The Microsoft 365 outage of November 12, 2026 took Outlook, Teams and SharePoint away from many users because several of its servers went offline. On the 13th, Claude had sign-in and performance errors; only one of those, the Microsoft 365 connector, was tied to Microsoft. Here is what the official sources say, what is still unknown and how to prepare your business.
Information as of November 13, 2026. Both platforms were still recovering as we wrote this; check their status pages for the latest.
What happened in the Microsoft 365 outage?
A number of servers went offline and traffic overloaded the rest. Microsoft opened advisory MO1550821 on Thursday, November 12 at 1:26 p.m. Colombia time (the same as New York in November), according to BleepingComputer. The affected services were Exchange Online (Outlook email), Teams, SharePoint Online, Microsoft 365 Copilot and Microsoft Defender XDR.
Microsoft first pointed to elevated load on its infrastructure. It then said the cause was "a number of servers being unexpectedly unavailable". It paused deployments, rebalanced traffic and began restarting the affected servers.
On Friday the 13th at 8:05 a.m. it reported "encouraging recovery progress" and extended to 48 hours the time it keeps retrying delayed email. It has not yet confirmed a final root cause.
What happened with Claude?
There were three separate incidents on Anthropic's status page. Times are Colombia time.
| When | What failed | What Anthropic says |
|---|---|---|
| Nov 12, 12:52 p.m. | Microsoft 365 connector and Claude for Outlook | An upstream provider problem; at 5:23 p.m. it points to a Microsoft 365 service incident |
| Nov 13, 4:58 a.m. | Sign-in on claude.ai and Claude Code | Fix deployed; resolved at 6:58 a.m. |
| Nov 13, 8:09 a.m. | Elevated errors on claude.ai, the API and Claude Code | At 9:36 a.m. claude.ai and the API are recovering, but errors may return and connectors remain degraded |
Anthropic links only the first incident to Microsoft. For the other two it has not published a cause.
Why does one tool's failure drag others down?
Because one tool can depend on another without you noticing. Think of a row of dominoes: Claude reaches your email and calendar through Microsoft 365. If Microsoft fails, that connection fails even though Claude itself is running.
This applies to any business. Your online store depends on the payment gateway, your sales email depends on your domain provider and your support agent depends on an API. We saw it in the AWS outage and in the Cloudflare outage of November 2025.
How do you prepare a simple backup plan?
With three habits you can start today without buying anything:
- Draw your dependency map. List each critical tool (email, sales, payments, support) and what it depends on. If two of them rely on the same provider, that is your weak point.
- Keep a second channel for urgent things. If email goes down, how do you tell a customer their order ships today? A WhatsApp Business number or a phone call is enough. Define it before the failure, not during.
- Follow the official status pages. Subscribe to the ones for your key providers. They tell you whether the problem is theirs or yours and save hours of restarting equipment for nothing.
If you use AI with your customers' data, also check what information each connection shares; we explain it in AI and customer data in Colombia.
When you do NOT need to worry
When a tool is convenient but not critical. If you could not use an AI assistant for a few hours, your business would keep selling. Sales email, payments and your website do deserve a backup plan. Spend your time there.
Wrap-up
A giant's outage is rarely your fault, but it is your problem: what counts is how fast you keep serving customers. Start with the dependency map and a second channel. If you want a second pair of eyes, our technology consulting does it without jargon.
Frequently asked questions
How long did the Microsoft 365 outage last?
It began Thursday, November 12, 2026 at 1:26 p.m. Colombia time, and on Friday the 13th at 8:05 a.m. Microsoft described gradual recovery without confirming a date for full restoration.
Was the Microsoft 365 outage a cyberattack?
Microsoft has not said so. It said the cause was servers unexpectedly unavailable and that a final root cause was not yet confirmed.
Were the Claude errors caused by Microsoft?
Only one of them. Anthropic attributed the Microsoft 365 connector failure to a Microsoft service incident. It has not published a cause for the sign-in and performance problems of November 13.
Where do I check whether a tool is down?
On the provider's official status page, such as [status.claude.com](https://status.claude.com) for Claude. It is more reliable than social media because the company itself updates it.
