Microsoft cites progress fixing outage that roiled the world

Microsoft Corp. said it was making progress in restoring access to its flagship software and end an outage that paralyzed government offices, traders and businesses around the world.

The company said in an emailed statement at about 12:30 p.m. New York time that it had “addressed the underlying issue” and was restoring service gradually to remaining customers experiencing problems. An update posted earlier in the day described improvements, but warned that users of more than 20 products and services may continue to encounter errors.

That included Outlook, Teams and the Copilot artificial intelligence tool, whose users experienced a second day of missed meeting invitations and unresponsive web pages on Friday. Previous disruptions have rarely lasted more than several hours. 

The outage cascaded around the world, with users in more than 70 countries reporting problems, according to Downdetector, which recorded hundreds of thousands of complaints. Office and government workers from Tokyo to London and New York struggled to adapt, often reaching for analog 20th-century tools like the telephone and, yes, paper. Some rejoiced that they basically got the day off because they couldn’t get any work done.

But their bosses were already counting the costs. Some companies, including Swiss luxury group Richemont and US-based truck and bus manufacturer Workhorse Group Inc., were forced to delay reporting their financial results. Critical services including police departments and hospitals were affected, with the UK’s National Health Service rescheduling some procedures, according to people familiar with the matter.

The failure and the time it took to find a fix was a black eye for Microsoft, whose Windows operating system and productivity apps remain critical infrastructure in the age of artificial intelligence. The company’s M365 software bundle has more than 450 million commercial users. Once again, critics highlighted the risks of entrusting so much power to one provider — even as many people understood that unshackling themselves from Microsoft’s ubiquitous products was unlikely anytime soon. Investors largely shrugged off the incident, with Microsoft shares little changed.

“We do not expect the incident to meaningfully change Microsoft’s competitive position in the near term, given the high cost and complexity of replacing deeply embedded productivity workflows,” William Blair & Co. analyst Jason Ader wrote in a note. “Longer term, however, it could make some customers more receptive to alternatives, particularly Google Workspace and emerging productivity offerings from Anthropic unveiled earlier this week.”

The error was introduced during a network capacity upgrade, a spokesperson said. The company’s response, detailed in hourly updates to the company’s status page, described some false starts as Microsoft struggled to pin down the issue.

More than 1,000 engineers and other employees gathered on video calls led by cloud chief Scott Guthrie and Charles Lamanna, who oversees the M365 bundle and other workplace applications. Sessions ran late Thursday and spilled into Friday morning.

Exchange, the digital backbone powering email and calendars in the M365 bundle, would start up on a server and quickly consume its available resources, leading it to become unresponsive. Engineers investigated whether the application was mistakenly firing traffic at databases rather than local memory, according to a person familiar with the matter. Some suspected a misconfiguration in Microsoft’s Azure cloud network could be at fault.

Publicly, Microsoft said it would provide more detail after the disruption was over, adding that there was no evidence of malicious activity. It was also unrelated to AI or agents, a spokesperson said.

“We know our customers rely on our services to get important work done every day, and regret the impact these outages had on those who depend on us,” Lamanna said in an emailed statement. “Our teams are reviewing what happened and how we can prevent similar issues going forward.”

The first signs that Microsoft was experiencing issues came at about 1 p.m. Thursday on the US East Coast. Reports began spiking on Downdetector, and disgruntled customers took to social media to gripe and say, essentially, “Here we go again.” Many recalled previous outages, including a 15-hour disruption at Amazon Web Services last year that hit hundreds of websites. And few have forgotten the mother of all outages in 2024, when a faulty software update from cybersecurity firm CrowdStrike Holdings Inc. grounded flights and crashed systems around the world, causing billions of dollars in damages.

The latest incident wasn’t so far-reaching, but impactful all the same. PVH Corp., the apparel conglomerate that owns the Calvin Klein and Tommy Hilfiger brands, announced Thursday that it would cut about 4% of its global workforce. But some of the difficult conversations, scheduled to start in Asia on Friday morning, didn’t take place because of the outages, leaving employees uncertain of their future, according to a person familiar with the matters, who requested anonymity to describe private matter. PVH didn’t immediately respond to a request for comment.

In Arapahoe County, Colorado, which includes some Denver suburbs, the sheriff’s office couldn’t process online bond payments to release detainees from jail because of missing email traffic. People were asked to post bond in person at the jail instead.

In the UK, Guy’s and St. Thomas’ warned patients that email was offline, though the London hospital’s procedures were going ahead as planned. A number of local police forces also said they couldn’t take crime reports online, according to the Guardian newspaper.

Traders in Singapore, Indian brokerages and officials in Tokyo also reported issues with Microsoft services on Friday. Euronext sent a warning that its email communications were unavailable.

Japan’s finance ministry said in a statement that it had been unable to receive emails for most of the day. Without access to their email, government ministries resorted to pinning up paper notices to inform journalists about daily schedules.

Staffers at the Reserve Bank of Australia were also impacted, with Outlook down across departments at the bank, but critical operations ran normally. In New Zealand, immigration officials paused the processing of all visa applications.