NAS vs cloud — the cost conversation

A client showed me a bank statement with eleven separate subscriptions on it. Four were for storage: a cloud drive for the company, a photo plan on somebody's personal account, a backup service nobody could remember signing up for, and an extra tier on a mail provider bought years ago to stop a mailbox filling up. Individually every one of them was small. That is the entire business model of a subscription, and it works.

He did not want a lecture, he wanted a number. So we built the only thing that settles this kind of argument: a spreadsheet with both columns filled in honestly, including the entries that make my own side look worse.

The shape of the two curves

Strip out the numbers and you are left with two shapes, which is the part that transfers between situations.

A subscription is a straight line from the origin. Nothing to pay at the start, a constant amount every month, forever, and the total grows without limit. It is also a line that tilts upward over time — prices rise, tiers get restructured, and the amount of data you keep only ever increases, which moves you into higher bands.

Owned hardware is a step followed by a nearly flat line. A large payment once, then electricity and the occasional replacement drive. The line is not perfectly flat — drives fail, hardware eventually gets replaced — but the slope is small enough that over a few years the two lines cross, and after the crossing point the gap widens every month.

Everything therefore depends on where the crossing is, and the honest answer is that it depends on two variables only: how much data you keep, and how long you intend to keep it. Everything else is noise.

The entries people leave out

Both sides cheat, and a comparison is only worth anything if you stop them.

On the subscription side, the number people quote is usually the current tier for one service. The honest number is every storage line on the statement added together — which for my client was four — plus a realistic view of what the tier will be in three years given that nobody ever deletes anything. Add the cost of leaving: how long it takes to get your data out, and what it costs in time and bandwidth if you ever want to.

On the hardware side, the number enthusiasts quote is the price of the box. The honest number adds the drives — which is frequently more than the machine — and then a second set of drives, because one copy on one device is not a backup and pretending otherwise is how this goes wrong. Add electricity for continuous running. Add your own time, priced at something, for updates and the Saturday when something breaks.

Where each side actually wins

Having done this exercise with a fair number of small companies, the pattern is consistent enough to state as a rule.

Subscriptions win decisively when the quantity is small and the requirement is uncertain. A few tens of gigabytes of documents, a business that may look quite different in two years, nobody in the building who wants to be responsible for a machine — pay the monthly amount and stop thinking about it. That is not a defeat, it is the correct answer, and I say so to clients regularly.

Hardware wins decisively on volume and on time. Photographs, video, scanned archives, anything measured in terabytes that you are legally or sentimentally obliged to keep for a decade — that is the workload where subscription pricing compounds against you and the crossing point arrives fast. A photographer, an architect's office, a company with a long document retention obligation: for those, the arithmetic is not close.

The middle is genuinely ambiguous and I distrust anybody who says otherwise. For a five-person office with a moderate amount of data, the two columns land close enough that the decision should be made on something other than money — on who will maintain it, on whether the data is sensitive, on whether the company wants that responsibility.

The variable nobody prices

One column in that spreadsheet cannot be filled in, and it is the one I would argue matters most over a long horizon: what happens when the terms change.

A subscription price is not a fact about the world, it is a decision somebody else makes annually. Tiers get restructured, free allowances shrink, a plan you have had for years is discontinued and migrated into a more expensive one. None of this is malicious and all of it is normal. But it means the straight line in your spreadsheet is drawn with somebody else's ruler.

A machine you own has a price that was fixed on the day you bought it. It has other risks — it can fail, be stolen, or burn — and those risks are real and manageable with a copy elsewhere. What it does not have is a counterparty who can change the deal. For some businesses that is worth nothing. For the ones I mostly work with, storing decades of client work, it turns out to be the deciding line in the spreadsheet, once somebody points out that it belongs there.