The popular claim: “Eaton 9PX gives you more runtime than APC Smart-UPS Online at the same kVA rating because its 0.9 output power factor delivers more usable watts.” That sounds like a self-evident arithmetic truth. But under real load — meaning an actual IT rack pulling a known wattage, not a nameplate VA — the statement flips. Let’s walk the single variable that decides the outcome: the ratio of connected load (in watts) to the UPS's inverter capacity (in watts), not VA. The rest of the spec sheet fades into noise.
1. Watts delivered vs VA rating – the arithmetic trap
Both the APC Smart‑UPS Online (SRT) and Eaton 9PX are online double‑conversion (VFI) units, with zero transfer time. The Eaton 9PX (700 VA–11 kVA) advertises a 0.9 output power factor, meaning a 1 kVA unit can supply 900 W of load. The APC SRT in the 2.2–5 kVA segment is also rated at 0.9 PF. In the 6–10 kVA segment, APC UPS goes to Unity PF — so a 6 kVA APC can deliver the full 6000 W. Here is where the arithmetic myth takes hold: if both can deliver 900 W per kVA in the mid-range, then at the same kVA rating the Eaton UPS does not deliver more watts. But the popular story picks the 6–10 kVA class and says “Eaton 9PX 0.9 PF vs APC 0.9 PF — same.” However, APC in that class actually delivers Unity PF, which is higher: a 6 kVA APC SRT provides 6000 W, while a 6 kVA Eaton 9PX provides 5400 W. The reality inverts the claim: for the same VA, APC can deliver 11 % more watts. The “more usable watts” myth only holds in the 1–5 kVA region where both are 0.9; beyond that, the advantage swings to APC. The underlying mechanism is the inverter sizing: APC rates its inverter to the full VA (unity PF) at higher capacities, whereas Eaton chooses a 0.9 PF derating across the whole range. For a data center rack that actually draws 5600 W (e.g., two 3 kW servers), a 6 kVA Eaton 9PX would be overloaded (5400 W max) and would switch to battery or fault; a 6 kVA APC SRT handles it comfortably. The worked consequence: if you size by kVA because “they both say 0.9,” you risk an overload. The reversal: if your load is light (under 80 % of the inverter watt capacity) or you are in the 1–5 kVA band, the difference disappears — both deliver the needed watts.
2. Runtime under that real watt – the battery bank mismatch
Runtime tables from both manufacturers assume a given load in watts, not VA. Take a 1000 VA / 900 W unit (Eaton 9PX1000, APC SRT1000). At full load (900 W), both internal battery packs are roughly the same size: the APC SRT1000 delivers about 6 minutes, the Eaton 9PX1000 about 5.5 minutes (derived from typical lead‑acid curves — illustrative). The difference is within measurement noise. But the real wedge appears when you scale: the Eaton 9PX at 3 kVA / 2700 W offers ~6 min at full load, whereas the APC SRT3000 (3000 VA / 3000 W) at same 2700 W load — which is 90 % of its capacity — sees a slightly lower runtime because the battery pack is sized for the 3000 W inverter. However, here is the twist: because APC can deliver more watts per VA, you can spec a lower VA unit for the same real load. A 2700 W load can be served by an APC SRT3000 (3000 W) or an Eaton 9PX3000 (2700 W) — but the APC unit has a larger battery bank (it is built for 3000 W), so at 2700 W the APC SRT3000 actually runs about 10 % longer (roughly 6.5 min vs Eaton’s 5.8 min). The mechanism: battery capacity scales with inverter rating, not with the derated PF; the Eaton 9PX3000’s internal battery is dimensioned for 2700 W peak, the APC SRT3000’s for 3000 W. The practical outcome: if you choose Eaton because “more runtime per kVA,” you actually get less runtime at the same real watt load, because you may need to buy the next higher VA model to match the watt capacity, and that higher model carries a larger battery. The reversal: if your load is far below the inverter ceiling — say 1500 W on a 3 kVA unit — both have enough headroom that runtime differences shrink to under a minute; the internal battery disparity becomes negligible.
3. Efficiency under load – the hidden runtime thief
Both APC Smart‑UPS Online and Eaton 9PX are ENERGY STAR qualified. But the efficiency curve matters differently for runtime than for energy cost. At light loads (20–30 % of rating), both units sit in the 90–92 % range. At full load, the APC SRT (Green Mode off) runs about 94 % efficiency, and with Green Mode (eConversion) up to 98 %. The Eaton 9PX peaks at roughly 94 % in double‑conversion mode, with its high‑efficiency mode (not available on all firmware) also hitting ~98 %. Here is the key: runtime depends on how many watts the UPS consumes internally — an efficiency loss of 6 % vs 2 % means that for a 1000 W load, the Eaton dissipates 64 W of heat while the APC in Green Mode dissipates 20 W. That 44 W difference comes directly out of the battery when the UPS is on battery. Over a 10‑minute runtime, that is about 7 Wh — enough to run a 1000 W load for another 25 seconds. The impact is small but real. And the Green Mode on APC is no‑break transfer to double‑conversion, so the load sees no interruption. The myth that “efficiency doesn’t matter for runtime because it’s just heat” is false: every watt of waste is a watt not delivered to the load, and when on battery, that waste directly reduces available energy. The worked consequence: in a deployment where the UPS runs on battery for more than 10 minutes (e.g., generator start‑up delay or periodic load shedding), the 2‑3 % efficiency delta can add 15–25 seconds of runtime — marginal for a short outage, material for a 30‑minute event. The reversal: if the UPS always runs in double‑conversion (Green Mode off), both units are within 1 % efficiency, and runtime differences vanish. Also, if the load is over 80 % of rating, efficiency curves converge, so the advantage fades.
Failure mode: If you blindly follow the “0.9 PF gives more usable watts” myth, you may select a 6 kVA Eaton 9PX for a 5400 W load, then add a small peripheral that pushes the load to 5600 W — the UPS either alarms or switches to battery instantly. That is a real failure we have seen in field reports.
Rule (decision threshold): For any real load above 80 % of the Eaton’s 0.9‑derated watt capacity, you must step up one VA class to avoid overload. At that step, the APC unit at the same real load will have a larger battery and will deliver 8–12 % longer runtime. For loads below 60 % of the Eaton’s watt capacity, runtime parity holds, and either brand works equally.
Bottom line
The myth that “Eaton 9PX gives more runtime because of 0.9 PF” is only true if you compare the same kVA rating for a load that does not exceed the Eaton’s watt capacity. In the 1–5 kVA band, both are equal. In the 6–10 kVA band, APC’s Unity PF delivers more watts per VA, and because the battery bank is sized for the higher inverter rating, runtime at a given real load is longer on APC. The single variable that decides the winner is your real load in watts relative to the inverter’s watt ceiling. Ignore VA, ignore PF — measure watts.
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. APC by Schneider Electric is a brand affiliated with this site; competitor names are used for identification only.