On September 8, 2025, Google Meet suffered a colossal global outage that started with a small effect in Central Europe and spread globally, impacting millions of people over the course of nearly eight hours. It began disrupting at midnight of the US Pacific time, and the situation only worsened dramatically when a configuration change aiming to rectify the first issue caused mass failures in service.
The first to be hit by the cache service overload were European users
Starting at 00:00 US/Pacific, 8 September 2025, Google Meet had problems loading meet.google.com for a total of 7 hours 57 minutes, as indicated by Google’s status dashboard on its Workspaces. There was a limited strike in Central Europe at 00:00 -7:05 US/Pacific. The first one occurred due to the congested occurrence of a cache service fleet, which is part of Google Static Content Service (SCS). Aspects of the Google Meet UI that are not specific to individual users or meetings are served using SCS.
There are indications of a certain set of Meet hot keys that will not respect internal caching layers since Sep 8, 2025, 12:00 AM US/Pacific (9:00 AM Central European Time). This was only experienced in a few cache servers, resulting in their overloading, hence impacting the error rates to Google Meet users. Engineers at Google Meet were notified of the first outage on 08 September 2025 at 03:12 US/Pacific through an alert the company received about the growing search traffic on Google Meet outages.
Service collapse globally is caused by a configuration change
A change of configuration that was trying to fix the first problem forever led to a global impact by setting up a dedicated cache of Google Meet. This special cache was not pre-warmed and was not sufficiently sized. Following bursts of traffic caused cache misses that overloaded the cache servers and caused the global outage.
Millions are strongly impacted and disrupted worldwide
Between 10:25 and 11:17 US/Pacific, there was a widespread global impact, as many customers were unable to load meet.google.com. According to WebProNews, there was a widespread disturbance in the Google ecosystem at the early hours of Thursday, which caused the shutdown of important Google services, such as Gmail, Google Meet, and even core Google authentication systems, for thousands of users. Around 8 a.m. ET, there started coming in reports of users experiencing error messages when trying to log in, effectively locking them out of emails, virtual meetings, and collaborative tools.
Alternative outage monitoring tools, like popular site DownDetector, recorded more than 6,000 reports during the outage at one time, with its complaints spread across the U.S., Europe, and Asia, exemplifying the extent to which the Google infrastructure extends globally. As a user clicks to meet.google.com and tries to attend a meeting, multiple requests are sent by the user’s browser to SCS.
Measures put in place in recovery and prevention
At 11:09 US/Pacific, the rollback of the configuration change that had been undertaken was confirmed, and at 11:17 US/Pacific, it was confirmed that the mitigation of the impact had been achieved. Google is working on ensuring that it does not repeat this problem going forward and is in the process of completing the following tasks that are in place: Set up early detection monitoring and alerting mechanism to identify future occurrence proactively (Target: Q3 2025), Upgrade configuration rollout safety to handle overload triggers to prevent a recurrence (Target: Q4 2025), and SCS Validate Static Content Service (SCS) to overload triggers to prevent recurrence (Target: Q4 2025).
The Google Meet outage is a harsh demonstration of the importance of how swiftly localized technical problems can spread into a global catastrophe that impacted the lives of millions of users across the world. The incident emphasizes the paramount significance of adherence to robust testing procedures when making configuration changes, as well as the necessity to have a backup system that is sufficiently scaled.