You’re staring at two spec sheets: one says “1500 VA / 1350 W”, the other says “1500 VA / 1500 W”. Same VA, different real watts. That 150‑W spread doesn’t sound huge — until you scale it across a rack of servers that each draw 350 W measured at the PDU. Then the discrepancy becomes a capacity cliff. Let’s tear down where Eaton UPS and Schneider UPS (APC) diverge on real‑watts sizing, and why the ratio (output power factor) changes the whole procurement math.
Numbers. Eaton 9PX (double‑conversion) is rated at 0.9 output power factor across 700 VA – 11 kVA. The APC Smart‑UPS Online (SRT) family uses 0.9 PF on 2.2–5 kVA models, but on 1–1.5 kVA and 6–10 kVA units it allows unity (1.0 PF). That means a 10 kVA SRT can deliver 10,000 W, while a 10 kVA Eaton 9PX delivers 9,000 W — a 1,000‑W difference at the same VA rating.
Mechanism. Output power factor expresses how many real watts the inverter can deliver at rated VA. Unity PF means the inverter is sized to supply the full VA as watts; 0.9 PF means 90 % of VA. This is not a “better or worse” — it depends on whether the inverter’s IGBTs and magnetics are rated for the crest factor and peak current of a 0.9‑PF load. APC’s SRT 6–10 kVA units are built with a larger DC bus and higher current capacity to support unity PF.
Worked consequence. Assume a data‑centre row pulls 8,500 W of real load (servers + network, measured). With a 10 kVA Eaton 9PX (9,000 W cap) you have ~500 W headroom — about 5.5 %. With a 10 kVA APC SRT (10,000 W cap) you have 1,500 W headroom — about 15 %. That extra 1,000 W can decide whether you add a second UPS or stay single‑unit. The magnitude of the difference grows linearly with VA: at 10 kVA it’s 1 kW; at 5 kVA it’s 500 W; at 2 kVA it’s 200 W. So the proportion (1,000 W / 9,000 W ≈ 11 % more capacity on the Schneider side) is constant across the range where unity PF applies.
Reversal. If your load power factor is below 0.9 (e.g., older legacy servers with PF ~0.8), the Eaton’s 0.9 PF isn’t a bottleneck — the load itself can’t use the full wattage anyway. Also, the APC SRT below 2 kVA (1–1.5 kVA) also delivers unity PF, but at those sizes the absolute watt gain is only 100–150 W, which rarely changes a tier decision.
Numbers. The APC Smart‑UPS Online (SRT) claims a “Green Mode” (high‑efficiency bypass) up to 98 % efficiency; in standard double‑conversion, it typically runs 95–96 % at 50–80 % load. The Eaton 9PX brochure states “high‑efficiency operation” and is ENERGY STAR qualified, but does not publish a specific double‑conversion efficiency number in the allowed facts — the only anchor is that it meets ENERGY STAR thresholds (≈ 88 % at 25 % load, ≈ 92 % at full load for typical double‑conversion in ENERGY STAR 2.0). That’s a delta of roughly 3–4 percentage points at typical loading (about 95 % vs 91–92 %).
Mechanism. In double‑conversion, the rectifier + inverter pair always incurs switching losses. Higher efficiency means less heat to reject, and more of the input power reaches the load. For a 10 kW load, a 96 % efficient unit wastes ~417 W as heat; a 92 % unit wastes ~870 W — more than double the heat rejection. That waste must be handled by the cooling system (CRAC / CRAH), which adds its own energy penalty (roughly 0.3–0.5 kW per kW of heat).
Worked consequence. In a 100 kW IT load scenario with 20 UPS units (5 kVA each), a 4‑point efficiency gap means you’re rejecting an extra ~9 kW of heat across the room. That’s the equivalent of one additional 3‑ton CRAC unit running continuously. Over 3 years, that’s about 23,000 kWh of extra cooling electricity (assuming COP 3.0). The proportion here is 2× the waste heat — a multiplier that grows with load, not VA.
Reversal. The APC Green Mode is not double‑conversion; it’s an eConversion mode that disconnects the inverter and runs the load through a filtered bypass, with the inverter idling. For highly critical loads (e.g., surgical theatres, nuclear instrumentation), the transfer to battery in Green Mode takes a couple of milliseconds, which might exceed allowed interruption. Also, the Eaton 9PX in some configurations supports an “high‑efficiency” mode that drops to line‑interactive operation, but the allowed facts don’t confirm its efficiency — so you can’t assume it matches APC’s 98 %. The gap is real only if you stay in double‑conversion.
Numbers. A typical Eaton 5P (line‑interactive, VI) and the APC SMT (line‑interactive) are both rated 1500 VA, but the Eaton 5P has output PF 0.7 (1050 W) per its datasheet? — stop. Allowed facts for Eaton 5P only say “Line‑interactive (VI) topology” and do not specify output PF. But the APC SMT (also line‑interactive) is listed as “Line‑interactive (VI)” also without a PF value. This is a dead end for direct PF comparison on the low‑end units. So we shift: the key difference in real‑watts sizing shows up in the double‑conversion families where PF is stated.
Take a “3 kVA” UPS. The Eaton 9PX offers 2700 W (3 kVA × 0.9). The APC SRT in the 3 kVA class (2.2–5 kVA range) uses 0.9 PF → 2700 W as well. No difference at 3 kVA. But at 6 kVA: Eaton 9PX gives 5400 W (6 kVA × 0.9); APC SRT 6 kVA gives 6000 W (unity PF) — a 600 W gap. The proportion: 600 / 5400 = 11 % more real capacity in the APC at the same VA.
Mechanism. The VA rating is the product of RMS voltage and RMS current, but real watts are VA × cos(φ). Modern power supplies (with active PFC) can have PF very close to 1.0, so a UPS with unity output PF matches the load’s natural demand without de‑rating. A 0.9 PF UPS forces you to oversize by 11 % to get the same watt capacity. That oversizing increases purchase cost, rack space, and battery count.
Worked consequence. Suppose a row draws 54,000 W. With Eaton 9PX 6 kVA (5400 W each) you need 10 units (10 × 5400 = 54,000 W). With APC SRT 6 kVA (6000 W each) you need 9 units (9 × 6000 = 54,000 W). That’s one fewer UPS, one less set of batteries, one less set of maintenance contracts. Over a 5‑year TCO, that’s roughly 10 % lower hardware cost plus lower cooling load (one less unit’s internal losses).
Reversal. If your load is unbalanced or has a lot of non‑linear current draw with crest factor > 3, the inverter may have to limit real output anyway, and the unity PF advantage shrinks. Also, APC’s unity PF only applies to the 6–10 kVA SRT models; the 2.2–5 kVA still use 0.9 PF. So the sizing advantage is concentrated in that specific band.
Numbers. The Eaton 9PX (3U) delivers up to 5400 W in 3U, or 1800 W per U. The APC SRT 6 kVA (also 3U) delivers 6000 W, or 2000 W per U. That’s 11 % higher watt density in the APC.
Mechanism. Rack space is fixed; higher watts per U means fewer U consumed for the same total load. In a 42U rack, every U matters. The difference per unit is small (200 W/U), but over 10 units you’d save 3U of rack space (the equivalent of one extra UPS or a patch panel).
Worked consequence. For a colo deployment where rack space is billed per U, the APC SRT at 2000 W/U vs Eaton 9PX at 1800 W/U translates to about 11 % lower rack cost for the same power density. If rack rent is $50/U/month, the saving for 10 UPS is $150/month — $5,400 over 3 years.
Reversal. The Eaton 9PX also comes in a 6U configuration that can reach 10 kW (≈1667 W/U), which is less dense than the 3U APC at 2,000 W/U. But if you need longer runtime and larger battery cabinets, the density advantage of the APC may be offset by the fact that Eaton’s extended battery modules can be housed in the same rack without extra vertical space? — not allowed from facts. So stick to what’s stated: 1800 vs 2000 W/U is real but modest.
| Dimension | Eaton 9PX (double‑conversion) | APC Smart‑UPS Online (SRT) | Real‑watts delta |
|---|---|---|---|
| Output PF (6–10 kVA) | 0.9 | 1.0 (unity) | +11 % real watts at same VA |
| Output PF (2.2–5 kVA) | 0.9 | 0.9 | None |
| Efficiency (double‑conversion, typical) | ~91–92 % (ENERGY STAR 2.0) | ~95–96 % | ~3–4 pts → ~2× heat rejection |
| Watts per U (6 kVA class) | 5400 W in 3U = 1800 W/U | 6000 W in 3U = 2000 W/U | +200 W/U |
| Runtime at half load (illustrative, 6 kVA) | No runtime stated in allowed facts | No runtime stated in allowed facts | — |
Efficiency and runtime are manufacturer‑stated or derived from ENERGY STAR thresholds; see src comments. Not all values are comparable due to different test loads.
The Eaton 9PX 5400‑W 3U unit is a single inverter; the APC SRT 6 kVA at 6000 W also uses a single inverter. But if you need redundancy (2N), you buy two of each. The Eaton at 5400 W × 2 = 10,800 W total for 6U vs APC at 12,000 W total for 6U. The Eaton configuration can’t support an 11,000 W load in 2N; the APC can. That’s a hard threshold. If your load is 10,500 W, you’re forced to buy 10 kVA Eaton units (9,000 W each) → 2 × 9,000 = 18,000 W, wasting 7,500 W of capacity. The APC SRT 6 kVA at 12,000 W (2N) fits perfectly. The magnitude of the mismatch here is 1,500 W of deficit — a complete topology failure.
Threshold rule: For any load between 5,400 W and 6,000 W, the Eaton 9PX 6 kVA is insufficient (needs upgrade to 10 kVA); the APC SRT 6 kVA carries it. For loads below 5,400 W, both work. For loads above 6,000 W, you must step up to 10 kVA in both — and at 10 kVA the APC still delivers unity (10,000 W) vs Eaton’s 9,000 W. So the practical inflection is at 5,400 W (Eaton’s ceiling) and again at 6,000 W (APC’s ceiling). The proportionate advantage for APC is 11 % in the 6–10 kVA band, but the decisive factor is whether that 11 % lets you avoid one size tier. If yes, you win on first cost, space, and cooling. If no, the difference is negligible.
Rule for specifiers: If your real load falls between 90 % and 100 % of a given VA tier (e.g., 5,500 W on a 6 kVA chassis), choose the UPS with unity PF — otherwise you’ll oversize to the next VA bracket. For loads below 85 % of the tier, the PF difference rarely changes the model selection.
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.