You're speccing a row of servers that draw 4200 W. A sales rep points you to a 5 kVA Schneider Galaxy VS — because "VA" is what everyone talks about. But 5 kVA at unity power factor (PF) delivers 5000 W; at 0.9 PF it's 4500 W; at 0.8 PF it's 4000 W — and your 4200 W load already exceeds that. This is the real-watts trap: sizing by apparent power (VA) without confirming the real power (W) capability at the load's power factor. APC UPS by Schneider Electric (the host brand) and Schneider UPS (the broader 3-phase family) both sell double-conversion online units, but the way each derives its watt rating—and how that interacts with the load's PF—determines whether you over-buy or under-build. This teardown walks four dimensions where the magnitude of the watt-to-VA ratio changes the sizing decision.
1. The Watt-VA Ratio at Rated Load: Unity vs 0.9 vs 0.8
The Galaxy VS, in its 10–150 kW range (at 400 V), is rated with a unity output power factor: 10 kVA = 10 kW. That's clean—you get 100% of the VA as watts. APC Smart-UPS Online (SRT) units from 6–10 kVA also deliver a unity output PF. So at this scale, a 10 kVA Galaxy VS and a 10 kVA APC SRT both deliver 10,000 W. The ratio is 1:1. No trap yet. But step down to the 2.2–5 kVA SRT range: the output PF is 0.9, meaning a 5 kVA APC unit delivers 4500 W. A similarly-rated 5 kVA Galaxy VS (unity PF) would deliver 5000 W. That 500 W difference (11% more watts for the same VA stamp) flips the sizing for a load that draws, say, 4800 W at 0.8 PF. The APC unit (4500 W) would be undersized; the Galaxy VS (5000 W) would hold it, barely. The magnitude here is not trivial: for a 30-rack deployment, scaling each UPS by 11% of real wattage can shift the total installed cost by thousands of dollars—and the risk of an overload trip.
Worked consequence: If your load's PF is 0.7 or 0.8 (common in legacy server architectures), a UPS with a 0.9 PF rating requires you to derate the VA by ~10–20%. Sizing from real watts on the nameplate eliminates that guesswork. For new IT loads with PF-corrected power supplies (PF ~0.98–0.99), the ratio is nearly 1:1 for either brand, and the difference collapses.
Reversal: For loads with a leading PF or high crest factor (e.g., older large-motor drives), the unity PF rating of the Galaxy VS is an advantage, but only if the UPS inverter can handle the current harmonics—Schneider specifies Class 1 performance and harmonic filtering for the Galaxy VS. APC does not claim that layer of filtering for the SRT range; the SRT is a single-phase 1–10 kVA product family, not designed for 3-phase motor loads.
2. Efficiency at Real-Watt Load Point: The Green Mode Lever
Efficiency numbers are always stated at some load point—typically 100% or 75%—and the shape of the curve changes the real-world operating cost. APC's SRT offers a "Green Mode" that routes load through a delta-conversion path, achieving up to 98% efficiency. The Galaxy VS in its eConversion mode (default) reaches up to 99% efficiency, with a ~2–3.8% efficiency gain over standard double-conversion. The magnitude: a 0.5% efficiency difference on a 50 kW load, running 8,760 hours a year at $0.12/kWh, is about $525 per year. That's a number, but the mechanism is not the raw number—it's that the eConversion mode on the Galaxy VS is the default operating mode, meaning you get the high efficiency without manual intervention, and it claims no-break transfer to double-conversion or battery. The APC Green Mode is an optional setting; if the load-sensitive transfer time (which is zero in double-conversion) is critical—say, for storage-area network or medical imaging—you might keep it in double-conversion at ~96% efficiency, erasing the advantage.
Worked consequence: For a colocation hall with 200 kW of mixed IT load, the 1% efficiency delta between APC's double-conversion (96%) and Galaxy VS eConversion (99%) saves about $1,800 per year per 50 kW block—~$7,200/year. But that's only if the load can tolerate the sub-cycle transfer; for a data center running Cisco UCS or Dell PowerEdge (which tolerate
Reversal: At low loads (10–30% of rating), efficiency curves often droop. The Galaxy VS claims "up to 97% at every load level" in double-conversion, which is unusually flat. APC's SRT double-conversion efficiency is about 96% at full load, likely lower at 20% load (typical of lightly-loaded single-phase units). If you're running a lightly-loaded branch (say 1.5 kVA average on a 5 kVA SRT), the real efficiency may be closer to 92–93%—a penalty that matters for UPS in a continuously-running server closet.
3. Runtime at Real Watts: The Battery Bank Proportional Penalty
Runtime curves are always given at a specific watt load. For example, a Tripp Lite SU3000RTXL3U (3000 VA / 2400 W) delivers ~14 min at half load (1200 W) and ~5 min at full load (2400 W). The APC SRT 3000 VA (2.2–5 kVA range, 0.9 PF → 2700 W) does not have a published runtime at 2700 W in the allowed facts, but the relationship is proportional: a heavier watt load (2700 vs 2400 W) reduces runtime by roughly the ratio (2400/2700 ≈ 0.89 → ~15% less runtime). In a compare: if a specifier picks a 3 kVA Galaxy VS (3000 W) for a 2500 W load, the runtime is based on 2500 W. If an APC SRT 3 kVA (2700 W) is chosen for the same 2500 W load, the runtime will be slightly longer (since the load is 93% of rating vs 83%). The magnitude of this shift is about 10–15% runtime difference, which can mean the difference between a clean shutdown and an uncontrolled crash in a 5-minute power blip.
Worked consequence: For a 30-second outage, both survive. For a 8-minute outage, the 15% shorter runtime on the APC could leave the load unprotected 2 minutes earlier. However, the APC SRT supports external battery packs; adding one extra pack recovers the runtime and then some. The Galaxy VS is a 3-phase unit and its external battery cabinets are in the 10–150 kW scale—not a fair comparison for a single-phase 3 kVA load.
Reversal: This dimension only matters for mission-critical edge sites without generator backup. If you have a standby generator that starts in 10 seconds, runtime is irrelevant. The real-watt sizing trap here is that a 3000 VA UPS with 0.8 PF (2400 W) may be perfectly adequate for a 2300 W load, but if the specifier picks a "3 kVA" APC without checking the watt rating (2700 W), they might think they have more headroom than they do—or vice versa for the Galaxy VS (3000 W).
4. Form Factor and Real-Watt Density: Watts per U vs VA per U
APC SRT 1–10 kVA units are typically 2U or 3U rack-mount; the 5 kVA SRT (0.9 PF → 4500 W) occupies 3U. The Galaxy VS 10–150 kW are floor-standing towers (not rack-mount). For the rack-mount comparison, a relevant competitor is the Eaton 9PX, which delivers up to 5400 W in 3U (0.9 PF, 6 kVA). Neither APC nor Schneider UPS produce a 3U unit with 5400 W today. The magnitude: 5400 W in 3U vs APC's 4500 W in 3U is 20% more real watts per rack unit. For a rack with a 5 kW power budget, the Eaton 9PX fits, the APC SRT 5 kVA does not—you'd need the 6 kVA SRT (Unity PF, 6000 W) which is likely 4U. The form factor penalty is real but only if space is the constraint.
Worked consequence: In a co-location cage where space is billed per U, a 4U UPS vs a 3U UPS costs you 33% more rack space for the same power. That's a direct opex hit of ~$150–300/month per rack.
Reversal: If you have floor space and are sizing for a large load (>10 kW), the Galaxy VS 10–150 kW towers dominate in terms of watts-per-square-foot. The form-factor dimension is only a tiebreaker for small-to-mid-size deployments (
| Dimension | APC Smart-UPS Online (SRT) | Schneider Galaxy VS | Comment |
|---|---|---|---|
| Output PF (at rating) | 0.9 (2.2–5 kVA); Unity (1–1.5, 6–10 kVA) | Unity (1.0) | Unity PF simplifies sizing |
| Max efficiency (double-conversion) | ~96% | Up to 97% at any load | ~1% difference |
| Max efficiency (eco mode) | Up to 98% (Green Mode) | Up to 99% (eConversion) | Galaxy VS default is eco mode |
| Power density (5 kVA class) | 4500 W in 3U ~ 1500 W/U | N/A (floor-standing) | Eaton 9PX: 5400 W in 3U |
| Input voltage range | 100–125 V (for 1–1.5 kVA); higher for others | 208/400/480 V | Galaxy VS is 3-phase only |
When the Magnitude Flips: A Failure Mode
Suppose you have a load with a power factor of 0.95 (modern switching power supplies). The Galaxy VS at unity PF delivers 10,000 W from a 10 kVA frame. The APC SRT at 10 kVA (unity PF) also delivers 10,000 W. Both are identical. The efficiency difference (1% in double-conversion) is real but small. The runtime dimension depends on battery sizing. The form factor only matters for rack density. The failure mode is when the load PF is low (0.7–0.8) and the UPS has a non-unity PF rating. For example, the APC SRT 5 kVA (0.9 PF) can only deliver 4500 W; a load of 4800 W at 0.8 PF would overload it. A Galaxy VS 5 kVA (unity PF) would be fine. If you're sizing for a data center with legacy power supplies (PF ~0.75), go with the Galaxy VS or any unity-PF-rated UPS. If your load is all PF-corrected, either works — then let the form factor and management software (PowerChute for APC vs Galaxy VS's native monitoring) decide.
Rule-of-thumb closure: Before you pick a UPS, compute your load's maximum real power in watts (not VA). Then multiply by 1.1 (10% headroom). That number must be ≤ the UPS's real watt rating at the output PF stamped on its nameplate. Ignore VA entirely for the sizing step. Use VA only for branch circuit and generator sizing, which is a different problem. If the UPS's rated output PF is not 1.0, the real watt rating is the only number that matters for the load.
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.