and control brands while the real question sits unanswered: does this shop need to own the cut at all? Ownership pays off when one part family repeats, the geometry is stable, and the process window does not move much between lots. If the parts change every quarter, ownership turns into a programming and setup burden you pay for on every job.

Run a simple test on your last six months of orders. If more than 70% of the hours sit on fewer than five part numbers, and those parts fit one envelope, a dedicated machine is usually the cheaper path over two to three years. If the mix is wide and volumes are low, outside capacity wins because the setup cost is shared across many customers.

The middle case is common and awkward. You have one anchor part that runs weekly, plus a long tail of prototypes. Buying one machine for the anchor part and sending the tail out is often better than buying a machine sized for the worst case. Oversized machines cut slowly on small parts and cost more per hour to run.

• 1
  Own it whenOne or two part families repeat, tolerances are tight, and schedule control matters.
• 2
  Outsource whenThe mix changes monthly or volumes stay under a few hundred pieces per year.
• 3
  Split the workKeep the anchor part in-house, send prototypes and overflow to a partner.

Part size sets the frame size, and frame size sets rigidity. A 500 × 500 × 450 mm machine will out-cut a 4,000 mm gantry on the same aluminum part because the structure is stiffer and the tool is closer to the support. Buy the smallest envelope that fits your largest routine part, not your largest possible part.

Spindle power matters more than top speed for steel and titanium. A 15 kW spindle at low rpm with high torque removes material in 4140 and 17-4PH without chatter. A 24,000 rpm spindle is built for aluminum and plastics, where surface speed is high and cutting forces are low. Choosing the wrong one means either slow steel cutting or burnt aluminum tools.

Axis count is the next filter. Three axes handle prismatic parts with features on one face plus two sides. Four axes add a rotary table for parts that need work around a cylinder, such as flanges and shafts. Five simultaneous axes exist for contoured surfaces, deep pockets at odd angles, and parts that would need five separate fixtures otherwise. Each added axis raises programming time and setup complexity, not just machine price.

• 1
  Aluminum at volumeHigh-speed spindle, light frame, fast tool changes.
• 2
  Steel and stainlessTorque at low rpm, heavy casting, coolant through spindle.
• 3
  Contoured surfacesSimultaneous 5-axis with a Ø400 mm rotary table class of work.

Write down the tightest tolerance on your drawing and halve it. That number is your machine target, because process capability needs margin. A part at ±0.025 mm is comfortable on many three-axis mills. A part at ±0.005 mm needs temperature control, a thermally stable frame, and in-process probing. The machine is only one part of that chain.

Surface finish follows the same logic. Ra 1.6–3.2 μm is normal as-machined output. Ra 0.8–1.6 μm needs sharp tooling, stable fixturing and controlled stepover. Ra 0.2–0.8 μm usually means a finishing pass with small tools and slow feed, or a secondary process such as polishing. If your drawing calls for fine finish on a large part, the machine needs the spindle speed and the rigidity to hold it across the full travel.

Do not forget inspection. A machine that holds ±0.005 mm is wasted if nobody measures to that level. Budget for a coordinate measuring machine, or plan to buy inspection as a service. On high-mix work, the measuring step often costs more than the cutting step.

Batch size tells you whether to chase cycle time or setup time. On a 10,000-piece run, a few seconds per part changes the payback by months. On a 20-piece run, setup dominates, so quick-change workholding, tool presetting and offline programming matter far more than spindle acceleration. Many shops buy for the big run and then lose money on the small ones.

Labor is the constraint people underestimate. Five-axis simultaneous work needs programmers who can think in tool vectors and check collision. Mill-turn work needs people who understand both turning and milling cycles. If you cannot hire or train that skill, the machine sits idle or runs conservative programs that waste its capability.

Floor space and utilities are the last gate. A large gantry needs foundation work, crane access and three-phase power at high current. A compact mill-turn cell needs less space but more coolant management and chip handling. Check ceiling height for crane lifts and door width for delivery before you sign anything.

• 1
  High volumeOptimize cycle time, automate loading, minimize tool changes.
• 2
  High mixOptimize setup, use preset tooling, keep programs offline.
• 3
  Skill gapChoose the simpler machine you can actually staff.

A machine price is maybe half of the real cost in year one. Add tooling, workholding, a CAM seat, a post-processor, and the time to prove out first articles. A simultaneous 5-axis center needs a CAM seat that supports it, and that license can cost as much as a small machine over five years.

Consumables scale with runtime. Coolant, filters, way lube, inserts and spindle service add up. A high-speed spindle has a finite bearing life and rebuilding it is a scheduled cost, not a surprise if you plan for it. Keep a maintenance budget of a few percent of machine value per year, and track spindle hours.

Then there is the learning curve. Expect several weeks of reduced output while operators and programmers climb the curve. Plan a low-risk part family for the first month. Do not put your most valuable job on a new machine in week one.

• 1
  Tooling and workholdingOften 10-20% of machine cost in the first year.
• 2
  CAM and postFive-axis seats are priced separately and renew yearly.
• 3
  Ramp-upBudget several weeks of lower throughput.

Used machines can be a good entry point if you can inspect them. Check spindle runout, backlash on each axis, way condition and the hours on the control. Ask for a test cut on a part you bring. A used machine with a tired spindle costs more than its price in scrap and downtime.

New machines bring warranty, training and parts availability. That matters most when the machine is on a critical path. If a breakdown stops a customer line, warranty response time is worth real money and should be part of the comparison, not an afterthought.

There is a third path that many shops ignore: keep the anchor work in-house and use a partner for peaks and prototypes. A partner with 127 high-precision CNC machines, including 16 simultaneous 5-axis centers, absorbs demand spikes without adding fixed cost. We run this model for teams that are not ready to commit capital. No minimum order quantity, and parts ship in 3–5 days once production starts.