Microsoft has finally explained what knocked Outlook and other Microsoft 365 services offline around the world last week. The answer is not one fault but three, and the first of them had been sitting unnoticed on company servers since July.
The explanation came from Charles Lamanna, Microsoft’s executive vice president of Copilot, Agents and Platform, in a post on X on Monday 16 November. It matches the preliminary post-incident report Microsoft shared with business customers on 15 November, with a final report due on 20 November.
Three problems lined up at once
Lamanna said the outage hit the networking gateway that sits in front of many Microsoft 365 services. The first problem was a latent defect that made servers overwhelmed and unresponsive under certain load conditions.
The defect came from a fix written specifically for those gateway servers. Microsoft said the defect and the fix do not apply to Windows more widely.
The second problem was a surge in load. About 15 hours before the outage, Teams moved from a second gateway onto the primary one that also carries Exchange Online, Microsoft’s business email service.
The third problem was capacity. Nearly 14 percent of the gateway’s server capacity was already out of service because of an unrelated issue from the previous day. On its own, Microsoft said, that was within its spare capacity.
Together, the extra traffic and the missing servers set off the July bug during the busiest hours of traffic in the United States. At the peak, more than 80 percent of the servers behind the gateway had stopped responding.
Why the bug stayed hidden since July
The faulty change began rolling out in July and ran on servers worldwide for more than three months without causing any trouble. Microsoft said its staged deployment cleared every check along the way.
The problem was lock contention inside an internal process, where too many tasks queue for one shared resource and the server stops answering. Microsoft said its load tests leaned on weaker hardware with fewer cores, while the contention grows worse on newer machines with more cores.
Lamanna joined Microsoft in 2013, when the company acquired his startup MetricsHub. He became executive vice president of Copilot, Agents and Platform in March 2026 and took over Office and Teams on 1 October 2026.
What the outage hit
The disruption ran from 16.20 UTC on Thursday 12 November to 16.10 UTC on Friday 13 November, close to 24 hours. Exchange Online, the email system behind Outlook, was hit hardest, and most users could not reach their mailboxes or send mail normally.
Teams, SharePoint, OneDrive, Copilot and several security tools were also affected where they depend on Exchange Online. The outage reached users in more than 70 countries, including government offices, hospitals and businesses.
Microsoft restored services gradually through 13 November and confirmed on Saturday 14 November that access was fully back. The company said from the start that the failure was a technical issue, not a cyberattack and not connected to AI agents.
The apology, and what comes next
We are deeply sorry for the impact to all our customers and recognize the impact that this has on their normal business & operations.
Charles Lamanna, executive vice president of Copilot, Agents and Platform at Microsoft
The apology was echoed on Monday by Ravi Bhat, Microsoft’s commercial solutions and AI officer for South Africa. Speaking in Cape Town at the Africa Tech Festival, he called it a technical issue with network grid capacity, and said it had nothing to do with AI or a cyber incident.
Microsoft has already reverted the faulty change, expanded server capacity and added safeguards against the same kind of cascade. The final post-incident report, due on 20 November, will carry the full list of repairs.
Business customers affected by the outage can check whether they qualify for service credits under Microsoft’s service level agreement. The episode is a reminder of how much everyday work runs through a small number of cloud services.
A defect that caused no harm for three months still took email offline for millions of people once two ordinary events lined up beside it. That is the part of this story that should worry every team that treats email as something that simply works.
Fake outage emails are now circulating
There is one more thing to watch for. Microsoft Threat Intelligence said on 16 November that attackers are sending fake Microsoft 365 service health emails that use the outage as their lure. The full warning is on Microsoft’s security blog.
The messages claim that emails held during the outage will be deleted within 24 hours unless the reader clicks a restore link. Microsoft said the messages come from domains such as moments-notice.com, planetmaison.com and newbooker.com, and lead to a page at reallifeaccounting.com.
Microsoft said the activity looks like credential phishing, but it has not confirmed that any credentials were stolen or any malware was delivered. Its Defender products are blocking the malicious domains. Genuine service health messages never send users to another website to recover held mail.