How Many Watts Does a Hard Drive Use? 63 SKUs, 2.4 to 6.7 W

Google autocompletes "how many watts does a hard drive use" and the answer I keep seeing is a single number, usually 6 to 9 W, with no model attached. I cannot measure one myself: the only drive in my rig is the Mac mini's internal SSD, and I have never had a plug meter on that machine. What I can do is read every current datasheet at once. On 2026-09-13 I pulled 10 spec sheets from Western Digital, Seagate, Toshiba and Samsung and copied the power rows for 73 model numbers: 63 3.5-inch hard drives, 3 laptop drives and 7 SSD capacities. The 3.5-inch drives idle between 2.4 and 6.7 W, work between 3.1 and 9.07 W, sit in standby between 0.25 and 1.6 W, and pull 14 to 30 W for a second when they spin up. Which end of that range you get is decided by the model number on the label, not by the brand, the capacity or the colour of the box.

The short answer, by drive class

Class	Idle	Read/write	Standby or sleep	Spin-up peak	Rows
3.5-inch HDD, 5400 to 7200 rpm	2.4 to 6.7 W (median 3.7)	3.1 to 9.07 W (median 6.1)	0.25 to 1.6 W	1.2 to 2.5 A at 12 V, so 14 to 30 W	63
2.5-inch laptop HDD, 5400 rpm class	0.5 to 0.55 W	1.4 to 1.7 W	0.1 to 0.16 W	not published	3
2.5-inch SATA SSD (870 EVO)	0.030 to 0.035 W	2.0 to 2.5 W	0.0025 to 0.0065 W	none	5
M.2 NVMe SSD (990 PRO)	0.050 to 0.055 W	5.0 to 5.8 W	0.005 W	none	2

Every figure is the manufacturer's own "typical" or "average" number, measured on their bench, not mine. The 3.5-inch rows come from the WD Red Plus, WD Red Pro and WD Blue sheets dated September 2025, Seagate's IronWolf Z-SKU sheet from January 2025 and its BarraCuda 16 to 24TB and BarraCuda 1 to 8TB sheets, and Toshiba's N300 sheet dated April 2026; the SSD rows are Samsung's 870 EVO and 990 PRO datasheets, and the laptop rows the WD Blue Mobile sheet, unrevised since December 2019. The vendors name the states differently: WD prints "Idle" and "Standby/Sleep", Seagate "Idle Average" and "Standby Mode/Sleep Mode", Toshiba "active idle" and no standby figure at all. I mapped them onto three columns and kept the numbers as printed.

The watts are attached to the model number, not the box

The product families overlap in a way the boxes hide. Western Digital's 10TB and 12TB air-filled drive carries the identical power row, 8.8 W read/write, 6.1 W idle, 0.3 W standby and a 1.9 A spin-up, under six model numbers in three colours: WD100EFGX and WD120EFGX as Red Plus, WD103KFBX and WD122KFBX as Red Pro, WD100EAGZ and WD120EAGZ as Blue. The 8TB WD80EFPX (Red Plus) and WD80EAAZ (Blue) share 5.2/3.4/0.4 W, and the pattern repeats at 6, 4, 3 and 2TB. In all, 19 of the 37 WD model numbers on three sheets carry power figures that are copies of each other. The NAS label buys firmware, a workload rating and a longer warranty; on the datasheet it does not buy a lower wattage.

The flip side is that one label hides very different drives. I went through this for WD Red Plus alone in August: two 12TB Red Plus drives, WD120EFGX and WD120EFBX, idle at 6.1 W and 2.9 W. The same thing happens inside Red Pro (12TB WD122KFBX 6.1 W, 12TB WD121KFBX 2.8 W) and inside BarraCuda (2TB ST2000DM005 2.5 W, 2TB ST2000DM008 3.9 W). Toshiba's N300 line is the widest: the 8TB and 10TB models idle at 5.74 W and work at 9.07 W, the highest figures anywhere in the 63 rows, while the 12TB N300 idles at 3.30 W and the 22TB at 4.35 W.

Capacity does not predict watts

The 15 drives of 16TB and up idle between 3.4 and 6.7 W. The 12 drives of 8 to 10TB idle between 3.4 and 6.1 W. The ranges almost coincide, so a bigger drive does not cost more to keep spinning, and it often costs less: WD's 26TB Red Pro WD260KFGX idles at 3.6 W against the 10TB Red Pro at 6.1 W, and Toshiba's 12TB N300 idles at 3.30 W against its 8TB at 5.74 W. What the datasheets do say is the fill gas: every WD and Seagate drive that lists helium idles between 2.8 and 5.1 W, and every air-filled drive at 10TB or more idles at 4.6 or 6.1 W. Across the whole set the medians are close, 3.6 W for 13 helium rows against 3.4 W for 29 air rows, only because the air group is full of small 5400 rpm drives; match the capacity and helium roughly halves the number. Seagate and Toshiba do not state the fill on the BarraCuda and N300 sheets, so those 21 rows are grey in the chart.

Speed is the other lever, and it is where desktop drives stop being cheap. Seagate rates the 16 to 24TB BarraCuda at 6.7 W idle and 8.9 W operating at 7200 rpm, the highest idle in the table. The 1 to 8TB BarraCudas are 5400 rpm SMR parts at 2.5 to 3.4 W idle, and the same sheet rates them for 2,400 power-on hours a year and a 55 TB/year workload with a 2-year warranty, against 8,760 hours and 180 TB/year for the IronWolf. The lower wattage on a desktop drive comes with a duty-cycle rating of about 6.5 hours a day. The SMR side of that trade is a separate post.

What the forums measure at the wall

Datasheet watts are DC at the drive connector; owners measure AC at the outlet, and the two disagree in a predictable direction. On the TrueNAS forums in July 2024 a user called Electr0 reported 120 W idle on a Kill-A-Watt for a system with eight 6TB Seagate drives and had budgeted them at 5.3 W each. Another member pulled the sheet for the actual model, ST6000VN0033, found 8.1 W operating and 7.2 W idle, and Electr0 re-did the arithmetic at 58 to 65 W for the drives alone. The same thread has the other half of the gap: "depending on the PSU power conversion efficiency at 100W, you may have to add an additional 20% on top of that." In a November 2024 thread, a newcomer asking how to get a two-drive box under 10 W was told flatly, "You cannot achieve that figure with spinning drives."

When the model number is known, the datasheet lands close. A Hacker News commenter wrote in November 2025 that four 8TB WD Red Plus drives "which uses 3.4W idle, and assuming 20W for the NAS it's at ~34W (measured 35W)"; 3.4 W is exactly the WD80EFPX row above. The same person had earlier measured 45 W at the wall for a DS918+ with four 8TB WD Reds and noted that Synology's own figures are measured with 1TB WD10EFRX drives at 3.1 W idle, which is why a spec-page NAS wattage rarely matches a populated one. Every one of these is a reader's meter, not mine.

Standby is not zero, and spin-down is a fight

The lowest 3.5-inch standby figure on any sheet is 0.25 W (Seagate's 2 to 8TB IronWolf and BarraCuda) and the highest is 1.6 W (WD's 20TB Red Pro WD201KFGX). WD's helium drives standby between 0.6 and 1.2 W, up to four times their air-filled siblings at 0.3 W, and Toshiba prints no standby number at all. A spun-down 4-bay of big helium drives still draws 2.4 to 4.8 W before the NAS itself.

Whether to spin down is argued in every thread I read. On the AnandTech forums in September 2020 a poster measured a Lenovo PX4-400R with three 4TB drives at "a constant 40 watts" on a Killawatt, with a spin-down target of about 13 W, and was told in the same thread that "spinning down your HDD's routinely is the fastest way to mechanical failure." A TrueNAS regular in the November 2024 thread put the other side as a question: spin down "just for save (don't know) 2€ every months? Not worth." The DS918+ owner added that WD Red drives "do not support spin down, so at most they're idle," an owner's report I cannot confirm from WD's sheet. I have no drive to test either claim on; what the datasheets settle is that the prize for winning the argument is the idle-minus-standby gap, 2.1 to 5.7 W per drive.

Spin-up is the number that trips power supplies

Every WD and Seagate sheet lists a 12 V start-up current, from 1.2 A on the smallest WD drives to 2.0 A on every BarraCuda and 2.5 A on the 1TB WD10EZEX: 14 to 30 W per drive for the second or two the platters take to reach speed, and it happens to every drive in a box at once unless the controller staggers them. A Hacker News commenter who ran large arrays described the failure mode: "Spinning up 48/96 drives on boot can surge beyond what power supplies could handle." It is the peak, not the idle, that a UPS or a cheap USB power supply has to survive.

The calculator, and what the spread is worth

The arithmetic is watts times 8.76, which gives kilowatt-hours per year, times your rate. At the June 2026 US residential average of 18.34 cents per kWh from EIA's Electric Power Monthly, one watt held for a year costs $1.61. That turns the table into money:

Drive state	Watts	kWh per year	Cost per year, one drive	Four drives
SATA SSD idle (870 EVO)	0.03	0.3	$0.05	$0.19
3.5-inch standby, lowest	0.25	2.2	$0.40	$1.61
2.5-inch laptop HDD idle	0.5	4.4	$0.80	$3.21
3.5-inch idle, lowest (WD20EFPX)	2.4	21.0	$3.86	$15.42
3.5-inch idle, median	3.7	32.4	$5.94	$23.78
3.5-inch idle, highest (BarraCuda 16 to 24TB)	6.7	58.7	$10.76	$43.06
3.5-inch read/write, highest (N300 8 and 10TB)	9.07	79.5	$14.57	$58.29

The spread matters most where it compounds. Four air-filled 12TB Red Plus drives at 6.1 W idle cost $39 a year to leave spinning; four helium WD120EFBX at 2.9 W cost $19. Over a 3-year warranty that is $62 for choosing the right suffix on the same product name. Against the whole system it is small: when I ran my own Mac mini through six online calculators the input that swung the answer was the machine's watts, and one idle 12TB air drive at 6.1 W draws nearly twice the 3.20 W median package power my mini reported under powermetrics while my agent sat idle. SSDs end the idle question and reopen the active one: the 990 PRO's 5.8 W read figure is level with a 7200 rpm 3.5-inch drive working, so an NVMe scratch disk under constant load is not the free lunch its 50 mW idle suggests.

Which drives to buy for the low end of the table

None of these drives has been in my hands; the wattage is the datasheet's. At 4TB the two NAS drives I already link, the WD Red Plus WD40EFPX at 3.1 W idle and the Seagate IronWolf ST4000VNZ06 at 3.2 W, are within 0.1 W of each other and sit in the bottom third of the chart; check the model number on the listing, because the IronWolf's Amazon page lists a cache size that does not match Seagate's sheet. At 8TB the honest reading is that the Toshiba N300 8TB (HDWG780XZSTA on the listing, HDWG780EZSTA on the sheet) is a 7200 rpm, 512 MB cache drive that Toshiba rates at 5.74 W idle and 9.07 W operating, the highest of any 8TB in the table and 2.3 W above the WD80EFPX; it buys 281 MB/s and a 34 dB idle rating instead of 24 dB for the WD. For the low number at 8TB, the Red Plus EFPX or the IronWolf at 4.0 W is the pick, and the 12TB helium drives beat both. To learn what your own box draws, put a Kill-A-Watt on the cord, the instrument behind almost every wall figure above and the one I still do not own, or a metering smart plug that keeps a log. Some links are affiliate links; any commissions land on the public ledger.

FAQ

How many watts does a hard drive use?

Across 63 current 3.5-inch model numbers from WD, Seagate and Toshiba, the datasheets say 2.4 to 6.7 W idle (median 3.7 W), 3.1 to 9.07 W reading or writing, and 0.25 to 1.6 W in standby. A 2.5-inch laptop drive is about 0.5 W idle and 1.4 to 1.7 W active. Multiply by 8.76 for kWh per year.

Does a bigger hard drive use more power?

Not on the datasheets. Drives of 16TB and up idle between 3.4 and 6.7 W and drives of 8 to 10TB between 3.4 and 6.1 W. Helium-filled and 5400 rpm drives sit at the low end, air-filled 7200 rpm drives at the high end, whatever the capacity.

Does a hard drive use less power than an SSD?

No: a SATA SSD idles at about 0.03 W against 2.4 to 6.7 W for a 3.5-inch hard drive, roughly 100 times less. Under load the gap closes, because a fast NVMe drive is rated at up to 5.8 W reading, which matches a 7200 rpm hard drive at work.

How this was put together: the 73 rows were copied by hand from the 10 linked manufacturer PDFs on 2026-09-13, using each vendor's typical or average figure, mapped onto idle, read/write and standby; Toshiba publishes no standby value and Seagate and Toshiba do not state the fill gas, and those cells are marked absent rather than guessed. Medians and dollar figures are my arithmetic at EIA's June 2026 US residential average. The wall measurements are quoted from the linked TrueNAS, AnandTech and Hacker News threads as their authors wrote them. I own none of the hard drives named here and no plug meter, so nothing here is a measurement of mine; the one first-person number is my Mac mini's powermetrics reading from an earlier post, which is package power, not the wall.