Last week Outlook, Teams calendars and Copilot were down worldwide, Latvia included, for a day and a half. Microsoft says a technical fault caused it, not an attack or AI. Any company that runs on one supplier's tools can learn from it.

Quick answer
  • The Microsoft 365 outage started on Thursday 12 November at about 18:00 Latvian time. In Europe services recovered on Friday afternoon, and fully on Saturday morning.
  • Microsoft calls it a technical issue not related to a cyberattack or to AI.
  • Companies can claim an SLA credit. Microsoft must receive the claim by 31 December.
  • The bigger point: spend 30 minutes writing down what stops in your company if email is down for 36 hours.

What happened?

On Thursday 12 November, at about 18:00 Latvian time, Microsoft 365 users could no longer send or receive email, open calendars in Outlook and Teams, or load Copilot Chat. Downdetector logged more than 130,000 reports of Outlook problems, and within the first hour they came from 74 countries. In Europe services began to improve steadily on Friday afternoon, and Microsoft restored everything on Saturday morning.

Microsoft's preliminary explanation is a failure in the systems that set up connections to Exchange Online. It surfaced after an earlier network capacity upgrade. The company stresses that it was a technical issue, not related to a cyberattack or to AI agents. Microsoft has promised a full incident report later.

What does it mean for Latvian companies?

Most mid-sized Latvian companies run email, calendars, files, Teams meetings and staff sign-in on one Microsoft 365 account. When it stops, more than email stops. You can't approve an invoice that arrived by email. You can't find the meeting link. In some companies you can't even reset a password, because the reset code goes to the same mailbox.

The outage showed one more thing: Copilot went down together with email. Many companies start using AI with the tool already in their Microsoft licence. That's an understandable first step, but it puts AI in the same basket as email and documents. Once AI is part of everyday work, one supplier's outage stops both the people and their AI assistants.

That doesn't make Microsoft a bad choice. Cloud services are generally more reliable than a server in the office cupboard. The problem is that most companies have nothing written down about what happens on the day they stop working.

Our view

You don't need to leave Microsoft. You need to know what your company depends on before the next outage shows you. Nobody in a factory is surprised by a backup generator, even though it runs once a year. In a digital office there's usually no plan B at all.

The more work you hand to AI, the more it matters that your data, instructions and processes belong to you and aren't locked inside one supplier's tool. Then on the day of an outage you switch to another tool instead of waiting.

What to do now

  • If you haven't thought about it yet: 30 minutes with your management team. Write down what stops if email is down for 36 hours, how you reach your key clients and suppliers without Outlook, and where their phone numbers are.
  • If you already use AI: list which AI tools and routines depend on the same supplier as your email. Check that password reset and two-step sign-in codes don't go only to that mailbox.
  • Everyone: check whether you're owed a Microsoft SLA credit. Microsoft must receive the claim by the end of the following calendar month, so by 31 December. You file it with Microsoft support through the Microsoft 365 admin center or through your licensing partner.

For administrators

  • The Exchange Online incident ID is EX1550543. Put it in the claim together with your tenant ID.
  • Microsoft's SLA asks the claim to include a description of the incident, the time and duration of the downtime, the affected resource names, the number and location of affected users, and a description of the errors. A claim without this can be rejected.
  • The credit is prorated by the length of the outage and covers only the affected services. Advisories usually don't count.
  • If you buy licences through a partner (CSP), the partner files the claim. Ask them this week, because a partner may have its own, earlier deadline.
  • Free services and previews aren't eligible for credits.

If you'd like someone from outside to look at where your company has a single point of failure, book a free 30-minute call.

Sources

  • Microsoft 365 Status: https://x.com/MSFT365Status
  • LETA / iFinanses, 13 November 2026: https://ifinanses.lv/zinas/actual-trauceta-microsoft-darbiba/32063
  • webhosting.today, "Microsoft points to a latent bug behind its day-long Outlook outage", 16 November 2026: https://webhosting.today/2026/11/16/microsoft-points-to-a-latent-bug-behind-its-day-long-outlook-outage/
  • Microsoft, "Service Level Agreement for Microsoft Online Services", 17 November 2026: https://www.microsoft.com/licensing/docs/view/Service-Level-Agreements-SLA-for-Online-Services