Backup & Recovery

Microsoft outage: Outlook down in 74 countries, now restored

Key takeaways

  • Outlook was hit hardest while Teams chats and calls largely kept working, so email is the service to plan around.
  • Restored servers kept failing again as traffic from other servers swamped them, which analysts say may explain why Microsoft's fixes did not hold.
  • Microsoft says there is no sign of a cyberattack, but has not published a root cause, so IT teams should test a minimum viable service plan now.

Microsoft said it had only "marginal success" restoring Outlook, Teams and Copilot more than 12 hours into a global outage affecting users in 74 countries, before declaring services fully restored on Saturday. Outlook, the email platform at the heart of Microsoft's workplace services, took the worst of it. For IT teams, this is less a product review than a stress test of what happens when a single supplier carries a business's mail.

But the outage gives a rare, live look at how its workplace services behave under failure, and the picture is instructive.

What held up

The first thing to note is what kept working. Teams chats and calls have largely continued, and the desktop version of Outlook may still show emails already downloaded. The problem lies with Microsoft's servers rather than users' own devices, so there is little that users can do themselves.

Analyst Rupert Goodwins said Microsoft has been quite open that its fixes are not really working yet. The company posted a series of updates on X about its progress, and in a statement on its service health page early on Friday it said it was monitoring activity across regions to assess server status and find ways to improve capacity. It said this included reviewing server performance and confirming that developed patches had been applied and were working as intended after infrastructure reboots were completed.

Charles Lamanna, Microsoft's executive vice president of Copilot, Agents and Platform, said late on Thursday there was "no indication this is the result of malicious or cyber activity". Goodwins said he believed Microsoft's position that an external attack had not happened, because there are ways to mitigate that kind of attack.

Microsoft also told Australia's Home Affairs Minister Tony Burke that it did not believe it was dealing with a cyberattack. Burke said the company believed it was dealing with something similar to the CrowdStrike incident, where an update had gone awry.

Where it fell short

The failures were widespread. Problems began at about 16:00 GMT on Thursday, when users reported they could not access Outlook. Within about an hour, Downdetector, the outage tracker, had logged reports from 74 countries and territories. By about 07:00 GMT on Friday it had received at least 43,000 reports of problems with Outlook from users in 72 countries. The true number is likely far higher, since the figure only counts people who visit the site to report an issue.

Users reported being unable to open their mailboxes via the web, the app or mobile devices, and many emails that did go through arrived late. The disruption spread beyond Microsoft's own products, hitting integrations which other companies, including the AI firms OpenAI and Anthropic, rely on.

The recovery itself has been the weak point. Microsoft first said it was investigating problems in the Americas, but later confirmed that users in any region could be affected. It said the affected infrastructure was experiencing elevated load across multiple regions, then said a number of servers had become unexpectedly unavailable and that the elevated load had not been caused by an increase in traffic. Late on Thursday it reported only limited recovery as it tried to route users' traffic away from affected servers, and later said the changes it had applied were not taking effect as expected.

Walter Pasquarelli, an affiliate researcher at the Bennett School of Public Policy at the University of Cambridge and an AI specialist, described the cascade: some servers failed automated health checks, traffic then went to other servers, and those became overloaded. His analogy was a highway with four lanes that suddenly has two, so that the entire thing shuts down.

Goodwins described the restart problem. Servers knocked offline usually attempt to reconnect so they can synchronise their files again. As things come back up, they are "immediately swamped by the traffic from all the other servers desperate to talk to them. So it all falls down again." Servers Microsoft had restored were failing again by Friday morning in Europe.

Goodwins also offered a pointed view on staffing and priorities: that Microsoft's focus on AI may have played an indirect role, with email and Teams seen as forgotten revenue sources that did not get the investment and management attention they should have had. He also said many experienced engineers have left Microsoft, and that there has been a draining of corporate memory of how these systems are built up and how they work. These are his opinions, not Microsoft's findings, and the company has not published a root cause.

Pasquarelli said the evidence so far points to a technical failure, and that he saw no sign of an extremely high amount of computational power or traffic caused by AI systems. The cause, he said, could be damaged infrastructure, a mistake in automated systems or simple incompetence.

The wider damage was real. Government systems in Singapore and Japan were disrupted, as were communications at the Supreme Court of the Philippines. In Australia, the country's spy agency and federal police were among the organisations hit.

How it compares

This is not Microsoft's first major outage this year. In January, Microsoft 365 went down across North America for almost 10 hours, which the company blamed on heavy service load combined with reduced capacity during maintenance. In late August, thousands of Outlook users worldwide were locked out of their mailboxes.

Other tech companies have suffered similar failures. In October 2025, an Amazon Web Services outage took down a large number of websites and apps. In July 2024, a faulty software update from the cybersecurity firm CrowdStrike crashed computers around the world, cancelling flights, disrupting hospitals and causing billions of dollars in losses.

Cost and duration

Microsoft gave no timeline for full recovery during the outage, though it said that as capacity returned, users may see service come back gradually as their apps reconnect, rather than all at once. The outage began at about 16:00 GMT on Thursday and was still unresolved on Friday morning in Europe, with Microsoft declaring services fully restored on Saturday.

Who should use it

About 430 million people use Microsoft services, according to Pasquarelli. The outage does not change that calculus overnight.

What it should change is how organisations plan around it. Pasquarelli's advice is direct: ask "what is the minimum viable service or the minimum viable product that we need to keep up if there's a global outage?" Teams that have not written this down, or tested it, have just had it tested for them. Organisations building a continuity plan can start with how recovery objectives are set; our explainer on RPO, RTO and MTD covers the terms, and the ransomware recovery and incident response plan guide offers a template for communications when core systems are unavailable.

Pasquarelli also framed the incident as first and foremost a communication risk. He noted that people take these tools for granted until they are suddenly down, and Goodwins said the short-term impact of such outages tends to be limited: you cannot log on to your email for a bit, it comes back up, and people grumble and get back on with life.

There is also a policy dimension. The outage comes as Europe pushes for greater technological sovereignty, aiming to run its own servers and depend less on foreign providers. Lacking tech giants of its own, the continent relies heavily on US firms for digital services.

Verdict

Keep using Microsoft's workplace services, but stop treating them as infallible. The practical step for IT teams is a documented, tested fallback for email and urgent communications. Microsoft has since said the incident arose from an issue during a network capacity upgrade after a recent network change, so customers can weigh how avoidable it was.

Frequently asked questions

Was the Microsoft outage a cyberattack?

Microsoft says there is no indication of malicious or cyber activity. On Friday it was rolling back a network change, though it did not say whether that change caused the problems. No root cause has been published.

What can users do while Outlook is down?

Little, because the fault is on Microsoft's servers. Desktop Outlook may still show emails already downloaded, and Teams chats and calls have largely continued working.

More from Future Technology HQ