How one network change disrupted Microsoft Outlook, Teams & AI tools that rely on them

A financial firm in Taiwan went back to the fax machine to confirm trades. At Europe’s Euronext exchange, email was disrupted. Shipping group Maersk warned customers that updates could take longer. The three businesses were among thousands of those caught up in a Microsoft outage that began on November 12 and took more than a day to ease. One of the world’s largest software companies had changed something in its network. A trading desk ended up reaching for a fax. Microsoft traced the outage to a problem introduced during an earlier expansion of its network capacity. The problem surfaced when the company made a subsequent network change. Microsoft ruled out a cyberattack and said AI played no role in causing the outage. Its public updates describe how servers dropped out of service, why some failed again after being restarted and how trouble in its email system reached applications well beyond Outlook. Microsoft says the Microsoft 365 outage that disrupted Outlook, Teams and Copilot for more than 24 hours is now fixed. In its final status update on the incident, the company said, “We have completed our efforts to address the underlying issue of the incident and restored service.” The disruption hit users in at least 74 countries and left businesses, governments, hospitals and airlines falling back on phone lines, paper records and, in some offices, fax machines. "I want to share what we know so far. This was not a cyberattack or an issue caused by AI tooling. Our amazing team was able to work through a very challenging situation to restore service. Our investigation points to technical challenges and failure modes (cascading failures + thundering herds at our network perimeter) that made recovery more difficult than normal. I'm working closely with the engineering team to complete the full post-mortem. We will publish the post-mortem shortly covering the step-by-step timeline and our many repair items to avoid similar outages in the future. We take our responsibility to our customers seriously and will do everything we can to learn from this and prevent it going forward," wrote Charles Lamanna, EVP of Copilot, Agents, and Platform at Microsoft.

The same traffic, fewer servers

The failure centred on Exchange Online, the cloud service that powers Microsoft’s business email and calendars. As servers became unavailable during routine checks, the rest had to absorb their share of requests. Microsoft said total traffic had stayed steady. Engineers found that a process used to establish secure connections was getting overwhelmed on affected machines, causing them to stop responding. With less capacity available, the usual workload was enough to put the system under pressure. That also explains why restarting servers brought only temporary relief in some cases. A machine could appear healthy after coming back online, then stop responding as it took on work. Microsoft restricted the kinds of requests some servers received while engineers tested a configuration change on a small group of machines. The company also began reversing the recent network change. It brought traffic back in stages so engineers could see whether the servers would stay up as the workload increased. Even after people could get into their inboxes, Microsoft still had to deliver the messages that had accumulated. By Friday morning in the US, the company said the backlog had shrunk substantially, though some email could still arrive late. That made recovery hard to judge from the user’s end. Outlook could open normally while a reply approving a payment, confirming a shipment or changing a meeting time was still held up. For work that depends on a timely response, the disruption lasted until the message arrived.

Why the outage reached Teams, ChatGPT and Claude

Exchange Online also supplies data to other Microsoft products. The calendar in Teams draws on it, as do Copilot features that retrieve work email and appointments. That meant the effects varied by task. A person could still open a file stored directly in SharePoint while a feature in the same product that needed data from Exchange could fail. The list of affected products looked long because several of them called on the same underlying service. That dependency extended to rival AI assistants. Anthropic said the outage caused errors in Claude’s Microsoft 365 integration for roughly 25 hours, disrupting requests to retrieve messages, look up appointments or search for information. OpenAI reported that users could have trouble linking Microsoft services to ChatGPT and using tools that relied on those connections. Anthropic also recorded a separate problem with signing into Claude on November 13. It specifically traced the failures in its Microsoft connector and Claude for Outlook to Microsoft’s service incident. For businesses, the outage exposed how easily a second tool can run into the same problem as the first. Opening Teams to check an appointment, or asking an AI assistant to find the email it came from, can seem like two alternatives to opening Outlook. Both still need access to data held by Exchange. The Taiwanese firm’s use of a fax was a more direct workaround. It gave the firm a way to confirm trades while its usual communications were disrupted.

Microsoft's apology and a 'promise'

Microsoft has promised a full incident report. Its account so far explains how the later network change triggered the outage and why restarting machines was only part of the repair. The report still has to explain what went wrong during the original upgrade and why the subsequent change could take enough servers out of service to strain those that remained. "We are deeply sorry for the disruption this incident has caused our customers and partners. We know the trust they place in our services and take that responsibility extremely seriously. We are seeing broad recovery and stability across our services. This was a technical issue not related to cyber or AI / agents. The incident was introduced by an issue during an earlier network capacity upgrade that surfaced following a recent network change. We have addressed the underlying issue and continue to restore service to remaining customers and regions. Given the scale of demand, we restored services gradually, validating stability at each stage. While this took longer, it helped ensure a stable restoration. We will provide a full post-incident report once our review is complete," the company said in a statement. The answer will help establish whether Microsoft can catch a similar problem while it is still confined to a small part of the network.