Microsoft Explains 30-Hour Outlook and Teams Outage: It Was the Network, Not AI

⚡ The gist

• Microsoft has named the cause of the 30-hour outage that hit Outlook, Teams, and other services: a bug in a recent network bandwidth update.
• Companies and government agencies in at least 74 countries were affected; the fix rolled out in stages, with stability checks after each step.
• The company ruled out hackers and AI agents, and promised a full report after its internal investigation.

🔍 What happened

According to Microsoft's status page, the outage began Thursday at 11:05 and was resolved Friday at 17:10 — nearly 30 hours. Then the backlog of emails took all night to clear. Companies and government institutions in at least 74 countries were hit: in Japan, agencies fell back on paper, banks on fax machines, and luxury holding Richemont delayed publishing its half-year results. AI services like Claude and ChatGPT lost access to email, calendars, and files in Microsoft 365 through their connectors. iXBT Live recaps the details.

Microsoft's version: no hackers, no AI agents. The cause was a bug in a recent network bandwidth update that only surfaced after recent changes to the network infrastructure. The company rolled back the configuration change and patched the component responsible for authentication and routing. Restored servers kept crashing as soon as traffic hit them — hence the staged approach and checks after every step. On Friday evening, November 13, Microsoft apologized on X: "We deeply regret the inconvenience this incident has caused our customers and partners." A full report is promised after the internal investigation. Even early on, Australia's Home Affairs Minister Tony Burke said Microsoft itself had blamed a faulty update.

🧠 Our take

What gets me here isn't the outage itself — it's that Microsoft had to explain, specifically, that this wasn't AI. Posts were circulating on X and LinkedIn tying the outage to layoffs — the company had urged tens of thousands of employees to take voluntary buyouts as it pivots toward AI. IT entrepreneur Tom Lawrence sarcastically asked whether Microsoft was now hunting for the fired engineers' phone numbers. Our read: the technical explanation is honest, but it's answering a mood, not a technical question — the corporation is selling AI agents and can't afford the word "agent" showing up next to the word "outage."

What does this actually change? First, it exposed just how dependent businesses are on connectors: Claude and ChatGPT without access to Microsoft 365 email and calendars instantly lose half their point. Second, it raises an SLA question: if Exchange Online guarantees availability, customers now have leverage for compensation. We think the "AI broke everything" narrative will resurface after every major outage, because the public has no verifiable data — just a status page and a postmortem that hasn't been written yet.

Who wins? Cloud competitors — all they have to do is mention 30 hours and 74 countries. Who loses? Anyone who built workflows around the "email + AI assistant" combo with no plan B. In our experience, large corporations tend to have one; small businesses rarely do.

⚠️ What we don't know

The biggest weakness: we're reading a recap, not the primary sources — we haven't seen Microsoft's status page or its final report on this incident, and the SLA text cuts off at the words "Exchange Online availability." There's no loss figure, no count of affected accounts, and no confirmation that any compensation is coming at all. The date in the source — "Friday, November 13" — is worth verifying yourself via Microsoft's official channels. And yes, the finding rests on a single source.

Forecast: within the quarter, Microsoft will publish a postmortem with a timeline and name the specific configuration change — or explain why it's keeping the details in-house. Check the services status page.

Disclaimer: officially confirmed by Microsoft, but the details come from a single recap — the company hasn't released its full report yet.

Source: iXBT Live