Microsoft's Capacity Upgrade Triggered a Day and a Half of Global Outage

By Grace Kim

Downdetector logged nearly 130,000 reports of Outlook problems in the U.S. alone at the peak of a Microsoft outage that knocked out email, Teams and other Microsoft 365 services for roughly a day and a half.

Microsoft confirmed on Friday that the trigger was its own infrastructure work. In a statement, the company said: "The incident was introduced by an issue during an earlier network capacity upgrade that surfaced following a recent network change."

That's the whole explanation so far, and it leaves plenty unclear. Microsoft hasn't defined what it means by "network capacity upgrade" or "recent network change." What's evident is the sequence: one change planted a latent fault, a second change exposed it, and the result was a global outage.

What actually broke?

According to Microsoft's updates, the failure hit systems that manage traffic between servers. Those systems became overwhelmed. Servers stopped responding. Traffic rerouted to other servers, which then failed too.

That cascade explains the slow recovery. Once engineers identified the fault, returning traffic flooded back into the system and knocked servers offline again.

Microsoft said it was "deeply sorry" for the disruption and has promised a full post-incident report.

Takeaway: a routine capacity upgrade left a hidden fault that a later change turned into a day without email for millions.

Why did Microsoft mention AI at all?

One line in Microsoft's statement stood out beyond the apology: "This was a technical issue not related to cyber or AI / agents."

The cyber clarification makes sense. Businesses running on Microsoft's stack need to know immediately whether their data was at risk. It wasn't a breach.

The AI denial is the more revealing part. As Inc.com tech columnist Jason Aten noted, "when a company has to explicitly reassure customers that AI wasn't responsible for breaking its products, that's a problem."

Does the trust math add up?

Microsoft has spent billions of dollars on AI and embedded it across its business — in Windows, in Office, on Surface laptops. It hosts AI workloads for enterprise customers. And, as the source notes, it has cut thousands of jobs while describing AI as transforming how work gets done.

To be clear, there's no evidence those layoffs contributed to this outage. But as Aten points out, reasonable people will ask whether fewer employees and heavier reliance on AI affects a company's ability to maintain the complicated infrastructure millions of customers depend on.

Takeaway: Microsoft's pitch is that customers can trust it to run the technology they rely on — and when that fails, the customers absorb the cost.

What happens next?

Two commitments are on the table: fix the fault and publish a full report. The harder question, per Aten, is why Microsoft "didn't catch the problem before it caused a day with no email."

The reputational issue outlasts the outage. A company that feels compelled to volunteer that AI didn't break its products is signaling it knows customers don't fully trust that technology yet. Until that trust gap closes, every future outage — AI-related or not — will carry the same question.

Opinion portions of this report reflect analysis by Inc.com tech columnist Jason Aten (@jasonaten).