Microsoft 365 Goes Dark Worldwide as Outlook, Teams and Copilot Fail in Ten-Hour Global Outage

Microsoft’s email and workplace software stopped working for millions of people around the world on Thursday, in a global outage that knocked out Outlook, Teams and the company’s Copilot assistant for about ten hours and was still not fully fixed as Friday began. The failure, which Microsoft traced to its own servers dropping out during routine health checks, hit businesses, government agencies and emergency services across 74 countries and territories, and it exposed once again how much of the working world runs on a single company’s cloud.

The trouble began on Thursday afternoon in the United States. According to USA Today, reports of problems with Outlook on the outage tracker Downdetector surged to a peak of 128,874 at 1:09 p.m. Eastern Time, while reports for Microsoft 365 as a whole climbed above 47,000. Reuters reported that at the peak, Downdetector showed more than 130,000 reports of Outlook problems in the United States, and that the disruption spread across the Americas, Europe, Asia and Africa within about an hour. By Friday morning, Microsoft was reporting partial recovery in some regions, according to The National, but services remained disrupted as engineers worked to restore normal operations.

The outage centred on Exchange Online, the cloud service behind Outlook email, calendars, contacts and tasks. In a service health notice quoted by USA Today, Microsoft said the features of Microsoft 365 that depended on Exchange Online and email were affected, while elements that did not rely on Exchange Online remained functional. Users reported being unable to open mailboxes, send email, join some meetings or reach files. The company’s own status page listed Outlook, Teams, SharePoint, Copilot, Defender, To Do and Purview among the disrupted services, and administrators tracked the incident under the reference numbers EX1550543 for Exchange Online and MO1550821 for Microsoft 365.

It is important to be clear about what the numbers mean. Downdetector counts reports submitted by users, not the number of people affected, so its figures are a measure of the scale of complaints rather than a headcount of those locked out. But the pattern was unambiguous: a sharp spike at the start of the American working afternoon, a slow decline through the evening, and thousands of reports still coming in late at night. More than 18,000 users in the United States were still reporting Outlook problems at around 11:30 p.m. Eastern Time on Thursday, according to USA Today, with more than 5,000 reporting problems with Microsoft 365. Downdetector also showed outages for several Microsoft services in India, Australia and Japan early on Friday local time, as the working day began across Asia and the Pacific.

What Microsoft says went wrong

Microsoft’s own account of the failure changed several times through the day, and the company was unusually candid about how badly its repair efforts were going. In its first updates, the company said it was investigating reports of users experiencing issues with Microsoft 365 services, primarily with Exchange Online, and that the problem might affect connectivity for users in any region. It then said a portion of its infrastructure was experiencing elevated load across multiple regions, which was likely contributing to the connection errors.

The more specific explanation came later. Microsoft said it had determined that the elevated load was not driven by an increase in traffic, but by a number of servers becoming unexpectedly unavailable, which strained the infrastructure that remained. Affected servers, the company said, had failed during automated health checks and were therefore treated as unavailable. In an update at 5:09 p.m. Eastern Time, Microsoft said it had found that a core process on affected servers was becoming overwhelmed, causing capacity loss. “We’re opening all available pathways to restore service, as our actions taken thus far have not provided the anticipated relief,” the company said in that update, according to USA Today.

The company tried several remedies. It said it was redistributing connection traffic across its broader worldwide infrastructure, pausing all deployments and configuration changes, restarting affected servers and returning them to service in stages, and limiting which types of traffic could reach the affected infrastructure in some locations. At about 4 p.m. Eastern Time, Microsoft said it had validated an alternative method for carrying out the health checks, which allowed a subset of affected servers to return to service, and that it was applying the change more broadly and preparing automated recovery systems to detect and restart unresponsive servers continually.

But the recovery kept slipping. Shortly before 9 p.m. Eastern Time, Microsoft said it was seeing positive results from restarting affected infrastructure in isolation. About an hour later, it said the opposite: “We’ve monitored the status of the changes applied to improve serviceability and confirmed they’re not taking as expected,” the company said, according to USA Today. “We’re identifying alternate workstreams that can be applied to increase capacity in the face of sustained traffic management imbalances.” At around 11:30 p.m. Eastern Time, the company said it was working to improve server stability to improve traffic flow, and added: “We’re observing partial recovery across regions in which we’ve applied these adjustments; however, these actions aren’t resulting in sustained recovery, so we’re reviewing what additional actions we can take to improve overall infrastructure stability.”

Microsoft has not confirmed a final root cause, and it has given no time for full recovery. The National reported that engineers were investigating whether a recent network configuration change had played a part. A Microsoft spokesperson told the ABC, Australia’s public broadcaster, that there was no indication of a cyber attack, and a Microsoft executive, Charles Lamanna, told the BBC that the outage was not caused by a cyber-security incident, ruling out malicious or hacking activity.

Who was hit, and how badly

The outage reached far beyond office workers unable to open their inboxes. In Australia, the country’s spy agency and federal police were among the organisations caught up in the failure, Reuters reported. At a news conference on Friday, Australian Security Intelligence Organisation Director-General Mike Burgess and Australian Federal Police Commissioner Krissy Barrett confirmed their organisations had been affected. Home Affairs Minister Tony Burke said Microsoft had told him it did not believe the outage was caused by a cyber attack, and that the government had activated a formal agreement with the company to ensure Australia was prioritised in the restoration of services. “There is good security work that is done with Microsoft as a business and when it came to the formal cyber agreements which I’ve signed on behalf of Australia, the first of those was with Microsoft,” Mr Burke said at the news conference, according to Reuters. “That agreement has in fact been activated today to make sure that we get appropriate prioritisation of Australian interests in this.”

In the United States, the House of Representatives said the outage had affected its users and that its engineers were working with Microsoft, according to Reuters. In Southern California, NBC 4 Los Angeles reported that the Los Angeles County Sheriff’s Department and other government agencies in the area were experiencing issues with their communications services. A message on the Los Angeles County Medical Examiner’s website stated that its email was down “due to a nationwide Microsoft Outlook outage,” USA Today reported. In Arapahoe County, Colorado, the sheriff’s office said the outage was affecting online bond payments for people held at the county detention facility. “Because of the outage, jail officials are not receiving the email notifications needed to process online bond payments,” the sheriff’s office said in a statement on social media, according to USA Today. “Until Outlook email is restored, anyone needing to post bond for immediate release must do so in person at the jail.”

Businesses felt it too. Electric truck maker Workhorse Group, which had been expected to report third-quarter results on November 12, rescheduled the announcement to November 16, citing business disruptions caused by the global outage, Reuters reported. The failure also rippled into the artificial intelligence industry: OpenAI said it was investigating an issue with Microsoft connectors, noting that new connections might be broken and tools were not working as expected, while Anthropic said it had identified an issue with an upstream provider affecting its Microsoft 365 connector, according to USA Today.

The scale of the disruption reflects how deeply Microsoft 365 is embedded in working life. More than 430 million people use Microsoft 365 apps, according to Microsoft’s own website, as reported by the BBC, and tracking platforms put the number of companies using the service at between three and four million. Outlook is the default email system for a large share of the world’s companies and governments, and Teams is the meeting room for millions of remote and hybrid workers. When both fail at once, the effect is not a single app going dark but a working day losing its two most basic tools: the inbox and the meeting.

A familiar pattern, and an uncomfortable one

This is not the first time Microsoft’s cloud has failed on this scale, and the timing is awkward for the company. The latest disruption follows an outage last month involving Microsoft’s Azure cloud platform, which affected the company’s suite of productivity software and disrupted services across a range of industries worldwide, according to USA Today. That incident, in October 2026, was itself a reminder that the infrastructure behind Microsoft’s most widely used products can fail in ways that cascade across the whole suite.

The comparison many people reached for on Thursday was the global IT outage of July 19, 2024, when a faulty software update from the cyber-security firm CrowdStrike crashed Windows computers around the world. Microsoft estimated at the time that the CrowdStrike update had affected about 8.5 million Windows devices, or less than one percent of all Windows machines, yet the impact was enormous: thousands of flights were grounded, emergency services went offline in several American states, banks were disrupted and hospitals cancelled surgeries. Australia’s Home Affairs Minister, Tony Burke, drew the parallel himself on Friday, telling reporters that Microsoft believed it was dealing with something similar to what was seen with CrowdStrike, “where there is an update that has gone awry,” according to Reuters. The comparison should be treated carefully: the CrowdStrike failure was caused by a defective update from a third-party security company, while Microsoft has not said this week’s outage was caused by an update, and has said only that it is investigating whether a recent network configuration change played a part. But the parallel captures the same underlying lesson, which is that a fault in one widely used piece of software can paralyse organisations that had no part in creating it.

There is also a longer history. Microsoft’s cloud services have suffered several major outages over the years, and each one has prompted the same questions about concentration risk: whether it is wise for so much of the world’s email, documents, meetings and now AI assistance to depend on a single provider’s infrastructure, and whether the companies and governments that depend on it have workable fallbacks when it fails. Those questions were asked after the CrowdStrike outage in 2024, after the Azure outage last month, and they will be asked again now.

Analysis: Why It Matters

1. The health check that killed the patient

The most striking detail in Microsoft’s account is also the most ironic. The servers did not fail because they were overwhelmed by users, or because they were attacked, or because a data centre lost power. According to Microsoft, they became unavailable because they failed automated health checks, and the system then treated them as dead. The checks are supposed to protect the service by removing unhealthy machines from rotation. In this case, the protective mechanism appears to have taken healthy capacity offline faster than it could be replaced, and the remaining servers then buckled under the redistributed load. It is a classic cascade failure, and it is the kind of failure that is hardest to defend against, because the system did exactly what it was designed to do. The lesson for Microsoft’s engineers is not that health checks are a bad idea, but that a health-check system needs its own safeguards: limits on how much capacity it can remove at once, and a way to distinguish between a server that is genuinely broken and one that is merely struggling under a load spike. The fact that Microsoft’s fix was, in effect, a different way of doing the health check suggests the company has reached the same conclusion.

2. Ten hours is a long time for the world’s inbox

Outages happen to every cloud provider, and a few hours of downtime is, by itself, not a scandal. What makes this one notable is its duration and its stubbornness. Ten hours after the first reports, Microsoft was still telling users that its fixes were “not taking as expected” and that it was looking for “alternate workstreams.” Restored servers, the company warned, could become unresponsive again once they received traffic. That is an unusually frank admission of a recovery that is not working, and it matters because of what was down. Email is not a luxury service. For a hospital, a police department, a jail or a legislature, email is the channel through which decisions, records and releases move. The Arapahoe County case is a small but telling example: a person who could have been released on bond had to wait, or travel in person, because an email notification never arrived. Multiplied across thousands of organisations, that kind of friction is the real cost of the outage, and it is a cost that does not show up in Downdetector charts.

3. The cloud is a single point of failure wearing a distributed costume

Microsoft’s cloud is, by any measure, one of the most distributed systems on earth, with data centres in dozens of regions and traffic routed across continents. Yet on Thursday, a fault in one part of that system produced failures in 74 countries and territories within an hour. That is the paradox of modern cloud computing: the infrastructure is geographically distributed, but the software that runs it is logically centralised. Exchange Online is one service, run by one company, on one set of shared systems, and when its core process fails, there is no rival Exchange Online to fall back on. The same logic applies to the AI connectors that failed alongside it: OpenAI and Anthropic both depend on Microsoft’s platform to reach their users, so a Microsoft outage became, in part, an AI outage too. For the organisations that depend on these services, the practical question is not whether Microsoft is reliable in general, but whether they have any plan at all for the hours when it is not. Most, evidently, do not.

4. “Not a cyber attack” is good news, and also the bad news

Microsoft and the Australian government were quick to rule out a cyber attack, and that is genuinely reassuring: nobody’s data was stolen, no ransomware gang held the world’s email hostage, and there is no adversary to hunt. But the reassurance cuts both ways. A cyber attack, however damaging, is an external event that a company can harden itself against. An outage caused by the company’s own systems failing their own checks is an internal event, and it is harder to explain away. It suggests a flaw in the design or operation of the infrastructure itself, not a hole that an attacker exploited. That is why the investigation into a possible network configuration change matters so much: if a routine change contributed to the failure, then the question becomes how a change with that much destructive potential reached production without a safeguard catching it. Microsoft’s decision to pause all deployments and configuration changes while it investigates is a sign that the company is asking itself the same question.

5. Trust is the product now, and it is dented twice in two months

Microsoft sells reliability as much as it sells software. Its customers pay for Microsoft 365 not only because Outlook and Teams are useful, but because they expect them to work, always, everywhere. Two major outages in about a month, the Azure failure in October and this one in November, damage that expectation in a way that a single incident would not. Each outage on its own can be explained; two in quick succession start to look like a pattern, and customers begin to ask whether the pace of change inside Microsoft’s cloud has outrun its ability to keep the service stable. The company will publish a formal incident report in due course, and it will no doubt describe the technical cause in detail. But the commercial and reputational question is simpler: how many more times can the world’s inbox go dark before the organisations that depend on it start building real alternatives, or demanding contractual guarantees with real teeth? The answer will shape Microsoft’s cloud business for years.

What to Watch Next

1. Microsoft’s formal incident report: the company is expected to publish a detailed account of the root cause, including whether a network configuration change contributed. The report’s honesty about the health-check failure will be closely read by customers and rivals alike.
2. Full restoration: as of Friday morning, recovery was partial and uneven across regions. Watch the Microsoft 365 Status account and the status page at status.cloud.microsoft for confirmation that all services, including Copilot and Defender, are fully back.
3. Service credits and compensation: large enterprise customers with service-level agreements may be entitled to credits for the downtime. Watch for announcements from Microsoft and for any public complaints from major customers.
4. Regulatory attention: the outage affected a spy agency, a federal police force, a legislature and emergency services. Governments in Australia, the United States and elsewhere may ask Microsoft for formal explanations, as they did after the CrowdStrike outage in 2024.
5. The Azure connection: Microsoft said this outage was centred on Microsoft 365 and Exchange Online, not Azure, but the October Azure outage also hit the productivity suite. Watch whether the company explains how the two incidents relate, if at all.
6. OpenAI and Anthropic connectors: both companies reported problems with their Microsoft 365 connectors during the outage. Watch for confirmation that those integrations are fully restored, and for any changes to how they are built.
7. Customer fallout: the most important signal will be what Microsoft’s biggest customers do next. Any public move by a large company or government to diversify away from a single cloud provider, or to demand stronger guarantees, would mark a lasting consequence of this week.

Sources

• USA Today — Is Microsoft down? Updates as users report issues with Outlook, 365
• Reuters via BusinessWorld — Microsoft outage drags on for hours, disrupting Outlook, Teams for thousands
• Reuters — Australia’s spy agency, federal police hit by Microsoft outage
• The National — Microsoft identifies cause of global outage as Outlook, Teams recovery falters
• Dubai Standard — Microsoft outage affects Outlook and Teams users worldwide, disrupting businesses
• Cybersecurity Dive — CrowdStrike, Microsoft scramble to contain fallout from global IT outage