Quick Answer: A Microsoft 365 outage that began around 18:00 UTC on Thursday, 12 November 2026 disrupted Exchange Online and every service that depends on it, including email, Teams meetings, cloud PCs, and Copilot. Microsoft's fixes stalled because scaled-up virtual machines got stuck in a bad state, so changes never synced. Services that don't rely on Exchange Online stayed up.
A Microsoft 365 outage that runs past 15 hours is more than an inconvenience. For organizations that run email and calendars through Redmond's cloud, it stops the working day. The Register reported that the disruption, which Australian users were already calling a "Microsoft long weekend," kept going even after Microsoft announced fixes. Here's what actually broke, why the repairs didn't stick, and what you can do while you wait.
What Actually Broke (and What Didn't)
The outage hit Microsoft 365 services that interact with Exchange Online. Users couldn't reliably reach email, join Teams meetings, use cloud PCs, or access Copilot. Microsoft said roughly 15 other hosted services were affected as well.
The scope matters. Microsoft's own language made clear that services that don't need Exchange Online kept working. If your organization uses SharePoint or OneDrive without relying on mail flow, your experience may have been far milder than your inbox suggested.
Here's the practical takeaway. Before you assume the whole tenant is down, check which workloads are actually failing. In an outage like this, that distinction determines whether your staff can keep working on documents or are stuck waiting.
Why the "Fix" Kept Failing
Microsoft's first explanation was a "problem that's causing servers to fail automated health checks and become unavailable." That's a symptom, not a cause. Early in the incident, the company thought it had found a fix, then admitted that "actions taken thus far have not provided the anticipated relief."
The most revealing detail came in a later update. Microsoft said it was scaling up virtual machines to add capacity and reduce load, but "we are encountering issues with the machines being stuck in a bad state, preventing our changes from syncing as expected."
This is a failure mode worth understanding. When you add capacity to a degraded backend, the new machines have to pass through the same broken provisioning and configuration path as everything else. If they stall, your fix never reaches the servers that need it. The result is a loop: the team applies a change, the change doesn't propagate, and users see partial recovery that doesn't hold.
Microsoft's own wording reflected this. It reported "partial recovery across regions" but said "these actions aren't resulting in sustained recovery." That's the signature of a recovery problem, not a capacity problem alone.
The Cyberattack Question and Why the Wording Changed
Early on, a major outage at a cloud provider invites a cyberattack theory. Microsoft's language on that point shifted during the incident, and the change is worth reading closely.
At 04:11 UTC, the company said it could find "no indication this is the result of malicious activity." A later update, time-stamped 05:46 UTC, went further: "we are not aware of any malicious activity that is causing this incident."
Notice what these statements do and don't say. They describe what Microsoft has found so far. They don't rule out an attack, and they don't establish a root cause. Treat them as a current assessment that can still change. Once a provider has confirmed a cause, the post-incident review should be the document you rely on.
Here's where most people get it wrong: they read "no malicious activity" as "not a security incident" and stop thinking about it. For your own environment, keep the security question open. If you see unusual sign-ins, mailbox rule changes, or forwarding rules appearing during an outage, log them and report them, whatever the provider says.
Reading the Status Feed, Not Just the Status Page
Here's a counter-intuitive finding that most outage guides miss. The public status page is the wrong place to reconstruct what happened. As The Register noted, Microsoft's status page shows only the most recent update. Anyone who checked it at 12:00 UTC saw one message and no history of the failed fixes that came before it.
The RSS feed behind the status page preserves every timestamped update. The Register distilled that feed into a chronological list, and it tells a very different story: several rounds of "this is improving" followed by "these actions aren't resulting in sustained recovery."
If you're managing communications during an incident, that gap matters. Staff and executives who only read the latest status line will think recovery is closer than the timeline suggests. Subscribe to the RSS feed, keep a running log, and quote the actual timestamps when you brief people.
What to Do When Your Tenant Goes Dark
Waiting is not a plan. Use the outage to run a short checklist, in order:
- Confirm the scope in your admin center. Open the Microsoft 365 admin center and check the Service Health view for advisories that apply to your tenant. Don't rely on public news alone.
- Tell users which services work. If SharePoint or OneDrive still function, say so. That single message prevents a flood of helpdesk tickets.
- Activate your fallback channel. Use a pre-agreed alternative such as a phone bridge, a chat tool outside Microsoft 365, or a shared status doc. Decide this before the next outage, not during it.
- Log timestamps. Record when users first noticed, when they lost access, and when access returned. You'll need this for credits and your own review.
- Hold off on mass fixes on your side. Re-syncing mailboxes, reconfiguring connectors, or reissuing credentials during a provider-side incident can create a second problem. Wait for the provider's confirmed recovery.
The last step is the one teams argue about most. Re-running sync jobs feels productive. In a backend incident, though, extra load on already stressed infrastructure can slow the very recovery you're waiting for.
Comparing Your Outage Information Sources
No single channel tells you everything. This table shows what each one is good for:
| Channel | What it shows | Main limitation | Best use |
|---|---|---|---|
| Public status page | Most recent update only | No incident history | Quick check of current status |
| Status page RSS feed | Every timestamped update | Requires a feed reader | Building the incident timeline |
| Microsoft 365 admin center Service Health | Advisories scoped to your tenant | Requires admin access | Confirming your organization's impact |
| Your own mail-flow monitoring | Real delivery failures on your domain | You must build and maintain it | Independent detection and SLA evidence |
If you only have time to set up one thing, make it independent monitoring. The provider's status page tells you what the provider admits. Your own monitoring tells you what your users actually experience.
Setting Realistic Expectations on Uptime and Credits
Microsoft's service level agreement for Exchange Online commits to 99.9% uptime. That sounds strict until you do the arithmetic: 0.1% of an 8,760-hour year is about 8.8 hours of allowed downtime. A single incident running past 15 hours blows through that allowance on its own.
That said, the SLA is a financial remedy, not a guarantee of service. Service credits typically depend on a claim you file and documentation you supply. Check your contract and your admin portal for the claim process, and keep the timestamps from your own log. Don't assume credits will appear automatically.
The nuance here matters. Credits rarely compensate for lost productivity, missed deals, or the scramble to reach clients. Treat them as a partial recognition of the failure, and plan your continuity around the outage you'll inevitably face.
Frequently Asked Questions
What is a Microsoft 365 outage?
A Microsoft 365 outage is a period when one or more cloud services, such as Exchange Online, Teams, or SharePoint, become unavailable or degraded for users. Outages can be partial, affecting only some regions or workloads, or full. You can confirm one in your admin center's Service Health view or through Microsoft's public status channels.
How do I check Microsoft 365 service health status?
Open the Microsoft 365 admin center and go to Service Health for advisories specific to your tenant. For public, cross-customer updates, use the status page and its RSS feed. The RSS feed is more useful for incident timelines, because the status page shows only the latest update.
Why did Microsoft 365 go down?
Microsoft initially described a problem causing servers to fail automated health checks. Later updates pointed to virtual machines stuck in a bad state, which blocked the configuration changes meant to restore service. The company said it was still identifying the underlying cause and had found no evidence of malicious activity at the time of its updates.
Can I get credits for a Microsoft 365 outage?
You may be eligible for service credits under Microsoft's service level agreement, but you typically need to file a claim. Keep accurate timestamps of when your users lost and regained access. Credits generally don't cover lost productivity, so plan your continuity measures separately.
Are Teams and SharePoint affected when Exchange Online goes down?
Not always. Microsoft said the outage affected services that interact with Exchange Online, so Teams meetings and some other features were impacted, while services that don't depend on Exchange Online remained functional. Check which workloads are failing in your own tenant rather than assuming the whole suite is down.
Once the dust settles, the most useful step is to write down what your team learned. Run a 30-minute review this week, document which workloads failed, and pick one fallback channel to set up before the next incident. Pass this checklist to whoever owns your Microsoft 365 administration, especially if they're still stuck trying to explain a "long weekend" to the business.