UK’s trusted IT infrastructure partner since 2003
Servnet
FinanceToolsConfiguratorGet in Touch
Servnet Research · Data study

The rack runs out of power long before it runs out of space

We hold verified specifications for every current Dell PowerEdge, HPE ProLiant and Lenovo ThinkSystem server — 108 models, taken from the manufacturers' own technical documents. This study asks one question of all of them: how much power does a rack have to deliver, and what does that do to the number of servers you can actually install?

1600 W
median provisioned power per rack unit, across 81 rack models
100%
of models exhaust a 6 kW rack’s power before its height
15.0 kW
the densest single machine, in 4U (PowerEdge XE9680L)
between the densest and the least dense rack server

Updated 12 September 2026 · every figure computed from the manufacturers’ own documents · method and dataset below

01 — The finding

Power is the binding constraint at every realistic rack budget

Take each rack budget in turn, work out how many of a given server the power allows, and compare it with how many the 42U of height allows. At 6 kW, 100% of the 81 rack models run out of power first — and 19 of them cannot be powered at all. Even at 30 kW, a density most UK enterprise halls do not offer, 95% are still power-bound.
what the power allowswhat the 42U of height allows6 kW2 of 21100% of models run out of power before they run out of space10 kW3 of 21100% of models run out of power before they run out of space15 kW5 of 21100% of models run out of power before they run out of space30 kW10 of 2195% of models run out of power before they run out of space60 kW21 of 2169% of models run out of power before they run out of space

Scroll the chart sideways to see all of it →

Median across every rack model with node-level power supplies. “What the height allows” assumes 42U of usable space; real deployments lose several U to switching, patching and PDUs, so the gap is wider in practice.

Rack budgetModels that fit at allMedian that fit on powerMedian the height allowsPower-bound
6 kW62 of 81221100%
10 kW66 of 81321100%
15 kW75 of 81521100%
30 kW81 of 81102195%
60 kW81 of 81212169%

Scroll the table sideways to see every column →

02 — The spread

From 500 to 3750 watts per rack unit

Every rack server in the catalogue, plotted by chassis height against the power a feed must deliver per rack unit. The cluster along the bottom is the mainstream estate — one and two rack units, one or two supplies. The outliers at the top are accelerator platforms, and the liquid-cooled models sit almost entirely above the air-cooled ones.
01k2k3k1U2U4U6U8U10UW/Uchassis heightPowerEdge XE9680L — 4U, 3750 W/U, 15.0 kW provisioned, up to 8 accelerators, liquid-cooledPowerEdge XE9685L — 4U, 3750 W/U, 15.0 kW provisioned, up to 8 accelerators, liquid-cooledHPE ProLiant Compute XD685 — 5U, 3600 W/U, 18.0 kW provisioned, up to 8 accelerators, liquid-cooledHPE ProLiant Compute DL384 Gen12 — 2U, 3300 W/U, 6.6 kW provisioned, up to 2 accelerators, air-cooledPowerEdge XE7740 — 4U, 3200 W/U, 12.8 kW provisioned, up to 16 accelerators, air-cooledPowerEdge XE7745 — 4U, 3200 W/U, 12.8 kW provisioned, up to 16 accelerators, air-cooledHPE ProLiant Compute DL380a Gen12 — 4U, 3200 W/U, 12.8 kW provisioned, up to 10 accelerators, air-cooledHPE ProLiant Compute XD230 — 1U, 3200 W/U, 3.2 kW provisioned, air-cooledPowerEdge XE9680 — 6U, 2667 W/U, 16.0 kW provisioned, up to 8 accelerators, air-cooledThinkSystem SR780a V3 — 5U, 2560 W/U, 12.8 kW provisioned, up to 8 accelerators, air-cooledPowerEdge XE8640 — 4U, 2400 W/U, 9.6 kW provisioned, up to 4 accelerators, air-cooledHPE ProLiant Compute DL320 Gen12 — 1U, 2400 W/U, 2.4 kW provisioned, up to 4 accelerators, air-cooledHPE ProLiant Compute DL325 Gen12 — 1U, 2400 W/U, 2.4 kW provisioned, up to 4 accelerators, air-cooledHPE ProLiant DL320 Gen11 — 1U, 2200 W/U, 2.2 kW provisioned, up to 4 accelerators, air-cooledHPE ProLiant DL325 Gen11 — 1U, 2200 W/U, 2.2 kW provisioned, up to 4 accelerators, air-cooledHPE ProLiant DL360 Gen11 — 1U, 2200 W/U, 2.2 kW provisioned, up to 3 accelerators, air-cooledHPE ProLiant Compute DL360 Gen12 — 1U, 2200 W/U, 2.2 kW provisioned, up to 3 accelerators, air-cooledHPE ProLiant DL365 Gen11 — 1U, 2200 W/U, 2.2 kW provisioned, up to 2 accelerators, air-cooledHPE ProLiant DL380a Gen11 — 2U, 2200 W/U, 4.4 kW provisioned, up to 8 accelerators, air-cooledHPE ProLiant DL560 Gen11 — 2U, 2200 W/U, 4.4 kW provisioned, up to 2 accelerators, air-cooledThinkSystem SR630 V4 — 1U, 2000 W/U, 2.0 kW provisioned, up to 3 accelerators, air-cooledPowerEdge XE9780 — 10U, 1920 W/U, 19.2 kW provisioned, up to 8 accelerators, liquid-cooledPowerEdge XE9785 — 10U, 1920 W/U, 19.2 kW provisioned, up to 8 accelerators, air-cooledHPE Compute XD690 — 10U, 1920 W/U, 19.2 kW provisioned, up to 8 accelerators, air-cooledThinkSystem SR680a V4 — 8U, 1900 W/U, 15.2 kW provisioned, up to 8 accelerators, air-cooledPowerEdge R660 — 1U, 1800 W/U, 1.8 kW provisioned, up to 3 accelerators, air-cooledPowerEdge R660xs — 1U, 1800 W/U, 1.8 kW provisioned, air-cooledPowerEdge R6615 — 1U, 1800 W/U, 1.8 kW provisioned, up to 2 accelerators, air-cooledPowerEdge R6625 — 1U, 1800 W/U, 1.8 kW provisioned, up to 2 accelerators, air-cooledPowerEdge R670 — 1U, 1800 W/U, 1.8 kW provisioned, up to 3 accelerators, air-cooledPowerEdge R6715 — 1U, 1800 W/U, 1.8 kW provisioned, up to 3 accelerators, air-cooledPowerEdge R6725 — 1U, 1800 W/U, 1.8 kW provisioned, up to 3 accelerators, air-cooledPowerEdge XR5610 — 1U, 1800 W/U, 1.8 kW provisioned, up to 2 accelerators, air-cooledThinkSystem SR630 V3 — 1U, 1800 W/U, 1.8 kW provisioned, up to 3 accelerators, air-cooledThinkSystem SR635 V3 — 1U, 1800 W/U, 1.8 kW provisioned, up to 5 accelerators, air-cooledThinkSystem SR645 V3 — 1U, 1800 W/U, 1.8 kW provisioned, up to 5 accelerators, air-cooledThinkSystem SR675 V3 — 3U, 1733 W/U, 5.2 kW provisioned, up to 8 accelerators, air-cooledPowerEdge R760 — 2U, 1600 W/U, 3.2 kW provisioned, up to 6 accelerators, air-cooledPowerEdge R760xa — 2U, 1600 W/U, 3.2 kW provisioned, up to 12 accelerators, air-cooledPowerEdge R7625 — 2U, 1600 W/U, 3.2 kW provisioned, up to 6 accelerators, air-cooledPowerEdge R770 — 2U, 1600 W/U, 3.2 kW provisioned, up to 6 accelerators, air-cooledPowerEdge R770AP — 2U, 1600 W/U, 3.2 kW provisioned, air-cooledPowerEdge R7715 — 2U, 1600 W/U, 3.2 kW provisioned, up to 6 accelerators, air-cooledPowerEdge R7725 — 2U, 1600 W/U, 3.2 kW provisioned, up to 6 accelerators, air-cooledPowerEdge R7725xd — 2U, 1600 W/U, 3.2 kW provisioned, air-cooledHPE ProLiant Compute DL340 Gen12 — 2U, 1600 W/U, 3.2 kW provisioned, up to 4 accelerators, air-cooledHPE ProLiant Compute DL340e Gen12 — 2U, 1600 W/U, 3.2 kW provisioned, air-cooledHPE ProLiant Compute DL345 Gen12 — 2U, 1600 W/U, 3.2 kW provisioned, up to 6 accelerators, air-cooledHPE ProLiant Compute DL580 Gen12 — 4U, 1600 W/U, 6.4 kW provisioned, air-cooledThinkSystem SR650 V4 — 2U, 1600 W/U, 3.2 kW provisioned, up to 10 accelerators, air-cooledThinkSystem SR650a V4 — 2U, 1600 W/U, 3.2 kW provisioned, up to 8 accelerators, air-cooledThinkSystem SR650i V4 — 2U, 1600 W/U, 3.2 kW provisioned, up to 2 accelerators, air-cooledThinkSystem SR680a V3 with B200 — 8U, 1600 W/U, 12.8 kW provisioned, up to 8 accelerators, air-cooledThinkSystem SR850 V4 — 2U, 1600 W/U, 3.2 kW provisioned, up to 4 accelerators, air-cooledThinkSystem SR860 V4 — 4U, 1600 W/U, 6.4 kW provisioned, up to 8 accelerators, air-cooledPowerEdge R470 — 1U, 1500 W/U, 1.5 kW provisioned, up to 4 accelerators, air-cooledPowerEdge R860 — 2U, 1400 W/U, 2.8 kW provisioned, air-cooledPowerEdge R960 — 4U, 1400 W/U, 5.6 kW provisioned, up to 4 accelerators, air-cooledThinkSystem SR850 V3 — 2U, 1350 W/U, 2.7 kW provisioned, up to 4 accelerators, air-cooledThinkSystem SR860 V3 — 4U, 1350 W/U, 5.4 kW provisioned, up to 8 accelerators, air-cooledThinkSystem SR950 V3 — 8U, 1350 W/U, 10.8 kW provisioned, air-cooledHPE ProLiant Compute DL110 Gen12 — 1U, 1300 W/U, 1.3 kW provisioned, up to 2 accelerators, air-cooledThinkSystem SR650 V3 — 2U, 1300 W/U, 2.6 kW provisioned, up to 8 accelerators, air-cooledThinkSystem SR655 V3 — 2U, 1300 W/U, 2.6 kW provisioned, up to 8 accelerators, air-cooledThinkSystem SR665 V3 — 2U, 1300 W/U, 2.6 kW provisioned, up to 7 accelerators, air-cooledThinkSystem SR685a V3 — 8U, 1300 W/U, 10.4 kW provisioned, up to 8 accelerators, air-cooledPowerEdge R7615 — 2U, 1200 W/U, 2.4 kW provisioned, up to 6 accelerators, air-cooledHPE ProLiant DL345 Gen11 — 2U, 1100 W/U, 2.2 kW provisioned, up to 4 accelerators, air-cooledHPE ProLiant DL380 Gen11 — 2U, 1100 W/U, 2.2 kW provisioned, up to 8 accelerators, air-cooledHPE ProLiant Compute DL380 Gen12 — 2U, 1100 W/U, 2.2 kW provisioned, up to 8 accelerators, air-cooledHPE ProLiant DL385 Gen11 — 2U, 1100 W/U, 2.2 kW provisioned, up to 8 accelerators, air-cooledHPE ProLiant DL20 Gen11 — 1U, 1000 W/U, 1.0 kW provisioned, air-cooledPowerEdge R570 — 2U, 900 W/U, 1.8 kW provisioned, up to 4 accelerators, air-cooledPowerEdge R760xd2 — 2U, 900 W/U, 1.8 kW provisioned, up to 2 accelerators, air-cooledPowerEdge R760xs — 2U, 900 W/U, 1.8 kW provisioned, up to 2 accelerators, air-cooledPowerEdge XR7620 — 2U, 900 W/U, 1.8 kW provisioned, up to 5 accelerators, air-cooledThinkSystem SR250 V3 — 1U, 800 W/U, 0.8 kW provisioned, up to 1 accelerators, air-cooledPowerEdge R260 — 1U, 700 W/U, 0.7 kW provisioned, air-cooledPowerEdge R360 — 1U, 700 W/U, 0.7 kW provisioned, up to 1 accelerators, air-cooledPowerEdge R5715 — 2U, 550 W/U, 1.1 kW provisioned, air-cooledHPE ProLiant DL145 Gen11 — 2U, 500 W/U, 1.0 kW provisioned, up to 3 accelerators, air-cooledair-cooledliquid-cooleddot size = accelerators supported

Scroll the chart sideways to see all of it →

Hover any point for the model. Dot size is the number of accelerators the chassis supports; colour is the cooling the manufacturer specifies.

ModelHeightSuppliesProvisionedPer rack unitAcceleratorsCooling
PowerEdge XE9680L4U5 live of 6 × 3000 W15.0 kW3750 W8liquid
PowerEdge XE9685L4U5 live of 6 × 3000 W15.0 kW3750 W8liquid
HPE ProLiant Compute XD6855U6 live of 12 × 3000 W18.0 kW3600 W8liquid
HPE ProLiant Compute DL384 Gen122U3 live of 4 × 2200 W6.6 kW3300 W2air
PowerEdge XE77404U4 live of 8 × 3200 W12.8 kW3200 W16air
PowerEdge XE77454U4 live of 8 × 3200 W12.8 kW3200 W16air
HPE ProLiant Compute DL380a Gen124U4 live of 8 × 3200 W12.8 kW3200 W10air
HPE ProLiant Compute XD2301U1 live of 2 × 3200 W3.2 kW3200 Wair
PowerEdge XE96806U5 live of 6 × 3200 W16.0 kW2667 W8air
ThinkSystem SR780a V35U4 live of 8 × 3200 W12.8 kW2560 W8air

Scroll the table sideways to see every column →

The ten densest machines in the catalogue. Every one of them is an accelerator platform, and every figure comes from the model’s own manufacturer document — the same documents behind our 108 data sheets.

03 — Where the jump happens

Per rack unit it is a nudge; per machine it is four times the load

This is the part that surprises people. Density per rack unit barely moves as you add accelerators — a modern 2U server already offers supplies sized for a fully-loaded configuration. What changes is the size of the machine: an eight-accelerator platform is four to ten rack units tall and carries several supplies, so the whole machine asks for 12.8 kW where a conventional server asks for 3.2 kW.
median provisioned load for the whole machine, kWNo accelerator support3.2 kW11 models · peak 3200 W/U1–3 accelerators1.8 kW21 models · peak 3300 W/U4–7 accelerators2.6 kW23 models · peak 2400 W/U8 or more12.8 kW26 models · peak 3750 W/U

Scroll the chart sideways to see all of it →

The practical consequence: you do not plan for AI by finding space. You plan for it by finding a feed, and usually by finding a hall that offers 30 kW or more per rack — which is why colocation enquiries now start with power density rather than floor area. Our rack and power planner works the same arithmetic against your own estate.

04 — Cooling

Liquid cooling is 8% of the catalogue, and it sits at the top of the chart

9 of the 108 models are described by their manufacturer as liquid-cooled, direct-liquid-cooled or water-cooled. Their median provisioned density is 3750 W/U against 1600 W/U for the air-cooled models — so on this evidence the crossover is a power-density threshold, not a GPU count. Air-cooled machines still reach 3300 W/U, which is a useful thing to know if your facility cannot take water.
CoolingModelsMedian W/UHighest W/UHighest provisioned load
Air-cooled771600 W3300 W19.2 kW
Liquid-cooled43750 W3750 W19.2 kW

Scroll the table sideways to see every column →

05 — Method

What was counted, and what it does not claim

How these figures were produced

  • Population: all 108 current server models from Dell PowerEdge, HPE ProLiant and Lenovo ThinkSystem for which we hold a verified specification extracted from the manufacturer's own technical document. 81 of them are rack-mount with power supplies of their own and are included in the density figures.
  • Excluded: blades, multi-node trays, Synergy modules and rack-scale sleds. Their power supplies belong to an enclosure, not to the node — a Lenovo SC750 V4 tray lists 15,000 W because that is its N1380 enclosure shelf, and counting it as “15 kW in 1U” would be wrong. Two models have no supplies at all because the chassis powers them.
  • Provisioned power = the largest supply the vendor lists for the model × the number of supplies that must be live in the most demanding documented mode (the N of N+1 or N+N). Where a vendor documents no redundancy mode and the model takes two supplies, 1+1 is assumed; where it takes more than two and states no mode, half are assumed live. Every per-model figure, including the supply count and redundancy wording, is in the dataset so the arithmetic can be checked.
  • Chassis with separate power zones are read from their own zoned statement, not from the summary line. The Dell XE7740 and XE7745 list “7+1 redundant” across eight bays, but the technical guide defines full redundancy as 1+1 in the CPU zone plus 3+3 in the GPU zone — four supplies live, not seven. Reading the summary line alone overstated those two machines by 9.6 kW each, and they were the top of this table until it was corrected on 12 September 2026.
  • This is a facilities ceiling, not a measurement. It is what a rack feed must be able to deliver to a fully-configured machine. Actual draw depends on the configuration and the workload and is lower — usually much lower.
  • Rack arithmetic assumes 42U of usable height and no derating of the power budget. Real installations lose height to switching and patching, and most electrical designs leave headroom, so the power constraint bites sooner than shown here, not later.
  • No prices appear anywhere in this study, and no vendor sponsored it.
Download the dataset (JSON)The 108 data sheets behind it

Free to reuse with attribution to Servnet and a link to this page. If you are citing a single model’s figure, cite its data sheet too — each one names the manufacturer document and revision it was taken from.

06 — Questions

Server power density, answered

How much power does a modern server need per rack unit?

Across 81 current rack servers from Dell, HPE and Lenovo, the median is 1600 W per rack unit of provisioned power — the capacity a rack feed has to be able to deliver to a fully-loaded example. The range is 500 W/U at the bottom to 3750 W/U at the top, a spread of about 8×.

How many servers fit in a 6 kW rack?

On power rather than height: a median of 2. A 42U rack has room for far more — the median model would fit 21 times over by height — but 100% of models run out of power first, and 19 of the 81 models cannot be powered at all from a 6 kW feed.

Do AI servers really need that much more power?

Yes, but not in the way people assume. Per rack unit the difference is modest — the median machine with eight or more accelerators provisions 1900 W/U against 1600 W/U for one with no accelerator support. It is the whole machine that changes: 12.8 kW against 3.2 kW, because the chassis is four to ten rack units tall and carries up to 12 power supplies.

At what point do you need liquid cooling?

9 of the 108 models in this study are described by their manufacturer as liquid-cooled or direct-liquid-cooled — 8% of the catalogue. Their median density is 3750 W/U against 1600 W/U for air-cooled models, so on this evidence the practical crossover sits above roughly 2.5 kW per rack unit rather than at a fixed GPU count.

Is provisioned power the same as what the server draws?

No, and the difference matters. Provisioned power is what the feed must be able to deliver to a fully-configured machine: the largest supply the vendor lists, times the number of supplies that have to be live in the most demanding documented mode. A typical configuration running a typical workload draws less. Facilities are sized on the former; electricity bills are paid on the latter.

Why is rack space no longer the constraint?

Because power per rack has not grown as fast as power per server. A 42U rack is the same height it was twenty years ago; the servers in it now provision up to 15.0 kW each. At every budget below 30 kW in this study, the overwhelming majority of models exhaust the power budget while the rack still has empty U.

Work the numbers on your own estate

Rack & power density plannerYour kit, your racks, your feed — what fits and what does notOpen →AI GPU calculatorModel size to GPUs to power draw to a bill of materialsOpen →Study: what actually differs between the vendors68% of processors are sold by all three — so what are you choosing?Open →Compare any two serversEvery model in this study, head to headOpen →

Talk to a UK specialist

Get expert advice or a no-obligation quote — servers, storage, networking, maintenance, finance and cloud. We reply the same working day.

or call 0800 987 4111