I earn commissions when you sign up or buy through the affiliate links on this page, at no extra cost to you.

Homelab · Guide · By Mohammed Almuhanna · Updated June 2026

What Size UPS Do I Need?

I will be straight with you. I do not run a UPS myself, because they are stupidly expensive here in Saudi Arabia and our grid is stable enough that I have never been burned by it. But if your power flickers or cuts, you want one, and sizing it comes down to two numbers: how many watts your gear draws, and how many minutes you want it to keep running. My flat opinion before you spend a cent is to size a home UPS to ride out flickers and shut the server down cleanly, not to run the rack for hours. People chase runtime they will never use and pay for it in battery. Get the watts right, pick a runtime target you actually need, and the rest of the choice falls out from there.

If you just want your number, the calculator does the math both ways.

Add up your load in watts

List every device the UPS has to carry and add up their real running watts. For a typical homelab that means the server or NAS, the network switch, the router, and the modem. A small NAS might pull 30 to 60 W, a busy server 100 to 250 W, and switches and routers usually 10 to 30 W each.

Measure with a plug-in power meter at the wall rather than trusting nameplate ratings. The label on a power supply states its maximum rated draw, not what the box actually uses, so it overstates real consumption, often by a wide margin. Once you have the total, add headroom. Aim to load a UPS to about 50 to 80 percent of its rating, not 100. That keeps the unit running efficiently, leaves room for spin-up surges, and is easier on the battery over its life.

VA vs watts

UPS units carry two ratings, VA and watts, and the difference trips up a lot of buyers. VA (volt-amps) is apparent power. Watts is real power, the figure that actually matters for your load. The two are linked by the power factor:

watts = VA × power factor

Older units often have a power factor around 0.6, so a 1500 VA unit may only handle about 900 W. Newer units sit around 0.9 to 1.0, where VA and watts are close. Size to the WATTS rating against your measured load, because watts is the limiting number. Never assume VA equals watts, or you can buy a unit that looks big enough on the box and trips under real load.

How much runtime do you need

This is where most people overspend, so be honest about the goal. A home UPS is for riding out brief outages and giving the server enough time to shut down cleanly. It is not meant to run your rack for hours through a long blackout. Most setups want enough margin to cover a flicker or a short cut, plus a couple of minutes for an orderly shutdown.

Pick a concrete target, for example 5 to 15 minutes at your load. This matters because runtime is what drives battery size and cost, and it climbs steeply. Doubling the runtime costs far more than double, so set a target you actually need rather than padding it for comfort.

Runtime, battery, and the math

Runtime depends on the battery's stored energy versus your load. Battery capacity in watt-hours is volts times amp-hours:

watt-hours = battery volts × amp-hours

A 12 V, 9 Ah battery holds about 108 Wh on paper. Divide usable watt-hours by your load in watts for a rough runtime in hours, then subtract inverter losses and the headroom you keep so the battery never fully drains. A 100 W load on that battery lands in the area of 30 to 45 real minutes, less as the battery ages.

To go longer you need a bigger internal battery or external extension packs, and cost climbs fast. Keep the target modest unless you need long runtime, and reach for a small generator instead if you are trying to survive outages measured in hours.

Line-interactive vs online, and pure vs simulated sine wave

Line-interactive units are cheap and perfectly fine for most homelabs. They regulate minor voltage swings and switch to battery quickly when power drops. Double-conversion online units run the load off the inverter continuously so the load never sees the switch, but they cost more and run hotter, so save them for sensitive or critical gear.

Pay attention to the output waveform. Prefer a pure sine wave for modern power supplies. Many server and PC supplies use active PFC, and active-PFC supplies can misbehave on a stepped or simulated sine wave, ranging from buzzing to dropping the load at the moment of transfer. A pure sine wave output avoids that whole class of problem.

Do not forget clean shutdown

A safe shutdown needs more than a battery, and it matters more than the runtime number. Connect the UPS to the server with its USB data link and run software that watches it, either NUT or the vendor's own tool. When the battery runs low, the software tells the server to shut down gracefully before power is lost, so you never risk a hard cut to a NAS mid-write. A UPS without that link only buys you time you may not be awake to use.

Use the UPS calculator to estimate runtime for a given UPS, or to size the UPS and battery for a target runtime.

Common questions

What is the difference between VA and watts on a UPS?

VA (volt-amps) is apparent power and watts is real power, the figure that matters for your load. They are linked by the power factor: watts = VA × power factor. Older units sit around 0.6, so a 1500 VA unit may only handle about 900 W, so always size to the watts rating against your measured load.

How big a UPS do I need for a homelab?

Add up the real running watts of everything the UPS carries, the server or NAS plus the switch, router, and modem, then load the UPS to about 50 to 80 percent of its watts rating rather than 100. Measure with a plug-in meter at the wall, since nameplate ratings overstate real draw.

How much runtime should I aim for?

A home UPS is for riding out brief outages and giving the server enough time to shut down cleanly, not running the rack for hours. A target of about 5 to 15 minutes at your load is sensible. Runtime drives battery size and cost, and doubling it costs far more than double.

Do I need a pure sine wave UPS?

For modern PC and server power supplies, yes, prefer one. Many supplies use active PFC, which can misbehave on a stepped or simulated sine wave, ranging from buzzing to dropping the load at transfer. A pure sine wave output avoids that whole class of problem.