A Microsoft 365 outage began on the evening of November 12, 2026 (UTC) and ran deep into November 13. It hit Exchange Online, Outlook, Teams, Copilot, and several admin and security tools. Microsoft says a core process on its servers became overwhelmed and stopped responding. It sees no sign of a cyberattack. Early fixes brought only partial, short-lived relief. For millions of people, Friday began with an empty inbox and a spinning Outlook icon. Calendars would not load. Copilot would not answer. Teams still worked for basic chat, but anything tied to email stalled behind it. This post covers what we know so far: when it started, what Microsoft says went wrong, who felt it, and what the company is doing to bring services back.
Quick Answer:
Microsoft puts the start of impact at about 5:15 am AEDT on Friday, November 13. That is roughly 6:15 pm UTC on Thursday, November 12, or close to 11:45 pm in India.
The first public notice was short. Microsoft said users in the Americas were having trouble reaching Exchange Online mailboxes through more than one connection method. It later said any region could be affected. The problem touched business and enterprise customers, not Microsoft’s consumer products.
Outage tracker Downdetector saw reports climb quickly. By its count, Outlook complaints ran above 2,000 times their normal level at the peak, and nearly seven in ten reports came from the United States. The Week, citing the same tracker, put Outlook reports at close to 140,000 around 6 pm GMT on Thursday.
Microsoft’s first explanation was simple. A large number of servers had become unavailable without warning. That pushed their traffic onto the servers still running, and those began to struggle too.
As the day went on, the picture sharpened. Microsoft said a process on the affected servers, the one that sets up secure connections, was becoming overwhelmed. Once that happened, the server stopped handling requests. Microsoft also said failed health checks seemed to come from a component that processes them, so healthy-looking machines kept dropping out of the pool.
The company then pointed to a recent network configuration change that lines up with the start of the trouble. It began rolling that change back. Microsoft has not named it as the confirmed cause, and it has said other configuration updates are still under review.
One point is settled for now. A Microsoft spokesperson said there is no indication of a cyberattack. Australia’s Home Affairs Minister, Tony Burke, told reporters that Microsoft and the national cyber security coordinator do not believe they are dealing with a cyber event.
The common thread is Exchange Online. Anything that leans on it felt the hit.
| Service | What users saw |
| Exchange Online and Outlook | Mailboxes would not open on web, desktop, or mobile sync. Mail was delayed. |
| Microsoft Teams | Calendar, meeting scheduling, and some chat creation failed. Core chat and meetings mostly held up. |
| SharePoint Online | Files stored directly in SharePoint stayed reachable. Features that need Exchange, such as shared mailboxes and calendars, were unreliable. |
| Microsoft 365 Copilot | Copilot Chat would not load or reply. It also lost access to email, calendar, and files. |
| Defender and Purview | Portal pages failed to load. Safe Links could not check URLs. Several Purview services were disrupted. |
| Planner and To Do | Task operations failed or the apps would not load. |
| Microsoft Graph | API calls that read mail, calendar, and contacts timed out. |
| Power BI | Exports of reports with sensitivity labels to PDF failed at times. |
Other services on Microsoft’s list included Intune, Universal Print, Microsoft 365 admin center pages, and Windows 365 Cloud PCs that use Purview Customer Key.
The effects reached past Microsoft’s own apps. OpenAI and Anthropic both reported problems with their Microsoft connectors during the outage.
The time of day mattered. Because the trouble began in the US evening, offices in Asia and Australia hit it as the workday opened, while Europe and the Americas dealt with it overnight and into Friday.
Small businesses felt it fast. A furniture store in Karratha, Western Australia, said half its business runs on email and expected to be a week behind. An accommodation manager in the same town could not send invoices. A florist could not see same-day orders.
Many teams hit a quieter wall. When login codes are sent by email, a mail outage can lock people out of other apps too. Several workers told ABC News they could not get into tools unrelated to Outlook for that reason.
Public services took hits as well:
There was good news too. Queensland Health said it knew of no clinical impact. The Australian Energy Market Operator reported no disruption to energy systems. Australia’s government also activated a digital resilience agreement with Microsoft so local needs would get priority.
Microsoft has posted frequent updates on its service health page. Here is the sequence of work as it has described it:
1. Restarting the affected servers and returning them to service in small batches, checking each batch for stability first.
2. Switching to a different health check. A new method on a subset of servers let them return to service. Microsoft then began applying it more widely.
3. Building automated recovery to spot unresponsive servers and restart them on their own.
4. Rolling back a recent network change that matches the start of the incident.
5. Limiting traffic to affected locations, then adding it back in steps once servers held steady.
6. Rebuilding some servers on an earlier software configuration to see if that stops them from freezing.
7. Adding capacity by scaling up virtual machines to ease the load.
Recovery has been bumpy. Microsoft said servers it brought back could go unresponsive again once real traffic returned. Later, it said one change was “not taking as expected.” In its update around 4 pm AEDT, it reported a surge of calls from routing services to backend servers, and some machines stuck in a bad state that blocked its network change from syncing. It saw partial recovery in some regions, but not a lasting one.
At the time of writing, Microsoft had not given a restoration time.
You cannot fix Microsoft’s servers. You can limit the damage on your side.
Outages like this are a reminder that cloud tools are only as steady as the plumbing under them. Teams that plan for a bad day recover faster on the day it arrives.
