
Ethan Caldwell
Crypto Analyst
TL;DR
Microsoft says every customer and region is back online after a Microsoft 365 outage that began Thursday and lasted more than 24 hours. The company now blames a network capacity upgrade, not a cyberattack or AI agents. Outlook, Teams, government email and AI connectors were hit, and a detailed incident report is still to come.
Key takeaways
- Microsoft reported all remaining customers and regions restored by Saturday, citing incident MO1550821.
- Microsoft first blamed an overloaded component that manages secure connections, then said a network capacity upgrade and a recent network change caused the trouble.
- Microsoft called it a technical issue 'not related to cyber or AI/agents.'
- The Tenth Circuit clerk's office could not send or receive messages for most of Thursday, and Japan's Finance Ministry had no inbound email on Friday.
- OpenAI and Anthropic both reported failures in their Microsoft 365 connectors, which have since been marked as working.
More than 450 million paid commercial seats. That's the figure Microsoft gave in an October filing for its 365 suite, and it explains why a fault that began Thursday and ran past 24 hours became a story for finance ministries and courts, not just IT desks.
By Saturday, Microsoft reported that every remaining customer and region was back online, and it pointed administrators to incident MO1550821. The Microsoft 365 outage is over. The explanation took two tries.
Microsoft's Second Explanation Blames a Network Upgrade
The first account came Friday. As Cointelegraph reported, Microsoft initially pointed to a component that manages secure connections. It had been swamped, and the servers behind it quit responding to requests. Engineers restarted infrastructure, shifted traffic elsewhere and brought more server capacity online. By the 7:25 am UTC update, a patch had helped on some machines. A broader rollout was continuing, and no completion time was given.
The later account was more specific. Microsoft now says the trouble traced to a network capacity upgrade, with a problem surfacing after a recent network change. The company called it a technical issue and said it was "not related to cyber or AI/agents." Recovery happened in phases, and each phase was checked for stability before the next began, according to Cointelegraph's follow‑up.
Microsoft apologized. It also committed to publishing a detailed write‑up of the incident when its internal review is finished. I'll be reading that closely. An overloaded connection process and a botched capacity upgrade could be one chain of events seen from two angles. They could also be two different problems. The company hasn't said.
Email Outages Reach Courts and Ministries
Exchange Online is the hosted service that holds the mailboxes and calendars, so when it stumbles, everything leaning on it stumbles too. Outlook went down for some users. Teams users struggled to schedule meetings and send invites, though most calls and chats kept running. Copilot and Defender for Office 365 features were hit as well.
The public‑sector fallout was concrete. A federal appeals court, the Tenth Circuit, said its clerk's office was unable to send or receive messages through most of Thursday. Japan's Finance Ministry had no inbound email on Friday. Singapore's Government Technology Agency, speaking to CNA, listed government email among the systems the disruption touched.
Then there's the AI plumbing, which I didn't expect to matter this much. OpenAI reported errors in the Microsoft 365 connector inside ChatGPT (the link that lets the chatbot read your mail and calendar). Anthropic said the bulk of mail, calendar and search calls routed through its Claude connector were failing. Both have since marked their connectors as working. Companies that wired assistants into workplace accounts learned the assistant is only as available as the mailbox behind it.
Two Things to Watch Next
First, the post‑incident report: does it name the specific change, and does it explain why the fix took more than a day? Second, whether large customers, especially government ones, start asking for fallback email paths and contractual remedies.
Crypto readers have seen this movie. A single cloud provider's bad day froze trading at Coinbase not long ago. Concentration risk, meaning too many eggs in one vendor's basket, keeps turning up in places people assumed were diversified. An outage with no attacker behind it is, in some ways, the less comforting kind. There's nobody to blame and nothing obvious to harden.
