Here’s the question I’d ask before anyone buys an ATM: if you have to lock $3,000 of your own cash inside a steel box to earn $300 a month, what are you actually earning? Most “how much do ATMs make” answers stop at the monthly number. I think the monthly number is the least interesting part. The better question is what that income costs you in capital and in hours, and how it compares to what the same money and time could do somewhere else.
So that’s what this post is. Not a pitch. A way to think about an ATM as a small investment, with the math laid out so you can plug in your own numbers. I’m not a financial advisor and nothing here is financial advice; it’s how I’d reason through it.
An ATM is a tiny bank that rents out access to your cash
When I owned vending machines, the product was snacks. With an ATM, the product is access to cash, and the inventory is your own money. A customer puts in a card, agrees to a fee on the screen, and walks away with your twenties. The processor pays you back the withdrawn amount, usually the next business day by bank transfer, plus the fee.
That means two separate things are happening:
- Your cash is rotating. It leaves the box, comes back to your account, and goes back in the box. It is never spent, but it is never free either.
- You are collecting a fee per rotation. Independent machines typically charge somewhere between $2.50 and $3.50. For perspective, Bankrate’s 2025 study had bank-owned ATMs averaging a record $3.22, with people’s own banks piling on another $1.64 or so. People are used to paying close to $5 to get their own money out of a machine that isn’t their bank’s.
There’s a little extra called interchange, a few cents to a quarter per withdrawal that flows back through the card networks, but I’d never build a plan on it.
The only number that matters: net per month divided by everything you put in
Here’s how I’d set it up. Every placement has two kinds of capital in it: the machine (call it $2,500 for a decent used or entry-level unit) and the float (the cash you keep inside, often $2,000 to $5,000 depending on how busy the spot is). Then there’s the monthly net: withdrawals times the fee, minus whatever the location owner takes, minus wireless, processing and repairs.
Let me run three hypothetical placements. These are not my numbers and not anyone’s real results. They’re built from commonly published ranges: quiet spots doing 30 to 80 withdrawals a month, a planning average around 150 to 180, and busy bars doing a few hundred.
| Hypothetical placement | Capital in (machine + float) | Monthly net | Yearly net | Cash-on-cash, pre-tax |
|---|---|---|---|---|
| Quiet laundromat, no split | $2,500 + $2,000 = $4,500 | ~$114 | ~$1,368 | ~30% |
| Convenience store, 25% split | $2,500 + $3,000 = $5,500 | ~$323 | ~$3,876 | ~70% |
| Busy bar, 30% split | $2,500 + $4,000 = $6,500 | ~$815 | ~$9,780 | ~150% |
Those percentages look wild next to an index fund, and I want to be really careful here, because this is exactly where people fool themselves. Three things that table does not show:
- The machine loses value. The float comes back to you when you quit. The $2,500 machine mostly doesn’t. Part of that “return” is just getting your machine money back over time.
- Your hours aren’t in it. A self-loaded ATM takes something like 2 to 4 hours a month: bank run, counting, driving, loading, balancing, clearing a jam. That’s labor. You’re paying yourself for it out of the net.
- It assumes you found the good spot. The table starts after the hardest part. A machine sitting in your garage while you hunt for a location earns 0% on everything.
Now put your hours back in
This is the part I’d push hardest on, because it’s the part almost every income screenshot leaves out. Monthly income feels great until you divide it by the time it took.
Take the convenience store. Say it takes 3 hours a month. If you decide your time is worth $30 an hour, that’s $90 of labor out of the $323, leaving about $233 a month as the true return on capital. On $5,500 that’s still about 51% a year before taxes. Very good, if it holds.
Now the laundromat. Take $90 of labor out of $114 and you’re left with $24 a month. On $4,500 that’s about 6% a year, and you’re carrying cash around town to get it. You’d be close to what a high-yield savings account or a Treasury bill pays, without the hassle or the risk. That one isn’t a business. It’s a hobby with a cassette.
This is the same lens I wrote about in evaluating dollar per hour: some work looks profitable until you price your own time, and then it doesn’t.
The iceberg is melting, slowly
The other honest thing: fewer people pay with cash every year. The Federal Reserve’s 2026 Diary of Consumer Payment Choice found consumers still used cash for about 1 in 7 payments in 2025, and most people still carry a little. That’s not zero, and it’s concentrated in exactly the places ATM operators like: bars, laundromats, flea markets, cash-only shops. But it means I would never model an ATM as growing. I’d model it as flat at best, drifting down over years.
That changes the comparison. With an index fund, I’m betting on a long-run upward drift (with no guarantees, and some ugly years along the way; I ran that math in compound interest is boring until you run the numbers). With an ATM, I’m collecting high cash yield today from an asset whose demand is slowly shrinking. Those are different animals. Neither is wrong. Just know which one you’re buying.
Where it gets interesting: stacking
One ATM won’t change your life. What makes this model worth thinking about is what happens when you stack good placements close together, because the hours don’t scale linearly. One reload trip can service three machines on the same street.
Picture a hypothetical operator, call him Marco (illustrative, not a real person), who keeps his day job in sales. His first machine goes into a liquor store and nets around $300 after the owner’s cut. He sends every dollar of it at his car loan. Eight months in he has three machines on the same stretch of road, and the car is paid off early. Two years in he has seven, averaging about $300 net each, so roughly $2,100 a month before taxes, for maybe 20 hours of work a month. That’s not quit-your-job money. It’s “the job becomes a choice sooner” money, which is the kind I care about.
But look at the other side of Marco’s ledger. Seven machines at roughly $2,500 each plus about $3,000 of float each is around $38,500 of capital. One of his placements flopped and got moved. He spends Saturday mornings counting twenties. That’s the real deal: meaningful income, bought with real capital and real hours. Anyone who shows you only the first half is selling something.
Who I think this fits
Honestly? Someone who already has some savings they’re comfortable tying up, lives in an area with cash-heavy businesses, doesn’t mind a weekly route, and wants income that isn’t tied to their job. Someone who’d rather own something than just draw a paycheck, which is the whole theme of never rely on just one job.
Who it doesn’t fit: someone using money they’ll need in the next year, someone who hates handling cash, or someone hoping it’ll be fully passive. It’s low-touch. It isn’t no-touch.
There’s a more hands-on companion to this piece on VendBuddy (full disclosure: VendBuddy is my company): How Much Do ATM Machines Make? Surcharge Math by Location (2026). Start there if you’re already past the “should I” stage.
The next step, if you’re curious
Don’t buy a machine first. Find the spot first. Make a list of every bar, laundromat, convenience store and cash-heavy shop within 15 minutes of your house, cross off anything next to a bank, and figure out who actually owns each place. VendBuddy (vendbuddy.io/app) does that list for any ZIP and includes the owner’s contact info; you can buy a single credit pack instead of subscribing if you just want one batch of leads.
Then run the table above with your own inputs, and be brutal about your hours. If the math still works after that, it might be worth doing.
