There is no single right answer to “how should I source a wire EDM machine”

I have been running machining-equipment procurement for a 180-person job shop since 2019. In that stretch I have signed off on wire EDM purchases four different ways — two new machines straight from the OEM, one refurbished unit through a sourcing broker, and one hybrid order that mixed both. The right answer changed every time, and it wasn’t because the market moved. It was because our tolerances and our cash position moved.

So instead of telling you to always buy new, or never buy refurbished, I’m going to lay out the three situations I have actually bought in, what each one cost us, and the four questions that will tell you which situation you are in. Pricing below reflects quotes we collected between November 2025 and March 2026 — I wish I had tracked this data going back to 2019, but I only started building the TCO spreadsheet in October 2024. Verify current numbers before you budget; electronics supply and ocean freight have both been moving fast enough to shift landed cost by 8-12% inside a single quarter.

Situation A: Your tightest true-position tolerance is under ±5 μm — buy new, and buy through the OEM’s authorized channel

If your prints call out true position in the 3-5 μm band, and you are cutting hardened tool steel or carbide where a single out-of-tolerance part scraps $800 of material, do not try to save 40% by going through a broker.

The reason is that the accuracy spec you are paying for is defined against a specific set of test conditions:

ISO 14137:2015 — Test conditions for wire electrical-discharge machines (wire EDM) — Testing of the accuracy. The standard specifies the environmental and measurement conditions under which positioning accuracy and repeatability are proven.

A refurbished machine can absolutely meet those specs. But proving that it does requires a verification chain most brokers cannot supply, and the ones who can will charge you for it. That hidden cost is the part the spec sheet never shows.

What we paid: about $215,000 per machine for the two new units we bought in 2023 and 2024 — including installation, operator training, and a two-year service contract. If I remember correctly, the second one came in closer to $208,000, though I might be misremembering after the freight surcharge. Either way, call it 30-50% more than the refurbished quotes we collected that same quarter. On a per-part basis, that premium paid itself back in about 14 months, because we stopped scrapping parts and stopped re-cutting.

The boring part matters more than I expected: OEM service response time. When our 2022 machine started throwing wire-break errors every 40 minutes, we had a technician on-site in 36 hours. When our refurbished unit (Situation B, below) developed a similar symptom, we waited nine days — no, ten, I’d have to check the maintenance log. Two weeks of downtime with one machine in the cell is a $40,000 revenue hit for us. That number never appears in the purchase comparison. It just appears in the P&L three months later.

Situation B: Tolerances of ±15-25 μm and a capital budget under $150,000 — refurbished through a sourcing service is usually the better buy

This is where the counterintuitive advice kicks in. Most first-time buyers assume refurbished equals risky. In my experience — six wire EDM purchases since 2020, three of them refurbished — the risk isn’t in the machine. It’s in the seller’s verification process.

It is tempting to think refurbished means 45% off new and to stop your analysis there. The percentage is the least interesting number in the deal. What the brochures won’t tell you is that rebuilt, refurbished, and reconditioned mean completely different things depending on who is using the word. At one end, it means cleaned, painted, and powered on. At the other, it means every servo, guide, and power supply was tested against OEM specs and replaced if out of tolerance. The price difference between those two definitions is around $60,000, and the performance difference is, well, everything.

What we paid for our one refurbished unit: right around $98,000, against an original new price of about $215,000. That’s roughly 45% off. It has held ±12 μm repeatability over 18 months of two-shift operation — fine for our roughing and semi-finish work. Two of our competitors told me they would never buy refurbished. Both of them have since bought refurbished. I don’t say that to win an argument; I say it because the objection is usually about the process, and the process is fixable.

The four verification steps that saved us from a bad purchase:

  1. Insist on a test cut before payment — not a demo, a test cut on your material with your wire and your program. If the seller hesitates, that hesitation is your answer.
  2. Ask for the service history, then call the service tech who signed it. Two of the four brokers I talked to said records were not available. One of them later admitted the “rebuild” had been cosmetic only.
  3. Check surface finish capability against VDI 3400 reference plates, not against glossy photos. VDI 3400 is the standard reference for EDM surface roughness — if the seller can’t produce a VDI chart or a recent Ra measurement from a test cut, walk.
  4. Confirm the control software version and parts availability. A 2014 control that the OEM dropped support for in 2022 is a $15,000 problem waiting to happen. Ask the OEM directly, not the broker.

Situation C: You need three machines in 18 months and delivery time is the deciding variable — hybrid sourcing wins

This is where I would push back on both the always-new advice and the always-refurbished advice. When you are adding capacity against a customer contract with an LD committed delivery date, the lead time on a new machine — typically 4-6 months for the major Japanese and Swiss platforms, longer for custom configurations — can kill the deal before price even becomes a conversation.

The move that worked for us in 2024: one new machine through the OEM for the tight-tolerance cell, two refurbished units through a sourcing broker for roughing and semi-finish, all arriving within a 90-day window. We ended up with three machines in the shop for the price of roughly 1.9 new ones. On our books, the hybrid order saved about $190,000 relative to an all-new purchase, and it hit the contract ship date that the all-new plan would have missed by roughly six weeks.

The catch is that this only works if someone on your staff can write separate programs for the two machine classes, or if your CAM software can route operations by tolerance band. We didn’t have that capability in 2022, and the hybrid order we tried then failed — we ended up re-cutting everything on the new machine anyway, which meant we paid the hybrid premium without harvesting the hybrid savings.

What most people don’t realize: the new machine in a hybrid order does not have to be the flagship model. We bought a mid-tier platform for the precision cell and parked the savings in better fixturing and a climate-controlled corner of the floor. That turned out to matter more for our repeatability than the machine spec did. This is the sort of thing that never makes it into a comparison sheet, because comparison sheets are built around the machine, not around the process that surrounds it.

Four questions to figure out which situation you are in

  1. What is your tightest true-position tolerance, and how often do you run it? Under ±5 μm on more than 20% of your volume — Situation A. Under ±5 μm on less than 5% — you are probably in B or C.
  2. Can you absorb a five-month lead time without losing the order? If not, C is your answer no matter what your budget says.
  3. Do you have in-house expertise to verify a refurbished machine? If you don’t, B is a much more expensive situation than the price tag suggests, and you should budget for an independent inspection before you sign.
  4. What is your annual operating volume on the machine? Under 2,000 hours per year, refurbished makes financial sense almost always. Above 4,000 hours, the parts-availability and service math tilts hard toward new.

One last thing, about sourcing services themselves. The ones worth working with will tell you when a machine is not right for your tolerance band, even when saying so costs them the sale. The broker we now use for all of our standard-tolerance purchases once told us, unprompted, that for our 3 μm work we should go straight to the OEM — they couldn’t guarantee the verification chain on anything they would sell us. That single admission is the reason we have bought three more machines through them since. A sourcing partner who cannot name their own limits is telling you something, and it isn’t good.

I don’t have hard data on the industry-wide first-year failure rate for refurbished wire EDMs. I have heard everything from 3% to 15%, and I don’t fully trust either number — different surveys probably define refurbished differently, which is exactly the problem. What I can say is that in our six wire EDM purchases, the one machine that failed early was the one where we skipped the test cut. That is a sample of one, and I would not build a strategy on it. But I’d also rather tell you it is a sample of one than dress it up as a statistic.