Outgrace, home

Technology & AI

Microsoft restores Teams and Outlook after near-24-hour outage

A gateway fault left more than 80% of one server fleet unresponsive at its lowest point. Morningstar DBRS puts insured losses at $300m–$500m.

By the numbers

Figures from this story's sources.

people use the affected Microsoft 365 apps, per Microsoft Source for 430 million (opens in a new tab)
430 million
Downdetector outage reports worldwide, per Ookla via CBC News Source for 7.4 million+ (opens in a new tab)
7.4 million+
global insured losses estimate, Morningstar DBRS via CBC News Source for $300m–$500m (opens in a new tab)
$300m–$500m

Microsoft said Teams, Outlook and Copilot were fully restored after an outage of more than 24 hours, according to CBC News. BBC News put the outage at nearly 24 hours. At its lowest point, more than 80% of servers in one networking gateway fleet were unresponsive, Business Standard reported, citing Microsoft.

Microsoft's root-cause account, as Business Standard reported it, names three factors. A latent defect from a July fix to gateway servers slipped past load tests that used weaker hardware. About 15 hours before the failure, Teams moved onto the primary gateway, which also serves Exchange Online. Roughly 14% of gateway capacity was already offline after an unrelated fault. Recovery then stalled when returning servers were swamped and failed again.

The BBC and The Guardian count 70-plus affected countries; Business Standard says 72 to 74, citing Downdetector. CBC reports more than 7.4 million Downdetector reports worldwide, while Business Standard cites more than 1.3 lakh Outlook problem reports at peak, a narrower scope. The start time also differs: The Guardian says about midday Thursday US time, Business Standard says late Thursday.

The Guardian reported that NHS England hospitals fell back on paper records. UK cyber minister Liz Lloyd told the BBC the impact on 999 calls was limited. Canada's national rail operator warned passengers might miss email trip updates, BBC News reported.

Morningstar DBRS puts global insured losses at $300m–$500m, CBC News reported, against $400m–$1.5bn for CrowdStrike's 2024 outage. BBC News reported that Microsoft's first cause statement pointed to a network capacity upgrade introduced after a recent network change. The Guardian reported that the fault was introduced during an earlier network capacity upgrade and that Microsoft said it was not related to cyber.

Why it matters

The lesson is concentration. One provider's gateway carried email and chat for airlines, courts, a national rail operator and trading firms, according to the reports. On Morningstar DBRS's estimate, insured losses are far below the CrowdStrike toll, and CBC News reported the hit to insurers is expected to be limited. Our reading: ask which processes still work with email gone. Taisho was handling orders by phone, according to The Guardian.

What’s next

Charles Lamanna, Microsoft EVP, promised a post-incident review "later this week", and BBC News reports the full report will follow once that review is complete.

Sources

Checked against 4 outlets

  • CBC News — Restoration announcement, outage duration, Downdetector totals, Morningstar DBRS insured-loss estimate, CrowdStrike comparison, expert comment and trading disruption
  • BBC News — Duration of nearly 24 hours, network capacity upgrade cause statement, 430 million user figure, first acknowledgement time, UK minister's remarks on 999 calls, Canadian rail impact and post-incident r
  • The Guardian Technology — Start time as midday US Thursday, country count, NHS England impact, earlier network capacity upgrade and not-related-to-cyber wording, and business and government disruption
  • Business Standard — Microsoft's root-cause account of three combined factors, the 80% server figure, Downdetector country count, Outlook report count, start time and Lamanna statement

Found something wrong? Our corrections policy explains how to tell us. How we report the news. Add Outgrace as a preferred source on Google.