Who wakes your Mac at night?

Updated 6 November 2026

Mostly macOS itself. We read the power logs of 5 Macs. On one of them, over 6 nights, 47 of the 88 wake-ups came from 2 built-in services: search indexing and disk-encryption reporting. Another 15 came over the network, and the last 26 from smaller macOS services. None of them is a bug or an app you installed. On 2 of the 3 lab Macs most dark wakes came over Wi-Fi. macOS writes every wake-up down in a log you can read yourself.

The services that woke the Mac

On the M3 Max over 6 nights, the 88 wake-ups by who asked:

All 73 wake-ups that ran on a timer trace to a named service or daemon; the other 15 came from outside, so nobody asked. Over that whole week, day and night, the requests themselves were placed by a few system daemons: dasd, the task scheduler (76% of the marked ones), powerd (13%), PowerUIAgent (6%) and mDNSResponder, the network discovery service (5%). Those are messengers; the info= field names whose task it is. macOS scheduled 777 requests that week and marked 258 of them as next to fire; 90% of all requests named the task behind them. That is the everyday cost of a Mac that is ready when you open it.

2 Macs over a week of ordinary nights

Both in everyday use. Wake-ups are counted from the sleep and wake events between 23:00 and 07:00.

On the M3 Max all 88 wake-ups were dark wakes, with no screen and no sound. The M1 Pro woke 10 times as often, but each wake-up lasted 5 seconds instead of 45.

3 lab Macs over 1 quiet night

Test machines without iCloud or third-party services, on battery for 9.5 hours with the lid open and the display off, each falling asleep on its own with Power Nap on. Here a dark wake is a DarkWake line in the log, a different count from the wake-ups above, so compare these 3 Macs with each other.

Same setup, same night, and the count runs from 36 to 121. Most of the gap is the network: 2 of the Macs woke over Wi-Fi 64 and 88 times, the third not once. The night before, each of them had 1 AI app left open after a single prompt; that comparison is on the AI page.

Who woke each Mac

On the 3 lab Macs, none of the Wi-Fi wake-ups had a marked request due at that moment: they come from the network. The rest ran on a timer, 37, 33 and 36 wake-ups that night, and all 106 matched a marked request due within 1 second.

The M3 Max block is a later night of everyday use, 23:00 to 07:00, on the charger, so it does not compare with the lab night: 32 wake-ups on a timer, all matched, and 4 over Wi-Fi. One of the 32 was marked before 23:00, so its wake-up counts here and its mark does not.

A mark is not a wake-up. Each Wi-Fi wake-up makes macOS print the schedule again, the next task's wakeAt= often moves a little, and that counts as a new marked request. So com.apple.appstored.Daily collected 97 marks and woke the Mac 27 times. On the macOS 14.2.1 Mac, with no Wi-Fi wake-ups, marks and wake-ups are equal. The M3 Max table at the top of this page counts wake-ups, not marks.

The top name changes from night to night. The night before, with 1 AI app left open on each Mac, it was com.apple.MobileAccessoryUpdater.deviceIdleCheck (26 of 38 timer wake-ups), com.apple.chronod.nextScheduledTimelineRefresh (17 of 34) and com.apple.rosetta.report (9 of 37). The shape held on both nights: on 2 Macs 1 task took half or more of the timer wake-ups, on the macOS 14.2.1 Mac no name reached 25%.

Read the log on your own Mac

No app needed: macOS keeps the log itself. In Terminal:

pmset -g log > ~/Desktop/power.txt

Lines containing DarkWake are the silent wake-ups. Lines labelled Wake Requests carry process= (which daemon asked), wakeAt= (when) and often info= (whose task it is).

Count each request once. The list is a schedule, reprinted in full every time the Mac goes to sleep, so a single request appears many times: identify it by its info= and wakeAt= together. The request macOS has lined up next carries an asterisk before process=; count the asterisks, not the lines. To confirm that a marked request really woke the Mac, match its wakeAt= to a DarkWake line with rtc in its reason, within 1 second. We did that on the 3 lab Macs; on the M3 Max, WattMate had made the same match when it read those nights.

The direct download of WattMate reads the same log for you and turns a night into plain language: how long the Mac slept, how often it woke and who asked. The latest night is free; earlier ones come with Pro. The App Store edition can't read this log, so it shows how long the Mac slept but not who woke it.

How it was measured

Our own measurements on 5 Macs, the M3 Max twice, read from each Mac's power log.

What this doesn't show

• 5 Macs, not a population. Yours will differ: more calendars, more mail accounts or a busier network mean more wake-ups; Power Nap off and Wi-Fi disconnected mean fewer. Read the numbers as a picture of what the log holds, not as an average for all Macs.
• No price per wake-up. The battery's finer reading, in milliamp-hours, is published only about once a minute, wobbles and bounces back after a wake, and a 45-second wake is lost inside that. What can be measured is the count, the duration and the requester's name.
• No price for the night either. On the 3 lab Macs the morning battery reading came within 33 seconds of waking, when that reading bounces, so we print no figure in milliamp-hours. The percentage in the log went from 80 to 76, from 80 to 79 and stayed at 100, too coarse to rank them.
• Silent doesn't mean harmful. Dark wakes are the design working as intended. The reason to look is the outlier: a service that wakes the Mac hundreds of times or holds it awake for minutes shows up in this log and nowhere else.

Methodology and limits for everything we measure: how we count. Related: what to expect after the macOS 27 update · does quitting apps actually help? · where the charge goes with the lid closed