Microsoft says months-old latent bug triggered near-daylong global outage
Microsoft says a latent bug present since July, a failed Teams gateway and 14% capacity loss combined to take Outlook, Teams and Copilot 365 offline for nearly a full day worldwide.

Why it matters
- A latent bug present in Microsoft's systems since July was one of three root causes of the outage.
- The outage took Outlook, Teams and Copilot 365 offline for almost a full day.
- Nearly 14 percent of the primary gateway's capacity was out of service due to an unrelated issue from the previous day.
- Microsoft will release a final post-incident report later this week.
- Charles Lamanna, EVP for Copilot and AI Agents, disclosed the root causes on X.
Microsoft has confirmed that a latent bug sitting in its systems since July was one of three root causes behind the outage that knocked Outlook, Teams and Copilot 365 offline for almost a full day, disrupting businesses, hospitals, law enforcement and government organizations around the world. Charles Lamanna, the company's EVP for Copilot and AI Agents, disclosed the chain of failures in a post on X, drawing on Microsoft's preliminary post-incident report.
The admission matters well beyond one bad afternoon. Organizations that have moved core operations onto Microsoft's cloud productivity stack felt the failure directly, from hospitals to government agencies to police departments. And the explanation points to a familiar and uncomfortable pattern in large-scale software: a defect can sit dormant for months, invisible until an unusual combination of load and reduced capacity wakes it up.
What exactly went wrong?
Lamanna explained that Microsoft normally runs two network gateway systems. These gateways connect separate systems and let data pass between them, effectively acting as the plumbing between parts of Microsoft's infrastructure.
On the day of the outage, that plumbing failed in a compounding way. According to Lamanna, one of the two gateways wasn't working for Microsoft Teams. That failure pushed all the traffic onto the primary gateway, overwhelming it.
The timing made things worse. Nearly 14 percent of the capacity for that primary gateway was out of service at that moment. The capacity reduction stemmed from an unrelated issue that had occurred the previous day, meaning Microsoft entered the incident with less headroom than usual.
Then came the trigger. The combination of higher load and reduced capacity caused the gateway to hit a latent bug that had been present in Microsoft's systems since July. The defect had never fired before. It took this specific pileup of conditions to expose it.
The result was a buildup in traffic requests until Microsoft's servers gave out and stopped functioning. The failure landed on a Thursday afternoon, during peak usage hours in the United States.
Why did recovery take so long?
The outage stretched across almost a full day, and the preliminary report helps explain why Microsoft couldn't simply flip a switch.
Lamanna said Microsoft worked on fixing the issue overnight. But because the bug had been in the systems since July — months before the incident — a straightforward rollback wasn't available. The company could not quickly undo changes tied to code that had been running, dormant, for that long.
There was a second complication. When Microsoft brought its servers back online, waiting traffic flooded in. The returning servers were overwhelmed by the backlog, and they became unresponsive again. Recovery, in other words, wasn't a single moment but a struggle against pent-up demand.
That detail will resonate with anyone who runs large distributed systems. Traffic surges after a widespread outage can turn a partial fix into a second failure, and Microsoft's own preliminary account confirms that dynamic played out here.
Lamanna addressed customers directly in his post. "We are deeply sorry for the impact to all our customers and recognize the impact that this has on their normal business & operations," he wrote.
The quote is brief, but it puts a named executive on the record for an incident that hit critical institutions. It also frames the disclosure itself as part of the response — Microsoft chose to publish root-cause detail within days rather than wait for outside pressure to force it out.
How unusual is a three-cause failure like this?
The structure of the incident, as Microsoft describes it, is a textbook cascade rather than a single point of failure.
The three contributing factors, according to the preliminary report:
- One of Microsoft's two network gateway systems was not working for Microsoft Teams, which pushed traffic onto the primary gateway and overwhelmed it.
- Nearly 14 percent of the primary gateway's capacity was already out of service due to an unrelated issue from the previous day.
- The elevated load and reduced capacity triggered a latent bug that had existed in Microsoft's systems since July, causing traffic requests to build up until servers stopped functioning.
Each element alone would likely have been survivable. A gateway with reduced redundancy can still pass data. A latent bug causes no harm while it stays latent. Combined, on a peak-hours Thursday, they produced a global outage.
This is why the story matters for the broader market. Enterprises keep consolidating their productivity, communication and AI tooling onto a small number of hyperscale platforms, and Microsoft has positioned Copilot 365 as the AI layer across its suite. When the underlying infrastructure for that suite degrades, the blast radius now includes healthcare delivery, government operations and public safety — the organizations Microsoft listed among those affected.
The incident also illustrates a stubborn reality of software engineering at scale. Testing cannot exhaustively cover every combination of load, partial capacity loss and code path. A bug shipped in July carried no visible symptoms until a gateway failure and a 14 percent capacity shortfall landed on the same day. Microsoft's preliminary report is, in effect, a case study in how dormant defects and operational bad luck intersect.
What happens next?
More detail is coming. Microsoft said the information released so far comes from its preliminary post-incident report, and the company committed to releasing a final report later this week.
That final report will be the document to watch. It should spell out what remediation Microsoft intends — whether that means changes to how gateway capacity is tracked, how latent bugs are hunted before they fire, or how the company manages traffic surges when services come back online.
For the customers affected, from hospitals to law enforcement agencies, the stakes are straightforward: they need assurance that the same three-factor collision cannot repeat. Until the final report lands, Lamanna's X post and the preliminary findings are the fullest public account of why a months-old bug brought a day of disruption to organizations worldwide.
Source: engadget.com
More from Elena Vasquez
Show full bio
Market editor covering technology and product news in artificial intelligence and machine learning at AI In Context.
708 articles
Related articles
- Microsoft outage passes 14 hours as fix fails 'as expected'
- Microsoft 365 Outage Drags Into Second Day, Exposing Global Fragility
- Microsoft 365, Outlook and Teams outage passed five hours Thursday
- Microsoft says Microsoft 365 recovery is underway after two-day global outage
- Microsoft outage hit 74 countries, and the company says it wasn't AI