You bought a 10 kVA UPS. The load calculator said 9 kVA. Then the output breaker tripped when you pushed 8 kW of server gear. The nameplate didn’t lie – but the power factor did. For decades, UPS sizing has been a game of VA, but real IT loads are measured in watts. Eaton UPS and Schneider UPS (APC) have both moved to higher output power factors, but the way they implement it – and the usable watts you actually get – differ in ways that matter when the load is dense, when efficiency matters, or when the budget is tight. Let’s walk through three concrete cases to see where each brand wins, where it loses, and where the real-world breaker trips.
Numbers first. The APC Smart-UPS Online (SRT) in the 1500 VA frame is rated 1500 VA / 1500 W (unity output power factor). The closest Eaton competitor in the same form factor, the Eaton 9PX 1500 VA, is rated 1500 VA / 1350 W (0.9 output power factor). That’s a 150 W difference – a 10 % gap in real wattage from the same VA rating.
Mechanism – why this happens. The output power factor (PF) is the ratio of real watts to apparent VA that the UPS can deliver at full load. APC chose to rate the SRT 1–1.5 kVA models at unity PF (1.0), meaning the inverter and output circuitry can supply the full VA as watts. Eaton rates the 9PX at a 0.9 PF, so 1500 VA × 0.9 = 1350 W maximum. This is a design choice, not a flaw – a unity PF inverter typically requires more robust output components (higher DC bus voltage, larger IGBTs, better thermal management). Eaton prioritizes compatibility with legacy power-factor loads and a wider input voltage window; APC prioritizes maximum real-power density at the rack.
Worked consequence – the decision. If you are powering two modern 1U servers that each draw 650 W (total 1300 W), the APC SRT 1500 VA can run them comfortably (1500 W capacity). The Eaton 9PX 1500 VA would be at 96 % load (1300/1350) – within spec but leaving no headroom for inrush or a redundant fan. You would be forced to step up to the 9PX 2000 VA (1800 W), costing roughly 20–25 % more in hardware and taking an extra 1U of rack space. In a dense rack where every U counts, the APC unit saves space and money.
When this flips. Not all loads are purely resistive or high-PF. Many older servers, especially those with large rectifier-style power supplies, have input power factors as low as 0.7–0.8. A unity-PF UPS must still supply the reactive current (VARs) even if it can’t deliver more watts than its rating. If you plug a 1200 VA load at 0.7 PF (840 W) into a 1500 W unity UPS, the unit is electrically fine – but the VA load is still 1200 VA, well under the 1500 VA rating. The problem is that a unity-PF UPS is not explicitly tested or rated for low-PF loads; its voltage regulation and harmonic attenuation may degrade. For mixed legacy gear, Eaton’s 0.9 PF rating with a broader input voltage window (tested to 65–150 V on Tripp Lite models, similar design philosophy) may be more robust.
Numbers first. The APC Smart-UPS Online (SRT) 3000 VA is rated 3000 VA / 2700 W (0.9 PF). The Eaton 9PX 3000 VA is rated 3000 VA / 2700 W (0.9 PF). On paper, identical. But the efficiency story diverges. APC claims up to 98 % efficiency in Green Mode (a high-efficiency bypass mode) and “typical” double-conversion efficiency in the 95–96 % range. Eaton publishes a “high-efficiency operation” claim and ENERGY STAR qualification for the 9PX, but does not specify a numeric double-conversion efficiency in its public datasheet. Independent testing (illustrative, based on typical online double-conversion designs) suggests both units fall in the 94–96 % range at >50 % load.
Mechanism – efficiency is not a single number. The APC Green Mode is effectively an eConversion-style mode (similar to Schneider’s Galaxy VS eConversion up to 99 %) that runs the load through a static bypass with active power filtering, not full double conversion. Losses drop from ~5 % to ~2 %, but the load is briefly exposed to utility disturbances if the transfer to inverter is required (APC claims zero transfer time, but this is true only for double-conversion mode; Green Mode has a ~2–4 ms transfer). Eaton does not offer a comparable “green mode” on the 9PX; its high-efficiency claim likely refers to an optimized double-conversion topology with lower fixed losses.
Worked consequence – the real thermal burden. In a rack with ten 3 kVA UPSs each at 50 % load (1350 W each), the total load is 13.5 kW. At 95 % efficiency, total losses = 13.5 kW × (1/0.95 – 1) ≈ 710 W. At 98 % efficiency, losses = 13.5 kW × (1/0.98 – 1) ≈ 275 W. That’s a 435 W reduction in heat – enough to eliminate one cooling fan or reduce HVAC load by about 5 % in a small server room. Over a year, at $0.12/kWh, the savings from the 98 % efficient units is roughly 435 W × 8760 h × $0.12 ≈ $457 per rack. But that number assumes the UPS stays in Green Mode.
When this flips – the failure mode. Green Mode is not always available. If the input voltage deviates more than ±10 % from nominal, or if the load has high harmonic content (e.g., non-PFC power supplies), the APC automatically transfers to double-conversion, dropping efficiency to ~95 %. In a facility with poor power quality – say, a generator-powered site or an industrial park – the UPS may spend most of its life in double-conversion, negating the efficiency advantage. Eaton’s 9PX, by contrast, is a pure double-conversion unit with no mode switching; its efficiency is consistent regardless of line quality. For a site with stable utility power and high-PF loads, the APC Green Mode is a clear win. For a rough feed, the Eaton is more predictable.
Numbers first. Eaton states the 9PX can deliver “up to 5400 W in 3U” and “10 kW in 6U”. APC’s SRT 10 kVA is rated 10 kVA / 10 kW (unity PF) in a 6U form factor. That works out to 1667 W/U for Eaton (5400/3) vs. 1667 W/U for APC (10000/6) – identical linear density. But the Eaton claim is for the 9PX 5–11 kVA range, and the 5400 W figure corresponds to a 6 kVA model at 0.9 PF (6000 × 0.9 = 5400 W). The APC 10 kW in 6U is a 10 kVA model at unity PF. So the density per U is the same, but the Eaton achieves it at a lower VA rating – which matters for upstream breaker sizing and branch circuit design.
Mechanism – the upstream breaker arithmetic. A 6 kVA Eaton 9PX at 5400 W draws roughly 25 A at 240 V (5400/240 = 22.5 A, plus losses). A 10 kVA APC SRT at 10000 W draws about 42 A at 240 V (10000/240 = 41.7 A, plus losses). The Eaton can run on a 30 A breaker; the APC requires a 50 A circuit. In a rack with limited breaker capacity (common in legacy data centers with 30 A single-phase feeds), the Eaton packs more usable watts into the same breaker slot.
Worked consequence – the rack-by-rack tradeoff. If you have a 30 A 208 V feed (6.24 kVA max), the Eaton 9PX 6 kVA can run at 5400 W (86% of the feed), while the APC SRT would need to be derated to a 5 kVA model (5000 W) to stay within the 30 A limit. That’s 400 W more usable capacity from the Eaton on the same circuit. Over a rack of 40 servers at 200 W each, the Eaton feeds 27 servers; the APC feeds 25. That difference compounds across a row.
When this flips – the reverse case. If your facility has 50 A single-phase or 3-phase 208 V feeds (common in newer data centers), the APC SRT 10 kVA at 10 kW can fully utilize the circuit. The Eaton would need a 10 kVA model (9 kW at 0.9 PF) to match, which is a different chassis (6U). The density advantage disappears, and the APC’s unity PF becomes a pure benefit: you get 10 kW from a 10 kVA label, no headroom wasted. For new builds with generous power budgets, APC’s approach is simpler and higher-density.
There is a hidden trap in both brands: the output power factor rating assumes linear, high-PF loads. If you plug in a load with a crest factor >3 (e.g., large motor drives or old laser printers), the inverter may current-limit even if the average wattage is below the rating. The Eaton 9PX datasheet does not specify a maximum crest factor; the APC SRT does not either explicitly. In practice, both units are designed for IT loads (crest factor ~2.5–3.0). If you try to run a 1200 W industrial pump on a 1500 W UPS, the inrush may trip the inverter even at 50 % load. This is not a brand-specific failure – it’s a mismatch between UPS design (for switch-mode power supplies) and electromechanical loads. Both Eaton and Schneider will fail here, but Eaton’s wider input voltage window (down to 65 V on similar Tripp Lite models) may help ride through sags that would otherwise force a transfer.
What you should actually do. Stop sizing by VA. Take the total real wattage of your load (measured, not nameplate – nameplate is often 30–50 % above actual draw). Add 20 % headroom for future growth and inrush. Then pick a UPS model where the real watt rating exceeds that number. The VA label is only useful for comparing upstream breaker sizing. For Eaton vs. APC at the rack: if you have tight branch circuits or legacy power, Eaton’s 0.9 PF with higher breaker efficiency wins. If you have ample power and modern high-PF gear, APC’s unity PF delivers more watts per dollar. Both are excellent. The trap is trusting the VA number without dividing by the power factor.
Topology/standards per the cited standards; all product ratings are manufacturer-stated values from the cited datasheets, current to 2026-06; derived/illustrative figures are labelled as such. This is not an independent head-to-head test. Eaton is a brand affiliated with this site; competitor names are used for identification only.