Buy the Boring ATM: New vs. Used and the Hidden Cost of Downtime

If I were buying my first ATM this year, I would buy the most boring machine I could find. Not the cheapest, not the one with the 17-inch video screen. The boring one: the model every repair tech in town already has parts for, that my processor already supports, and that will still pass security rules in three years. Here’s why I think the choice of machine is less about features and more about how many Saturday nights it stays online.

Disclosure: this post contains an affiliate link. If you buy through it, I may earn a commission at no extra cost to you. I only point to things I would use myself. I’m not a financial advisor, and nothing here is financial advice.

Some context on where I’m coming from. I’ve owned vending machines, and I eventually sold that business. I have not run an ATM route. So everything below is how I’d reason through the purchase using what I learned from equipment that earns money only when it works, plus dealer prices and specs I checked in September 2026.

The machine is a commodity. Downtime isn’t.

Most retail ATMs are the same idea: a steel safe, a cash dispenser, a card reader, a PIN pad, and a small computer that talks to a processor. The two brands you’ll see most in small independent routes are Genmega and Hyosung (long sold as Nautilus Hyosung), with Hantle also in the market. New retail models from these makers list at dealers for roughly $2,400 to $2,900. At that level, they’re all good enough.

What they don’t all have equally is an ecosystem near you. When a cassette jams or a card reader dies, what matters is how fast someone can fix it. And that is where real money hides. Let me show you with made-up but realistic numbers.

The downtime math

Say a bar machine averages 8 withdrawals a day and you keep about $2.50 of each surcharge after the bar’s share. That’s roughly $20 a day. Now compare two scenarios over a year:

• Machine A (common model, local parts): down about 4 days a year. Lost income: roughly $80.
• Machine B (cheaper, older, or odd brand; parts shipped in): down about 20 days a year, a few long outages waiting on parts. Lost income: roughly $400.

A $320 difference per year, from one machine, from something that never shows up on the price tag. And that ignores the soft cost: a bartender who sees “out of service” three weekends in a row starts telling customers there’s a gas station down the street. Downtime quietly retrains your customers. If Machine B saved me $300 up front, it pays that back to the universe in about a year and keeps charging me after.

These numbers are hypothetical. Your volume and your venue’s split will be different. The point is the shape: at any location worth having, reliability is worth more than a few hundred dollars of sticker savings.

The short list I’d actually consider

I’m not going to rank ten machines. Here’s how I’d think about the few that come up over and over:

• Hyosung Halo II (about $2,500 to $2,900 from dealers). Compact, 10.1-inch screen, UL 291 business-hours safe and 1,000-note cassette standard, with a 2,000-note option. If I had to pick one without knowing the venue, this is the kind of machine I’d default to. You can see seller listings on Amazon, though I’d compare against an ATM dealer, who usually programs the machine and handles setup.
• Genmega Onyx (about $2,550). Same job, bigger standard screen, lots of cassette options. If my local tech preferred Genmega, I’d buy this and not think twice.
• Genmega G2500 (about $2,400). The budget workhorse. Smaller screen, still does everything a laundromat needs.
• Hyosung Force (about $2,700). I’d only step up for a venue that truly empties a standard cassette between visits.

What I’d skip for a first machine: the big-screen models built to run video ads, which run closer to $3,400 to $3,600. Nobody pays me for the video. And through-the-wall units at $4,500 or more make sense for a landlord-owned storefront, not for a bar where I’m a guest.

New vs. used: a decision rule

This is where I think most people get the math wrong, because a used ATM for $900 looks like a steal next to a new one for $2,700.

The problem is that ATMs get forced into retirement by rules, not wear. Card networks shifted counterfeit-fraud liability to machines without chip readers in 2016 and 2017. Then PCI required a newer encryption key format, called TR-31 key blocks, for ATMs starting January 1, 2025. A machine whose PIN pad or software can’t support that can be refused by the processor. It’s still a perfectly good steel box. It just can’t earn.

So my rule would be simple:

Buy used only if the seller can tell me the exact PIN pad model and firmware, and my processor confirms in writing that this exact machine will run. Otherwise, buy new.

Older units that need an EMV kit can add around $500. Add a new PIN pad and labor, and the “cheap” machine often lands within a few hundred dollars of new, with no warranty and a shorter runway before the next rule change. I’d rather pay the extra once.

Think about the exit on day one

Selling a vending business taught me to look at every asset through a buyer’s eyes. What a buyer wants is simple: equipment they understand, clean records, and locations with agreements. I’d apply that to ATMs from the start. A route built on Halo IIs and Onyxes with current PIN pads is a route another operator can take over without flinching. A route built on a mix of orphaned models is a route a buyer discounts heavily, if they buy it at all.

That’s the freedom angle for me. I don’t buy assets to own them forever. I buy them to have options later: keep them, sell them, hand them to someone else to run. Boring hardware keeps those options open.

The order I’d do it in

1. Find the venue first. Get a real yes from an owner with cash-paying customers. If I were doing this today, I’d use VendBuddy (full disclosure: VendBuddy is my company) (vendbuddy.io/app) to pull bars, laundromats and convenience stores in my ZIP along with the owner’s contact, and I’d buy one credit pack rather than a subscription while I tested the idea.
2. Ask my local ATM tech which brand they stock parts for. That answer probably picks the brand for me.
3. Get processing quotes and confirm the exact model is supported.
4. Size the cassette to the venue’s busiest week.
5. Only then order. Machines sitting in garages are the most expensive machines there are.

If you’re earlier than that, still weighing whether a cash business fits your life, I’d read my honest take on why vending is still underrated and the tradeoffs in vending machines vs. real estate. ATMs rhyme with both.

The detailed operator guide on this exact topic is Best ATM Machines to Buy in 2026: Models, Prices, and Picks, on VendBuddy’s site. It goes further into the specifics than I have room for here.

What I’d tell a friend

Here’s an illustrative example, a made-up person, not a reader or a customer. A friend calls me excited about a $1,000 used ATM on a marketplace. I’d ask three questions. Does your processor accept that exact PIN pad? Where will it go? Who fixes it on a Saturday? If he can’t answer all three, the $1,000 machine is the expensive option. If he can, maybe it’s a great deal. Either way, the machine is the last thing to decide, not the first.

Your move this week: before you look at a single listing, write down three venues where you’ve personally seen people ask “is there an ATM here?” That list is worth more than any spec sheet.