By Microsoft’s global outage disrupted Outlook, Teams and Copilot into Friday, affecting businesses and government agencies while interrupting Microsoft-connected features in ChatGPT and Anthropic’s Claude.
The disruption began on Thursday and lasted more than 12 hours, with Downdetector recording more than 130,000 US Outlook problem reports at its peak.
Users reported missing emails, failed logins and difficulty accessing workplace services.
Microsoft traced the disruption to unavailable servers and overloaded processes handling connections, although its final explanation remains pending.
The problems exposed a wider vulnerability, as when a major software provider fails, the consequences can reach AI assistants, financial transactions and corporate communications far beyond its own applications.
How Microsoft’s server problem reached AI assistants
The incident began around 16:17 UTC on Thursday within Exchange Online, the cloud service underpinning Outlook email.
Microsoft found that some servers had stopped responding, transferring more work to those still operating.
Attempts to restore capacity were complicated when restarted servers became unresponsive again under renewed traffic.
Engineers redirected connections, restarted equipment and tested configuration changes while investigating the underlying cause.
“And so you end up with a cascading problem,” RMIT University engineering academic Mark Gregory told The Guardian, explaining how overloaded surviving servers can also fail.
That disruption reached AI companies through integrations with Microsoft services. OpenAI warned that new Microsoft connector connections might fail and associated tools might not work correctly.
Anthropic reported failures affecting its Microsoft 365 connector’s email, calendar and search functions, alongside Claude for Outlook.
The problem was access to linked workplace information, rather than the underlying chat systems becoming universally unavailable.
Microsoft said it had found no indication of malicious activity.
Why Microsoft outage became a business problem
The disruption extended beyond office inboxes. Exchange-dependent functions in Teams and Copilot were affected, including calendar access and the ability of AI tools to retrieve work information.
Government organisations in Australia, Japan and the United States reported difficulties. Indian airlines IndiGo and Akasa Air said customer email support was disrupted.
Reuters reported problems at Taiwanese financial institutions, including an unnamed company that cancelled a planned bond repurchase transaction.
US electric-truck manufacturer Workhorse postponed its quarterly earnings call until 16 November, explicitly citing Microsoft’s outage.
These examples show how a communication failure can delay financial decisions and investor announcements, even when the businesses involved are otherwise operational.
“Enterprises have priced Microsoft 365 as a licence and never as a single point of failure,” Greyhound Research chief analyst Sanchit Vir Gogia told CIO.
He argued that companies need practical contingency arrangements rather than assuming one provider’s services will remain continuously available.
Could repeated outages weaken Microsoft’s AI ambitions?
Microsoft stock closed Thursday at $523, up 0.36%, suggesting investors had not immediately treated the disruption as a major earnings threat.
That reaction does not settle the longer-term question. Microsoft wants companies to rely more heavily on Copilot and interconnected cloud services, making reliability increasingly important to its commercial strategy.
“I think investors will take this in their stride for now,” eToro analyst Josh Gilbert told Reuters, noting that enterprise customers cannot easily abandon deeply embedded Microsoft systems.
But repeated incidents could encourage technology departments to reconsider concentrating communications, security and AI tools with one supplier.
For customers, backup communication channels and rehearsed contingency plans may matter as much as software features, subscription prices and service guarantees.
Microsoft reported some recovery as servers were restarted and traffic rerouted, without confirming complete restoration in the latest updates.
Its next challenge is demonstrating that the infrastructure connecting everyday business software and newer AI products can withstand another failure without spreading disruption across customers’ operations.
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 Microsoft outage spreads to OpenAI and Anthropic: how did it get this bad? appeared first on Invezz