How to Choose an Online UPS: Why the Cheapest Quote Is Almost Always the Most Expensive Option

If you're comparing online UPS systems by price-per-VA, you're going to overpay.

I don't care what the spec sheet says. I've been a procurement manager at a 250-person industrial engineering firm for six years now. During that time, I've managed our electrical infrastructure budget — roughly $180,000 annually — and I've personally reviewed quotes from more than 20 switching power supply manufacturers and UPS vendors, ranging from large multinationals to specialist modular UPS manufacturers for data center applications.

When I first started, I assumed the vendor with the lowest price per VA won. Every time. I downloaded the quotes, found the cheapest line item, and pushed it through.

That approach cost us. Two battery replacements, one near-miss capacity crisis, and a TCO spreadsheet later, I finally understood something that should have been obvious: for an online UPS three-phase 100kVA system, the lowest quoted price is almost always the highest total cost over 10 years.

Here's why — and what I do instead.

Argument 1: You cannot ignore battery replacement costs

Let me give you a concrete example from a purchase we made in Q2 2022.

We needed a new 100kVA three-phase online UPS for our second data hall. Two vendors made the shortlist:

• Vendor A: $18,500 — valve-regulated lead-acid (VRLA) batteries, 94% efficiency, 3-year warranty
• Vendor B: $22,800 — VRLA batteries, 96% efficiency, 5-year warranty, annual maintenance included

I almost approved Vendor A on the spot. A $4,300 gap seemed like a clear win.

But our TCO spreadsheet said wait. What about battery replacement?

VRLA battery strings in a 25-30°C environment typically need replacement every 3-4 years. For a 100kVA system, a full string replacement runs $6,000 to $9,000 depending on battery quality. I checked with our facilities team — that matched what they'd seen across three sites.

Here's what the 10-year math actually looked like:

Vendor A (the "cheap" option):

• System: $18,500
• Battery replacement #1 (Year 3.5): $7,500
• Battery replacement #2 (Year 7): $7,500
• Maintenance (self-performed, no service contract): ~$1,200/year × 10 = $12,000
• Cooling overhead (94% efficiency means more heat in the room): roughly $1,500 over 10 years
• 10-year total: ~$47,000

Vendor B (the "expensive" option):

• System: $22,800
• Battery replacement #1 (Year 5): $7,500
• Battery replacement #2 (Year 10): $7,500
• Maintenance: included
• Cooling overhead (96% efficiency): roughly $900 over 10 years
• 10-year total: ~$38,700

Vendor B was $4,300 more expensive on paper and $8,300 cheaper in reality.

That hidden difference — it's not a small gap. It's the kind of gap that gets noticed when you're reporting to a CFO.

Argument 2: Modular UPS wins on total cost, not just on paper

Here's a misconception I encounter constantly: "Modular UPS is only for large data centers."

Wrong. It's for anyone who wants to stop paying for capacity they don't use.

When you buy a monolithic 100kVA UPS, you're buying a fixed block of capacity. If your actual load runs at 60kVA, you still paid for 100kVA. And that system may run less efficiently at 60% load than at its sweet spot.

In early 2023, we evaluated both options for a facility expansion:

• Monolithic 100kVA: $22,000 (4 VRLA strings)
• Modular 100kVA chassis + two 25kVA modules: $28,500 (VRLA, 5-year warranty)

That $6,500 gap looked steep. But here's what changed the picture: we weren't fully loading the system for 18 months.

With the monolithic unit, we'd have been running at roughly 45% load for over a year — a range where efficiency drops 3-5% below the rated number. We'd have been paying for 100kVA while effectively burning the efficiency penalty of a much smaller load.

The modular system let us start with two 25kVA modules, add a third when the load grew, and keep the load factor in that 40-70% sweet spot where online UPS systems actually perform best.

And there's another factor: mean time to repair (MTTR). With a monolithic UPS, a failure means downtime measured in hours. With hot-swappable modules, we're talking minutes. For a facility running production workloads, that difference has a cost attached to it — and it's not a small one.

If I were evaluating a lithium ion UPS system for industrial deployment today, I'd apply the same logic: don't just look at the upfront cost. Look at the cycle life, the replacement schedule, and the efficiency curve.

Argument 3: The highest efficiency rating is not the same as the lowest TCO

This one surprised me.

When comparing efficiency specs, most buyers look for the highest percentage. 96% beats 94%, right?

For online UPS systems, the efficiency curve matters more than the rated peak. A 100kVA UPS might hit 96% efficiency at 100% load — but if your actual load runs between 20% and 60%, you might be operating in the 88-90% range.

According to our monitoring data across three sites, online UPS systems tend to perform best between 40% and 70% load. Below 40%, efficiency falls off a cliff.

So if you buy a 100kVA UPS and only load it to 35%, you're paying an efficiency penalty every month — one that never shows up on the invoice.

This is the real argument for modular: you buy the capacity you need, you run it in the efficient zone, and you add modules when demand grows. It's not about being fancy. It's about not wasting money on electricity you convert into heat instead of backup power.

What about the counterarguments?

"Modular takes up more floor space."

True — in our case, the modular chassis used about 15% more floor area than the monolithic unit. If space is a hard constraint, that's a real consideration. But before you dismiss modular on space, calculate the annual efficiency loss from running a monolithic system at partial load. In our case, that penalty would have paid for the extra square footage in Year 2.

"Lithium-ion is too expensive upfront."

Our 2024 quotes showed lithium-ion systems running 2-3x the upfront cost of VRLA. But over 10 years:

• VRLA: 3 replacements at ~$7,500 each = $22,500
• Lithium-ion: effectively zero replacements (in a temperature-controlled room, 10-year expected life is realistic)
• Plus: lithium-ion produces 15-20% less heat during charge/discharge, which reduces cooling load

Our breakeven calculation landed at roughly Year 8-9. If you expect the system to run longer than that, the math shifts in lithium-ion's favor. I'm not saying everyone should rush to lithium-ion. But if you're still using a 5-year replacement cycle on VRLA batteries, you're probably leaving money on the table.

"My management won't approve the modular premium."

I've been there. What worked for us was a procurement policy change: any electrical equipment purchase over $50,000 now requires a TCO comparison, not just an upfront quote comparison. Once you show the 10-year picture side by side, the conversation changes.

Let me restate my point

If your process for choosing an online UPS is "find the lowest price per VA," you are almost certainly overpaying.

Total cost of ownership — not sticker price — is the right question.

The questions I ask now before signing any UPS purchase order:

1. What does the 10-year TCO look like, including battery replacements and maintenance?
2. What is the efficiency curve at 30%, 50%, and 70% load — not just at rated load?
3. What's the MTTR, and how much does one hour of downtime cost us?
4. What's the growth path if our load doubles in three years?

When I ran those numbers on our last purchase, the option that looked 20% more expensive upfront ended up being 23% cheaper over its lifecycle.

That's not just a better deal. That's the whole job.