Analysis · Microsoft 365 outage, November 2026

Microsoft Outlook and 365 outage: what happened, what Microsoft has said and what is still unknown

On Thursday 12 November 2026 people lost access to Outlook and Exchange Online mailboxes, first in the Americas and then in other regions, and Microsoft later said Teams, SharePoint Online, Copilot and Defender could also fail to connect. Downdetector reports peaked at roughly 130,000 for Outlook. Microsoft said it saw no sign of a cyberattack, first blamed elevated load, then said servers had become unavailable, and was still working on recovery on Friday 13 November with no restoration time. No root cause has been published. This piece sets out the timeline from the outlets that covered it, how Microsoft's explanation shifted, who was affected, what remains unconfirmed and what a business that depends on email should take from it.

Updated · 10 min read

01What happened, in short

PointWhat was reportedSource
StartProblems began shortly before 1 p.m. ET on Thursday 12 November; the Microsoft 365 incident is logged from 1:26 p.m. ETCNBC; BleepingComputer
First symptomUsers unable to reach Exchange Online mailboxes through several connection methods, in the AmericasBleepingComputer; SecurityWeek
SpreadAsia-Pacific and Europe: email was unavailable through much of the Asian business day and into the European morningCNBC
Services named by MicrosoftExchange Online, Teams, SharePoint Online, Microsoft 365 Copilot and Defender XDRBleepingComputer; SecurityWeek
Incident IDsEX1550543 (Exchange Online) and MO1550821 (wider Microsoft 365)SecurityWeek; BleepingComputer
Cyberattack?Microsoft: “we are not aware of any malicious activity that is causing this incident”CNBC; PPC Land
Status on FridayBoth incidents still open; users “may still be unable to connect”; no estimated time for full recoveryCNBC; CIO&Leader; PPC Land

02The timeline

Times are Eastern, as the outlets give them. Where a source does not give a time, the row says so.

WhenWhat happenedSource
Thu, about 9:45–10:53 a.m.A separate Teams incident (TM1550488), attributed to resource contention, affects North and South America and is later marked resolved. Microsoft has not said whether it is relatedBleepingComputer
Thu, shortly before 1 p.m.Connectivity problems begin; the Microsoft 365 incident is logged from 1:26 p.m. Windows Latest, via PPC Land, had reason to believe limited problems began earlier in the dayCNBC; BleepingComputer; PPC Land
Thu afternoonComplaints surge. Microsoft says it is investigating Exchange Online mailbox connectivity in the Americas, then cites “elevated load across multiple regions”BleepingComputer; CNBC
Thu, just before 3 p.m.Microsoft says affected servers are failing automated health checks, is testing a workaround that bypasses the component that processes them, and is looking at elevated processor usageSecurityWeek
Thu, 4 p.m.Microsoft says it validated an “alternative service health check method” that brought some servers back and is applying it more broadly, with automated recovery to detect and restart unresponsive serversCNBC
Thu, about 6–7 p.m.Microsoft says it is “returning a portion of affected servers to the service”, which “remain stable so far”, and describes “limited recovery”. It warns recovered servers can become unresponsive again once they receive trafficBleepingComputer; CIO&Leader
Thu, 9 p.m.Restarting servers is “showing signs of improvement”; recovery is not confirmedSecurityWeek
Thu, 11 p.m.Downdetector reports for Outlook are down to about 20,000, from a peak of about 130,000CIO&Leader
Fri, about 2:30 a.m.Users “may still be unable to connect”. Microsoft cites “positive results for server stability” after a patch on some infrastructure, gives no restoration time and says it will provide a full report once the problem is mitigated. Some Teams access is back; Outlook on web and desktop is still down for someCNBC
Fri, about 12:46 a.m. (05:46 UTC)Microsoft's status page, as copied by PPC Land: still pursuing “multiple workstreams to pinpoint the underlying cause of the traffic influx”, with the next update due by 7 a.m. UTC. PPC Land counts more than 11 hours since problems surfacedPPC Land

03How Microsoft's explanation changed

The cause was not fixed in one statement. The sequence below is the order the outlets report it; it matters because each step changes what a customer can reasonably do.

  1. 01Elevated load. Microsoft first cited “elevated load across multiple regions” (CNBC, BleepingComputer).
  2. 02Not more traffic, fewer servers. It then said the load was “not driven by an increase in traffic, but by a number of servers being unexpectedly unavailable”, which strained the rest of the infrastructure (BleepingComputer, CNBC).
  3. 03Health checks. Affected servers were failing automated health checks. CIO&Leader reports that the checks kept pulling servers out of service, and Microsoft's answer was an alternative check (SecurityWeek, CNBC, CIO&Leader).
  4. 04A connection process under strain. Microsoft traced failures to a core process that sets up secure connections and became overwhelmed, so affected servers stopped handling requests (BleepingComputer; CIO&Leader).
  5. 05A traffic influx again. On Friday, the status page copied by PPC Land spoke of pinpointing “the underlying cause of the traffic influx” and recovering backend servers. PPC Land notes this sits awkwardly beside Thursday's “not driven by traffic”. Microsoft may be describing returning traffic hitting recovered servers, which it warned about on Thursday evening; we do not know, and neither source says.

Mitigation, as reported: deployments paused, connection traffic redistributed worldwide and moved off affected systems, servers restarted, temporary traffic limits in some locations, and a configuration change tested on a subset of servers before wider rollout (BleepingComputer; CIO&Leader).

04What stopped working and who felt it

PPC Land lists 19 services on Microsoft's status page, all tied to dependence on Exchange Online. The failures were partial, which is why the experience differed so much from one customer to the next.

ServiceWhat was reportedSource
Outlook and Exchange OnlineMailbox access failing on web, desktop and mobile; Calendar and external services that connect to Outlook features also hitCNBC; CIO&Leader
TeamsCalendar, live events, app chat and contacts affected; calls, meetings and standard-channel messages still workingPPC Land; CIO&Leader
SharePoint Online and Microsoft 365 appsPartial; files still accessible, recently used files affectedPPC Land
Defender for Office 365Safe Links fails to open rewritten links; quarantine release failsPPC Land
Microsoft GraphCalls that touch Exchange time outPPC Land
CopilotAffected, including the OpenAI and Anthropic connectors for Microsoft 365PPC Land; CIO&Leader
Admin toolsMicrosoft 365 admin center affected; CIO&Leader names status.cloud.microsoft as a fallback for administratorsPPC Land; CIO&Leader
AzureStatus page: “No Azure issues detected at this time”PPC Land
  • Scale. Downdetector reports peaked at roughly 130,000 for Outlook, one tally put the high at 138,058, and reports covered 74 countries (CIO&Leader). PPC Land, citing ABC News Australia, gives more than 120,000 for Outlook and more than 45,000 for Microsoft 365 on the US site. These are user reports, not a count of affected customers.
  • Public bodies. New York City Emergency Management warned of delays reaching city agencies by email, and Service NSW in Australia asked customers to delay visits to its centres (CIO&Leader).
  • Companies. Workhorse Group moved its third-quarter results from 12 to 16 November citing the outage (CNBC, CIO&Leader), and CNBC says Richemont's Friday earnings were slightly delayed. Air India temporarily rerouted emails and said flight operations were not affected (CNBC).
  • A comparison. Australia's Home Affairs and Cyber Security Minister Tony Burke said Microsoft believed it was dealing with something “in some ways similar” to the 2024 CrowdStrike outage, per Reuters via CNBC. That is the minister's description of Microsoft's view, not a finding.
  • Other providers. CNBC says AWS and Google users also reported problems on Downdetector on Thursday; AWS said its services were “operating normally” and that an unrelated disruption elsewhere was prompting inaccurate reports. SecurityWeek separately says some Google Workspace users had trouble reaching Google-hosted mail through Outlook clients. Neither shows a Google or AWS outage.
  • Consumers. The status page also marked Microsoft consumer products as degraded. PPC Land notes service credits are unlikely to apply to personal accounts.

05Why one mail system can take so much down

This section is our reasoning from the symptoms above, not Microsoft's finding. Many Microsoft 365 features read or write mailbox data: calendars, meeting scheduling, Copilot's access to your messages, security quarantine, compliance search, Graph calls. When the mailbox layer does not answer, features built on it stall even though the products themselves are up. That matches the pattern in the table: Teams calls worked while Teams calendars did not, and files opened while sensitivity-label operations failed.

The reported cause also matches a familiar failure shape. If some servers fail a health check, the system removes them; the remaining servers take more work and may fail their own checks; and a recovered server that is handed its full share of traffic at once can fall over again. Microsoft's own warning on Thursday evening, that recovered servers could become unresponsive when traffic returned, is that last step. We cannot say that is what happened here.

06What a business can do

These are general practices, informed by what the reporting describes. They are not Microsoft's instructions.

  • Keep a second channel. CIO&Leader's advice, which is the outlet's own and not Microsoft's, is to hold an out-of-band way to reach staff and customers that does not run through the same mail system.
  • Know where the real status is. Administrators can use the Microsoft 365 Service Health page and, if the admin portal is down, status.cloud.microsoft (CIO&Leader).
  • List what depends on email. Sign-in codes, invoice reminders, approval requests and alerts often arrive by email. If the mailbox is unreachable, work that waits on it waits too.
  • Do not assume sent means read. During the outage a message could be delivered yet not readable by its recipient. Chase time-critical items by another route.
  • Wait for the report. Microsoft has committed to full details of the root cause and repair items in a Post-Incident Report once the incident is mitigated (PPC Land; CNBC). Service-credit terms are in your agreement, not this article.

07If you take payments and your inbox went dark

An email outage does not stop a payment from being made. The question is what your business learns, and when. These points describe how Payaries works, from its integration guide; they do not say Payaries was or was not affected by this outage.

  • Status does not depend on your inbox. Invoice status is on the invoice itself and is delivered to your server by webhook when it changes, so an order system that listens for the webhook does not need anyone to read an email. You can also fetch an invoice by its id. See verifying crypto webhook signatures.
  • A late-noticed payment is still on time. If a customer pays before the deadline and the system only notices after it, for example during a service interruption, it is treated as on time and credited through the normal flow, per the integration guide.
  • Some account steps use email. Confirming a new account is done with a one-time link, valid for 24 hours, and a new link can be requested. If that email sits in an inaccessible mailbox, the link can be re-sent once the mailbox is back.

How crypto invoices and settlement work is in how a crypto payment gateway works. This describes a payment flow; it is not an assurance about any third party's availability.

08What we could not verify

09Sources and as-of date

Questions

Was the Microsoft Outlook outage on 12 November 2026 a cyberattack?
Microsoft said it was not aware of any malicious activity causing the incident, and CIO&Leader reports an executive vice-president saying it was not a cyberattack. No root cause had been published when we wrote this.
When did the Microsoft 365 outage start?
Problems began shortly before 1 p.m. ET on Thursday 12 November 2026, and the Microsoft 365 incident is logged from 1:26 p.m. ET. Windows Latest, via PPC Land, had reason to believe limited problems began earlier that day.
Which Microsoft services were affected?
Exchange Online and Outlook first, then Microsoft named Teams, SharePoint Online, Microsoft 365 Copilot and Defender XDR. PPC Land lists 19 services tied to Exchange Online. Teams calls and standard-channel messages largely kept working.
What did Microsoft say caused it?
Microsoft first cited elevated load, then said a number of servers were unexpectedly unavailable, then said a process that sets up secure connections became overwhelmed. Servers were also failing automated health checks. A recent network configuration change is reported but not confirmed as the cause.
Is the Outlook outage over?
Not as of the reports we reviewed on Friday 13 November: both incidents (EX1550543 and MO1550821) were open and Microsoft gave no restoration time. Check the Microsoft 365 Service Health page for current status.
Does an Outlook outage stop crypto payments?
No. A payment is made on the blockchain and does not need your inbox. What you lose is anything that arrives by email. On Payaries, invoice status changes are sent to your server by webhook, so an order system that listens for them does not rely on someone reading mail.

Next

Run a real invoice before you commit

Test keys work immediately, with no approval step. Payaries runs on real mainnet networks, so use a small real payment to test.