Microsoft Outlook Outage: What Happened to Microsoft 365, Teams and Copilot?
Last updated: November 14, 2026. Microsoft has said it will publish a fuller post-incident report, so details on the root cause may change.
On Thursday, November 12, 2026, Outlook, Teams, Copilot, and other Microsoft 365 services stopped working for users around the world, the Microsoft Outlook outage dragged on through Friday, and Microsoft did not say it had finished restoring service to the last affected customers and regions until Saturday morning, people could not reliably send email, open calendars, or join meetings, and the disruption reached governments, hospitals, airlines, and banks.
This post covers what happened, what Microsoft says went wrong, and what is worth taking away from it.
What happened
Complaints about Outlook began climbing at about 17:25 GMT on Thursday, which was midday on the US East Coast, according to Ookla, the company that owns Downdetector, Microsoft posted its first public alert shortly before 1 p.m. Eastern, within an hour, Downdetector was receiving outage reports from 74 countries and territories, so what began as a US-daytime problem turned global very quickly, reports spiked again as workdays started in Asia and Europe.
Microsoft’s early explanations were vague, the company first said part of its infrastructure was under elevated load across multiple regions later it said the load was not caused by a surge in traffic instead, a number of servers had become unexpectedly unavailable, which strained the ones that were still working, and some servers were failing automated health checks, at one point it said a core process on the affected servers was becoming overwhelmed, causing a loss of capacity.
Key facts and timeline
| When | What happened |
|---|---|
| Thursday, November 12, about 12:25 p.m. ET | Outlook outage reports start climbing on Downdetector |
| Thursday, shortly before 1 p.m. ET | Microsoft posts its first public alert about connectivity problems |
| Thursday afternoon and evening | Microsoft cites elevated load, unavailable servers, and failed health checks, and works on backend servers in stages |
| Friday, November 13 | Microsoft apologizes, says the problem came from an earlier network capacity upgrade, and says it was not a cyberattack or caused by AI agents |
| Friday, November 13 | Microsoft extends its email timeout window to 48 hours and works through queued messages |
| Saturday, November 14, morning | Microsoft says it has finished addressing the underlying issue and restored service to the remaining customers and regions |
Which services were affected
Outlook was the most visible casualty, but it was not alone, in the Microsoft 365 admin center, the incident was logged under the ID MO1550821 and listed Teams, Exchange Online, SharePoint Online, Microsoft Defender XDR, and Microsoft 365 Copilot as services that could see connection failures.
Exchange Online is the engine behind Microsoft’s cloud email and calendars, so a problem there shows up in Outlook on the web, on desktop, and on phones.
Some users also reported trouble with other products, such as OneDrive and Word, but those reports came from users and were not part of the official service list, so treat them with some caution, the disruption also touched consumer accounts as well as business ones.
About 430 million people use Microsoft 365 apps, according to Microsoft.
How big was it
Reuters reported that Downdetector logged more than 7.4 million reports worldwide, peaking at nearly 20,000 a minute, which was more than 1,500 times the normal level, it is worth being precise about what that number means, it counts reports, not people, and one frustrated user refreshing a page can file several.
Other coverage cited a different Downdetector measure, more than 130,000 Outlook-related reports at the peak, which reflects how the counting is done and not a disagreement about the outage itself.
Reuters described it as one of the longest outages in Microsoft’s history and the company’s second service disruption this year, reports said the outage affected governments, courts, healthcare services, businesses, airlines, and banks.
In the UK, the NHS and police activated contingency plans to keep operating, the US House of Representatives said its technology staff worked with Microsoft to restore email for members.
Two Indian airlines, IndiGo and Akasa Air, said their email support was affected, and several Taiwanese financial firms told Reuters that fund management services were disrupted.
What caused it
Microsoft’s explanation, given on Friday, is short, the incident was introduced by an issue during an earlier network capacity upgrade, and it surfaced after a more recent network change, the company said it was a technical issue not related to cyber activity or AI agents, and that it had no indication of malicious activity.
A change made to add capacity to the network appears to have left a fault behind, that fault did not cause trouble until a later network change exposed it, at which point servers began failing and the rest of the system could not carry the load, this is a pattern familiar to anyone who has run production systems, where the change that breaks things is not always the one that happens just before the failure.
This is still an incomplete account, Microsoft has not yet explained exactly what failed in the upgrade or why it spread across so many regions, and it says a post-incident report will follow, until then, details such as the exact component involved remain partly unknown.
How Microsoft recovered
Recovery was gradual, Microsoft repaired affected backend servers in stages and checked that each change took effect before moving on, it added capacity using virtual machines, rebalanced traffic across its infrastructure, and then spent hours draining the queue of email that had piled up. On Friday it said that people might see sign-in and access improve before message delivery caught up, and it did not offer an estimate for full recovery.
It also extended the email timeout window to 48 hours, which was meant to give queued mail more time to get through.
On Saturday morning, Microsoft posted that it had finished its work on the underlying problem and restored service to the remaining affected customers and regions.
If you manage a Microsoft 365 tenant, check the Service health page in the admin center for the incident record, and test your key workflows such as sending and receiving mail, scheduling a Teams meeting, and opening SharePoint files.
If you sent something important on Thursday or Friday and are not sure it arrived, follow up with the recipient directly, since delayed delivery was one of the main side effects.
Organizations of any size gain from having a plan for the day their main email or chat tool disappears, that means a second way to reach staff, such as a phone tree or a separate messaging app, a way to reach key documents that does not depend on one provider, and a decision in advance about who communicates with customers while systems are down.
What developers can take from this
Many applications depend on Microsoft 365 behind the scenes, through the Microsoft Graph API for mail, calendars, and files, when a provider has a bad day, your code needs to fail gracefully.
A simple retry with exponential backoff, which waits longer between each attempt, avoids hammering a struggling service and lets your app recover on its own when the service returns.
async function fetchWithRetry(url, options = {}, maxAttempts = 5) {
let delay = 1000;
for (let attempt = 1; attempt <= maxAttempts; attempt++) {
try {
const response = await fetch(url, options);
// Retry on throttling and temporary server errors
if (response.status === 429 || response.status >= 500) {
const retryAfter = response.headers.get("Retry-After");
const wait = retryAfter ? Number(retryAfter) * 1000 : delay;
await new Promise((resolve) => setTimeout(resolve, wait));
delay *= 2;
continue;
}
return response;
} catch (error) {
if (attempt === maxAttempts) throw error;
await new Promise((resolve) => setTimeout(resolve, delay));
delay *= 2;
}
}
throw new Error("Request failed after retries");
}
The snippet respects a Retry-After header when the service sends one, which is how many APIs, including Microsoft Graph, signal how long to wait, pair it with a queue for anything you cannot afford to lose, such as outgoing email, and with clear status messages so your users know the delay is not your fault.
Another lesson sits in how the outage started, a capacity upgrade was done earlier, and a later change exposed its flaw, for anyone deploying software, that is a reminder to test infrastructure changes under realistic conditions, to roll out in stages where you can, and to keep a fast way to undo a change.
It is easy to treat a cloud provider as a utility that is always on, but this outage showed how many systems quietly depend on one company’s network, if your own project leans on a single provider for email, authentication, storage, or hosting, take an hour to map what would break if that provider went down for a day.
You will probably find at least one place where a fallback, a cache, or a clearer error message would help, the same habit applies on a small WordPress site as much as on a large platform, so it is worth starting with the dependencies you cannot easily replace.
Discover more from Developer Hint
Subscribe to get the latest posts sent to your email.
