Tech giant Microsoft announced on Saturday that it has fully restored services to its customers, bringing an end to a massive, over-24-hour global outage that crippled digital communications and disrupted public institutions, financial markets, and businesses across more than 70 countries.
​In an update posted to its official Microsoft 365 Status account on X on Saturday morning, the Redmond, Washington-based company confirmed the resolution of the crisis, stating that it had completed its efforts to address the underlying issue of the incident and restored service to all remaining impacted customers and regions.
​The sweeping outage began on Thursday, November 12, 2026, just before 1:00 p.m. ET (18:26 UTC), initially impacting users in the Americas before cascading globally. According to tracking data from Downdetector, more than 130,000 users reported immediate issues on Thursday afternoon.
The root of the failure centered on Exchange Online, the core infrastructure powering Outlook, which triggered a domino effect across interconnected workplace applications, including Microsoft Teams, SharePoint Online, OneDrive, Microsoft Defender, and the company’s Copilot AI assistant.
​By Friday morning, November 13, around 8:00 a.m. EST, Microsoft reported early signs of stabilisation, beginning with the Asia-Pacific region before gradually progressing through Europe and North America.
By Friday afternoon at 17:06 UTC, engineers had significantly reduced backlogged email queues after extending the email delivery timeout window to 48 hours to prevent lost messages, culminating in the full restoration declared on Saturday, November 14.
​Speaking on behalf of the company, Microsoft Executive Vice President Charles Lamanna addressed the incident and assured customers that a thorough investigation is underway.
Lamanna noted in a public update that he is working closely with the engineering team to complete a full post-mortem, which will be published shortly to cover the step-by-step timeline and the numerous repair items aimed at preventing similar outages in the future.
​Both Lamanna and Microsoft’s corporate communications team explicitly ruled out malicious activity, cyberattacks, or glitches related to artificial intelligence agents.
Instead, the company attributed the disruption to a technical error introduced during an earlier network capacity upgrade that surfaced following a recent network configuration change, causing numerous servers to fail automated health checks and overwhelm the remaining infrastructure.
In an official apology issued on Friday via X, Microsoft expressed deep regret for the disruption caused to customers and partners, acknowledging the trust placed in its services and affirming that the company takes that responsibility seriously.
​Because Microsoft serves more than 450 million paid commercial Microsoft 365 subscribers worldwide, the outage created severe bottlenecks across multiple continents and industries. Public sector operations were disrupted across the United States, the United Kingdom, Japan, Singapore, and Australia.
In Tokyo, Japan’s Deputy Chief Cabinet Secretary Masanao Ozaki confirmed that the outage impacted government IT infrastructure and noted that officials were keeping a close eye on the developments, while in London, a UK government spokesperson confirmed that officials were actively working alongside Microsoft to manage the disruption.
​The commercial fallout was equally widespread across aviation, finance, and the broader technology sector. Airlines warned passengers of communication lags, trading firms experienced operational delays, and several banks in Hong Kong reported service interruptions, while some corporations were forced to delay scheduled earnings releases.
Even rival and partner artificial intelligence labs, including OpenAI and Anthropic, reported that their integrations and connectors with Microsoft 365 tools temporarily stopped functioning during the height of the server failures.
​Engineers ultimately resolved the crisis by rolling back recent configuration changes, patching an upstream authentication and routing component, and gradually scaling virtual machine capacity while carefully throttling incoming traffic to avoid overwhelming recovering servers.
