Microsoft’s Outlook email service was still not working properly for many users on Friday morning, almost 19 hours after the outage began. Microsoft says its fixes are working only gradually and has given no time for full recovery.
The problems started at 16:17 GMT on Thursday, according to Microsoft’s incident record for Exchange Online, the system behind business Outlook mailboxes. They spread from the Americas to Asia overnight and hit Europe as offices opened.
At 11:11 GMT on Friday, the Microsoft 365 Status account said recovery work had been completed on “portions” of the affected infrastructure. Mailbox access and delayed email were improving, it said, but some users could still see intermittent problems.
Microsoft has said it is not aware of any malicious activity behind the incident.
Which Microsoft services are down?
The outage is centred on Exchange Online. Microsoft said users might be unable to reach their mailboxes through Outlook on the web, desktop Outlook and mobile mail apps.
Any Microsoft 365 feature that relies on Exchange has also been hit. That includes calendar scheduling in Teams, parts of SharePoint and Copilot, and several security tools, according to Microsoft’s advisory MO1550821. Features that do not depend on Exchange, such as joining Teams calls, have kept working.
Microsoft disclosed in April 2025 that it had more than 430 million paid Microsoft 365 commercial seats, which gives a sense of how many workers depend on the system.
Downdetector, which counts user reports, logged reports from 74 countries and territories within about an hour, NBC News reported. At the peak, there were 19,650 reports a minute across five Microsoft services, compared with fewer than 15 on a normal Thursday.
What Microsoft says went wrong
Microsoft has not published a root cause, promising a post-incident report once services are back. Its status updates, however, set out the problem in stages.
At 18:35 GMT on Thursday, it said part of its infrastructure was under “elevated load” in several regions. Less than an hour later, it clarified that the load was not caused by more traffic. Instead, a number of servers had become “unexpectedly unavailable”, straining the ones left running.
Those servers were failing automated health checks and being pulled from service. Microsoft paused all deployments and began restarting machines.
By early Friday, Microsoft’s admin centre updates said a process that sets up secure connections on the affected servers was becoming overwhelmed. Microsoft also said it had started reverting a recent network configuration change that “correlates with the start of impact”.
That is not a confirmed cause: Microsoft had earlier said it had found no change linked to the start of the problem.
Why Microsoft has been unable to bring it back
The main difficulty, in Microsoft’s own account, is that recovered servers did not stay recovered.
At 22:09 GMT on Thursday, Microsoft said its actions had “not provided the anticipated relief”. Two hours later, it reported that servers it had returned to service could become unresponsive again as soon as they received traffic.
At 03:13 GMT on Friday, it said the changes applied had not taken effect as expected. A later update described a surge of calls from its routing systems to back-end servers after traffic was rebalanced, and machines stuck in a “bad state” that blocked a network change from syncing.
On Microsoft’s account, each server that drops out pushes more work onto the rest. That is why engineers have been bringing capacity back slowly, limiting traffic in some places and draining, restarting and patching servers in stages.
The clearest progress came at about 07:24 GMT, when Microsoft said a patch to an “upstream authentication/routing configuration component” was improving server stability. It began extending the patch to regions entering business hours.
At 09:24 GMT, Microsoft reported “early stabilisation”. It also extended its email timeout to 48 hours, so that messages stuck in queues are held rather than bounced back.
Outside experts have been struck by how long it has lasted. Gergely Orosz, a former engineer at Skype and Uber, told the BBC he “cannot recall” a business email outage taking this long to return to normal.
Markets, traders and public services hit
The disruption has spilled into financial markets. Bloomberg, in a report syndicated by The Straits Times, reported that traders were doing business by phone, WhatsApp and even fax, and that trade confirmations were backing up. In Singapore, some banks were unable to complete financing deals.
Euronext told listed companies to phone through corporate information because its listing teams could not receive emails, Reuters reported. Workhorse Group postponed its earnings release, and Richemont briefly delayed its half-year results, NBC News reported.
In the UK, NHS England said staff were turning to paper records where needed, The Guardian reported. Japan’s government reported disruption to communications between ministries.
Microsoft shares were at $512.08 in pre-market trading at 11:19 GMT on Friday, according to Yahoo Finance data. That was 2.1% below Thursday’s close of $523.00.
The stock had touched $529.66 on Thursday before the outage widened, and closed up 0.4%.
Amish Patel, head of equity research at Raymond James, told The Guardian the direct financial impact was “likely to be negligible”. The bigger question, he said, is whether repeated outages make businesses rethink their reliance on one provider.
How it compares with other cloud outages
The outage is already longer than the two big cloud failures of October 2025.
Amazon Web Services’ outage in its US-EAST-1 region ran for about 14.5 hours on 19-20 October 2025, according to AWS’s own summary. Microsoft’s Azure Front Door outage on 29-30 October 2025 lasted about eight and a half hours, from 15:41 GMT to 00:05 GMT, according to Microsoft’s incident review.
The CrowdStrike incident of 19 July 2024 was different in kind. It was caused by a faulty security update that crashed about 8.5 million Windows computers, according to Microsoft, rather than by a failure in Microsoft’s own cloud.
What happens next
Microsoft’s focus is now on extending the fixes from Asia-Pacific to other regions and working through the backlog of delayed mail.
Its root-cause report, due after recovery, should show whether the network change was to blame and how Microsoft plans to prevent a repeat.
Until then, users may continue to see intermittent errors and late emails as service returns region by region.
AI disclosure: AI tools assisted with research and drafting of this piece. An Invezz editor reviewed the facts, sources and wording before publication. Any opinions or analysis are Invezz’s, not those of an automated system.
The post Why is Outlook down, and why can't Microsoft bring it back? What we know appeared first on Invezz