Microsoft’s Outlook email was down for almost exactly a day. By the company’s own clock the outage ran from 11:17 a.m. Eastern on Thursday, November 12 to 11:10 a.m. on Friday, and Microsoft didn’t declare the incident closed until late Friday night.
The failure took minutes. What I want to look at is why the fix took 24 hours, because the reason is the same one that keeps a dead power grid from being switched straight back on.
What Microsoft says went wrong
Microsoft’s statement on Friday blamed “an issue during an earlier network capacity upgrade that surfaced following a recent network change.” It said the problem was technical and “not related to cyber or AI / agents.” It also said it was “deeply sorry.”
That’s all the company has said about the cause. It promised a full report, and as I write this on November 16 I can’t find one on its own pages. One trade site says a preliminary report points to a Windows bug on gateway servers. I couldn’t confirm that, so I’m not treating it as fact.
Why it took a day
The more useful record is the string of status updates Microsoft posted while it was happening. An admin-focused site kept a log of them, and they read like a restart that kept failing. (I’ve converted its times to Eastern.)
At 1:19 p.m. on Thursday, Microsoft reported that its servers were under elevated load, and an hour later it said the load wasn’t coming from extra traffic. At 2:43 p.m. the overloaded servers failed their own health checks and were pulled out of service. By 5 p.m. the company said its fixes hadn’t brought the relief it expected.
Then came the part I keep coming back to. Engineers put servers back, and around 7 p.m. those servers went unresponsive again as soon as traffic reached them. Every one that returned met hours of waiting connections and fell over.
So Microsoft changed its approach. It began limiting traffic in some locations and restarting equipment in stages, and by 8:50 p.m. it reported “positive results.” Recovery reached parts of Asia Pacific at 6:04 a.m. Friday. Europe and North America improved over the next few hours.
Grid operators would recognize all of this. After a blackout they add customers back a block at a time, because a generator handed too much load at once shuts itself off again. I went through that in what a black start is. Microsoft ended up doing the same thing with email: limit the load, bring back a piece, let it settle, bring back the next.
Big systems are built to run. Very few are built to start.
What it looked like outside Microsoft
More than 7.4 million problem reports were logged worldwide, according to ABC News in Australia, which drew on Reuters and AFP. In Britain, a GMB union official said NHS staff taking emergency calls were “unable to refer patient details in the normal way.” Immigration New Zealand stopped processing visa applications. Some companies couldn’t publish their earnings.
None of those organizations made a mistake on Thursday. Each had put one supplier underneath a job that couldn’t wait.
What I’d take from it
Ask one question of anything you depend on: if this went down for a day, what’s my other way? For email the answer can be as small as a second account with a different provider and a printed list of the ten phone numbers you’d need.
For a household it’s the same question asked about maps, medical references and how you’d reach each other. That’s why the kit we sell keeps all of it on the device, and why the setup guide is free. Restarting is the hardest thing a big system does, so plan for it to be slow.


