Microsoft outage passes 14 hours as fix fails 'as expected'

Microsoft services have been disrupted for more than 14 hours, leaving email, calendars and AI tools unreliable for users across dozens of countries. The company has confirmed that its first attempted fix is not working as expected, extending one of the longest cloud productivity outages the vendor has suffered in recent memory.

According to the BBC, Microsoft has attributed the problem to "elevated load" across its infrastructure in multiple regions. A number of servers became "unexpectedly unavailable," the company said, pushing extra traffic onto those still running. That cascading failure pattern — surviving servers absorbing the workload of dead ones — explains why recovery has proven difficult rather than quick.

Microsoft executive Charles Lamanna ruled out malicious activity. "We regret that today's outages caused disruptions for our customers, and we are working to address the issue," Lamanna wrote in a post on X. He confirmed the incident was not caused by a cyber-security event.

The stakes are enormous. Microsoft's own website puts the number of people using its 365 apps at more than 430 million. When Exchange Online, Outlook and Copilot falter at the same time, the disruption propagates through governments, airlines, banks and rival AI companies within hours.

What went wrong with the recovery?

Microsoft's own status updates traced a deteriorating timeline. The company first said it was seeing "positive results" from efforts to restart affected systems. It later conceded the fixes were not working "as expected," according to the report.

Microsoft also warned that restarted services can become unresponsive again once traffic returns. In its latest update, the company said it was still recovering frontend servers (the machines that handle user-facing requests before they reach backend systems) while taking further steps to improve stability and capacity.

That warning matters. It means the outage is not a single point of failure but a load-management problem: capacity, not just uptime, is the constraint. Each wave of returning users risks tipping freshly restarted servers back into an unresponsive state.

How big is the disruption?

The numbers are stark:

• Outage-tracking site Downdetector first detected problems just after 16:15 GMT on Thursday.
• Reports about Outlook peaked at nearly 140,000 around 18:00 GMT — more than 2,000 times the normal level.
• Downdetector says the disruption reached 74 countries within about an hour.
• Microsoft counts more than 430 million users of its 365 apps.

A spike of that magnitude, 2,000 times baseline, indicates a systemic failure rather than a regional glitch. The geographic spread within a single hour shows how tightly coupled global workflows have become around a handful of cloud providers.

Which Microsoft services are affected?

Exchange Online, which includes the Outlook email app, is among the services disrupted for many customers, according to the report. Some users may be unable to view certain files or calendar details.

Copilot, Microsoft's AI assistant embedded across its productivity suite, has also been disrupted. Microsoft has said messaging and online meetings were largely functional.

The reality on the ground appears messier. A senior executive at a large Indian brokerage, who spoke anonymously, said Teams calls with big clients have been "a challenge." The gap between Microsoft's "largely functional" framing and user experience in regulated industries will likely draw scrutiny once post-incident reviews begin.

How far do the knock-on effects reach?

The outage has spilled beyond Microsoft's own customer base into the wider technology sector, including two of its most prominent AI partners.

OpenAI said it was investigating an issue affecting its connections to Microsoft's services. Anthropic said it had identified an issue affecting its links to Microsoft 365. Both companies run significant parts of their operations on Microsoft's Azure cloud infrastructure. Their acknowledgment turns a productivity-software story into a broader AI-industry story: when Microsoft's plumbing shakes, the model providers built on top of it shake too.

Which public services were hit?

Government and judicial systems across several countries reported operational strain:

• Singapore's Government Technology Agency confirmed some government systems were affected but said emergency services remain fully operational.
• Singapore's courts asked people to phone about urgent matters.
• The Philippines Supreme Court warned that email responses may be delayed.
• Courts from Los Angeles to Texas to Alberta issued similar notices.

In Australia, the spy agency ASIO and the federal police were also affected, though officials suggested the impact was minimal.

Australia's Home Affairs and Cybersecurity Minister Tony Burke provided the sharpest political readout of the incident. Burke said Microsoft officials in the country told him they believe the situation is similar to the CrowdStrike outage in 2024, where an update had "gone awry."

The comparison is pointed. The July 2024 CrowdStrike incident, triggered by a faulty software update, grounded flights and disrupted hospitals and banks worldwide. If Microsoft's internal assessment mirrors that event, it suggests a quality-control failure in routine infrastructure changes rather than an external shock.

How have businesses responded?

India's biggest airline, IndiGo, said its email support services have been hit. Rival Air India took a different posture, posting on X: "Emails can wait."

Banks and universities have also warned users to watch for fake offers of IT support — a familiar pattern in which attackers exploit high-profile outages to phish credentials through impersonation of help desks.

What happens next?

Technology analyst Paul Spain, speaking to the BBC, delivered the bluntest verdict on the incident. If no cyber-attack was involved, Spain said, the problem points to "incredible incompetence."

That charge frames the accountability question Microsoft now faces. A 14-plus-hour outage affecting 74 countries and more than 430 million app users, traced to load problems rather than an attack, will intensify pressure on the company to explain what failed in its capacity planning and failover design. UNILAD reported it had contacted Microsoft for comment.

Until Microsoft stabilizes its frontend servers and confirms they can hold under returning traffic, customers from courts in Alberta to brokerages in Mumbai should expect further intermittent degradation — and organizations that built single-vendor workflows around Microsoft 365, OpenAI and Anthropic connections will be reassessing how much redundancy their operations actually contain.