Signals/15 Nov 2026
Microsoft 365 outage lasts over 22 hours worldwide
A Microsoft outage lasting more than 22 hours hit Outlook and other Microsoft 365 services in over 70 countries. Microsoft says automated checks, not an attack, caused it.
In plain words
A Microsoft outage lasted more than 22 hours and cut off email and other Microsoft 365 services for people in over 70 countries. Microsoft says it was a technical failure and not a cyber-attack. Governments, courts and airline desks were affected, not only companies. If your work depends on one provider for email, it is worth having a second way to reach customers and colleagues. Plans for the day a key service fails are covered in our Day Zero plan.
What happened
The BBC reports that a critical Microsoft outage lasting more than 22 hours disrupted Microsoft 365 services such as Outlook. According to the report, the cause was routine automated checks that took some servers offline, which overloaded the remaining servers with traffic.
Recovery took nearly a full day while engineers gradually restored services. A Microsoft spokesperson said that "given the scale of demand" it restored services gradually, and that although this took longer it "helped ensure a stable restoration". Microsoft told the public late on Friday (GMT) that there was no "ETA" for full recovery.
The BBC says around 430 million people globally use Microsoft 365 and that issues were reported across more than 70 countries. Named effects include email systems for the US House of Representatives and Australian intelligence agencies, court systems in Singapore and airline support desks in India. Microsoft has apologised and says it is reviewing what went wrong, with the findings expected to inform steps to prevent a repeat.
What is confirmed and what is not
Confirmed by the BBC's account, with Microsoft as the source for the cause: the duration of more than 22 hours, the cause as automated checks and server overload, the apology, and the statement that it was not a cyber-attack. The scale figures (430 million users, more than 70 countries) and the named affected bodies are reported by the BBC.
Claimed rather than established: the view of Gergely Orosz that regional splitting would limit the impact, and Walter Pasquarelli's view that AI assistants plugged into email may spread such failures further. Both are opinions.
Unknown: the technical detail of why the automated checks removed servers, why recovery was slow, what the final review will find, and any figures for losses. The BBC report gives no date for the review's publication.
- 1Small and medium businesses that rely on Microsoft email9
- 2Government bodies and courts using Microsoft 3658
- 3Airline support desks and other service operators6
- 4Organisations linking AI assistants to email and files5
- 5Microsoft and its engineering teams7
Analysis
Businesses that run on one email provider lost their main channel
The most direct consequence falls on organisations for which Microsoft email is the working channel. Flora Hamilton of the Small Business Charter says that for many small and medium-sized enterprises, email access is fundamental to communicating with customers and suppliers and to generating income. The mechanism is simple: when one provider carries mail, calendars and files, an outage stops all three at once, and a small firm often has no second route. The BBC report gives no loss figures, so the scale of financial harm is unknown. We are confident in the mechanism and have no evidence on the size of the cost.
Public institutions and critical services
The BBC lists the US House of Representatives, Australian intelligence agencies, Singapore courts and Indian airline support desks. These are different kinds of dependence. Email for a legislature or an intelligence agency is an internal communications loss. Court systems and airline desks touch members of the public directly. The report does not say how long each body was affected or what workarounds they used, so the effect on each should not be assumed equal to the global headline of 22 hours.
Security
Microsoft says the cause was a technical failure and not a cyber-attack. That is a statement from the company, and the BBC gives no independent check. If it holds, the event is an availability failure and not a data breach. The security lesson is still real: a service that agencies rely on can disappear without any attacker, and an attacker who wanted the same effect would find a single point of failure. That second point is general reasoning, not something the sources show. Readers should also expect phishing that uses an outage as cover, a common pattern in general, though not reported here.
Technology and industry
The stated mechanism is worth reading closely. Routine automated checks took some servers offline, and the remaining ones took the load. A safety routine thus became the trigger, which is a known class of failure in large systems where protective actions shift load onto what is left. Orosz notes such outages are not unheard of, and what was unusual was the recovery time. Microsoft attributes the slow restoration to gradual recovery given the scale of demand. These two accounts fit together, but the BBC does not show that gradual restoration was the only option. A sceptical engineer would ask why checks could remove enough capacity to overload the rest, and whether limits on how many servers they may remove exist. Microsoft's review should answer this.
Orosz proposes splitting services per region, as the report says already happens with Amazon's cloud service AWS. If that works, a fault would stay inside one region. The trade-off, which the sources do not discuss, is added cost and complexity, and TOSWO Research cannot judge from the report whether it would have contained this fault.
AI assistants and growing dependence
Pasquarelli says that as AI assistants are plugged into email, calendars and files, a failure like this may spread further. The reasoning is that an assistant built on top of a mailbox inherits the mailbox's outage. This is his forecast, not a result. Our reading is that it holds only where the assistant has no fallback source for the same data. Earlier coverage of a related event is in our note on the Microsoft 365 and Copilot outage and on the 12 November disruption, and the pattern of repeat incidents can be browsed in the incident record.
Law and policy
The BBC frames the story as a question of how much vital services rely on a handful of major firms. Pasquarelli says the outage put vulnerabilities and dependence in plain sight for governments, businesses and critical infrastructure. The report does not describe any regulator's response or legal action. If regulators were to act, it would likely follow Microsoft's published findings, but that is our reading and nothing in the report says it will happen.
The strongest reason this view could be wrong
The whole piece rests on one report and on Microsoft's own account of the cause. If the review shows a different root cause, such as a software change or a security event, the analysis of mechanism and blame would change. The scale of effect could also prove smaller than the headline suggests if many named bodies had only brief or partial disruption.
What would settle it
Microsoft's written review of the incident would settle the cause, the sequence of events and the planned fixes. Statements from the affected bodies, such as the court system in Singapore or the US House, would show how long each was down. Microsoft's service status history for the event would give the timeline. We have only the BBC's account of these, so those primary documents should be read next.
What can be done
For individuals and small firms: keep a second way to reach key contacts, such as a different email account or a phone list, and know where critical documents are stored outside one provider. Treat unexpected messages during an outage with care.
For larger organisations: check which processes stop when one provider fails, including any AI assistant tied to that provider, and test a fallback. Our Day Zero plan and the live risk monitor give starting points.
For Microsoft: publish the review findings. Whether regional splitting is workable for its services is an open question that the review could address.
Questions readers ask
How long did the Microsoft 365 outage last?
The BBC reports it lasted more than 22 hours, with recovery taking nearly a full day as engineers gradually restored services. Microsoft had given no ETA for full recovery in a statement late on Friday (GMT).
What caused the Microsoft outage?
Microsoft says it was a technical failure and not a cyber-attack. The BBC reports that routine automated checks took some servers offline, which overloaded the remaining ones with traffic.
Who was affected?
The BBC says around 430 million people use Microsoft 365 and issues were reported in more than 70 countries. It names the US House of Representatives, Australian intelligence agencies, Singapore courts and Indian airline support desks.
Is there anything we should do now?
The sources give no instruction for users. As general good practice, keep an alternative way to contact customers and colleagues in case one provider fails.
Written by Toswo Research. This is not investment, legal or medical advice. To report an error, see corrections.
Toswo Signals, "Microsoft 365 outage lasts over 22 hours worldwide", https://www.toswo.com/signals/microsoft-365-outage-lasts-over-22-hours-worldwide, published 2026-11-15, retrieved 2026-11-22.