Microsoft's Outlook, Teams, SharePoint, Copilot and Defender security tool failed together on November 12, first with an error message about heavy load and then, Microsoft found, because a large number of its servers had become unavailable. The interesting part is what the list of casualties shows about how work now runs.
One fault underneath reaches everything on top
These services do not stand alone. They share servers, a sign-in system and the network that carries traffic to them, the way the lights, the lifts and the fire alarm in an office building can all hang off one electrical panel. When the panel trips, nobody asks which of the three was broken.
The November failure looks like that kind of fault. Microsoft first blamed load, then found that traffic had not gone up at all: servers had dropped out, and the ones still running were carrying their share and more. That is why services that seem unrelated all slowed or stopped together. The earlier failures, a configuration change in October 2025 and a sign-in fault at the end of August, spread across the suite in the same way.
What the AI assistant changes
Before assistants, a mail outage meant people phoned each other and worked from paper and memory. Now the assistant is often where work starts: it summarizes the meeting, finds the file and drafts the reply. When it is down with everything else, the tool that knows where things are is gone at the moment people most need to find them.
Take a sales manager who asks Copilot each morning for the day's customer follow-ups, pulled from her inbox and last week's meeting notes. During an outage she has no inbox, no notes and no assistant. If a 20-person team works at half speed for a four-hour afternoon, that is 40 hours lost, a full working week for one person.
The security tool is on the list too, and that matters for a different reason. When mail fails, people reach for personal email or a messaging app to keep a deal moving. Defender, which watches for odd sign-ins and risky attachments, was among the services that failed, so the busiest hour for improvising is an hour when the watcher is out.
The remedy on offer is small. Microsoft guarantees 99.9% uptime each month, and when it misses, a customer may get a service credit against the subscription. That does not repay a lost afternoon.
What to set up before the next one
Put one page outside Microsoft's systems, on paper or in a different service. On it, write the phone numbers of the ten customers or suppliers you cannot afford to lose, the three jobs that must happen that day whatever fails (the payroll cut-off, the afternoon shipment), and one messaging group the company has approved for emergencies, so improvising has somewhere safe to go. Then ask which jobs the assistant now does that nobody on the team could still do by hand, and keep one person who can.