Microsoft’s global Outlook outage has dragged into a second day, leaving users unable to access email and disrupting Teams, Copilot, and other Microsoft 365 tools.
It has also affected AI tools. Anthropic and OpenAI said in status update pages that their connectors to Microsoft 365 tools are disrupted. Anthropic also said that authentication issues were affecting Claude Code, but did not attribute this to the Microsoft outage.
Here’s what we know so far about the Microsoft outage that’s causing chaos for businesses.
What’s the latest?
Microsoft said in a 07:09 a.m. ET update on Friday that more of its services were coming back online and that email delivery was improving.
But the company warned that some users could still run into problems while it works to restore the rest of its systems.
Microsoft said it was now focused on clearing the backlog of delayed emails. It is keeping messages in the system for up to 48 hours, giving them more time to be delivered rather than immediately returning an error.
The company has not provided a timeline for a full recovery.
What happened?
Microsoft first flagged the outage on Thursday around midday ET, writing in an X post that users in the Americas were having trouble accessing Exchange Online mailboxes. It later said in a follow-up post that the issue could affect users “in any region.”
By Friday morning, Microsoft’s public service status page was still showing “service degradation” for both Microsoft 365 business customers and consumer products.
Microsoft said in a later X update on Thursday that some of its infrastructure was under elevated load across multiple regions. It said the strain was not caused by an increase in customer traffic, but because a number of servers had unexpectedly become unavailable.
Microsoft also said in that 3:34 a.m. update that it had “no indication” the disruption was caused by malicious activity.
In further X updates on Thursday, it said it had taken steps including restarting servers. Early Friday, Microsoft said that servers initially returned to service could become unresponsive again once traffic hit them.
What’s the impact?
The outage has caused significant disruption for businesses, with many workers unable to use systems such as email or Teams.
Downdetector recorded more than 130,000 reports of Outlook problems in the US at the outage’s peak on Thursday. That figure reflects user reports, not a confirmed count of affected people or businesses.
Other Microsoft services are affected too. It said in its service advisory that the disruption could affect Outlook, Teams, SharePoint, Copilot, Microsoft Defender, Intune, Power BI, and other Microsoft 365 products that rely on Exchange Online.
Reuters reported on Thursday that the US House of Representatives said the outage had affected its users, while electric-truck maker Workhorse Group postponed its quarterly earnings call because of the disruption.
Air India said in a Thursday X post that it had rerouted email services, while IndiGo said in an X post on Friday that its email support channel was disrupted. Both airlines said their flight operations were unaffected. AP also reported that government agencies in Singapore, the Philippines, and Japan had reported knock-on effects.
Is this unusual?
Microsoft — and other Big Tech companies — have been here before.
In November 2024, Outlook, Teams, SharePoint, and other Microsoft 365 services went down after a system change triggered a flood of automated reconnection attempts, putting pressure on the company’s servers.
Microsoft also suffered a separate outage for Azure, its cloud service, on September 30 and October 1, affecting customers across several regions. Microsoft said it rolled back a change after some customers had trouble using its cloud networking services
Other cloud providers have faced similar problems. Google said a global Google Cloud and Workspace outage in June 2025 lasted three hours.
Amazon Web Services, Amazon’s cloud computing arm, also had a major disruption in Northern Virginia in October 2025. It said it fixed the initial problem in about two and a half hours, though some services took longer to return to normal.
