Microsoft Stock: A July Update Knocked Out Over 80% of…

Tech Business

Microsoft 365 Outage Left Over 80% of Gateway Servers Unresponsive

Microsoft blamed a defective July gateway update for a Nov. 12-13 Microsoft 365 outage that left over 80% of servers unresponsive, with final findings due Nov. 20.

By
Grace Kim
Filed
Channel
Tech Business
Read
4 min read

Microsoft said more than 80% of its Microsoft 365 networking gateway servers were unresponsive at the low point of a global outage on 12-13 November 2026, after a defective July update entered production. The Register reported that Microsoft’s preliminary post-incident review said the faulty code ran in production for more than three months.

What caused the gateway failure?

Microsoft shared the review in the Microsoft 365 admin center. Charles Lamanna, executive vice president for Copilot, Agents and Platform, summarized it on X early on 16 November.

Microsoft’s review identified three interacting factors: lock contention from the July code, extra Teams traffic on the shared gateway, and nearly 14% of capacity already offline for an unrelated reason.

The July code caused lock contention inside an internal Windows process, exhausting the machine’s CPU and memory. Load tests used weaker hardware with fewer cores than the failed production servers. Microsoft said contention worsened on newer, higher-core machines, so the tests missed the production behavior.

Microsoft was reversing a recent network change on 13 November to test whether it contributed. The preliminary review instead placed the fault in the July code path. Microsoft scheduled the final report for 20 November.

How long did recovery take?

The sequence from Teams failover through restoration covered 39 hours, while the interval from detection to Microsoft’s healthy-service declaration lasted 24 hours and seven minutes.

  • 01:07 UTC, 12 November: A Teams failover moved traffic onto the shared gateway.
  • 16:03 UTC: Microsoft detected the anomaly.
  • 16:33 UTC: Microsoft logged its first support case.
  • 12 November close: Microsoft stock closed at $523, about five hours after detection.
  • 00:30 UTC, 13 November: Engineers traced the problem to the process negotiating connections.
  • 07:20 UTC: Mitigation began.
  • 16:10 UTC, 13 November: Microsoft declared services healthy, 24 hours and seven minutes after detection.

Lamanna wrote that “our normal rollback, flighting, and feature flags were unable to quickly revert the change.” He tied that limitation to a main defect latent for several months.

Which services remained unavailable?

Teams, OneDrive, SharePoint and Office applications shifted to another gateway and recovered sooner, but Exchange Online could not move for scale reasons, leaving Outlook mailboxes unavailable throughout the longer restoration. Defender and Intune also suffered disruptions.

Downdetector logged more than 7.4 million reports worldwide, peaking near 20,000 per minute, according to Reuters reporting carried by CBC News. CBC also reported that Copilot was among the affected services. Lamanna also leads the Copilot business.

A Microsoft spokesperson apologized for the disruption to customers and partners. Lamanna separately apologized for its impact on normal business operations.

What has Microsoft changed?

Microsoft has increased gateway capacity, improved monitoring and added safeguards against traffic redirection, while the corrective software only covers Microsoft 365 gateway servers and does not apply to broader Windows deployments.

Engineers built an update that returned the Windows code path to its earlier state. The fix applies only to Microsoft 365 networking gateway servers. Enterprise customers therefore have no Windows patch to install.

How large was the insured loss?

Morningstar DBRS’s Mario De Cicco estimated global insured losses at $300 million to $500 million, versus $400 million to $1.5 billion for the 2024 CrowdStrike outage. The figure remains an analyst estimate rather than a confirmed loss.

eToro market analyst Josh Gilbert told Reuters: “The bigger worry is that this is not the first Microsoft 365 outage this year.” He said reliability concerns could make IT departments think twice before placing more business with one provider. Microsoft 365 also went down in January 2026, and Azure had an outage in October.

How did investors react?

Microsoft stock closed at $517.99 on Monday, 16 November, down 1.12%. It traded at $515.90 before Tuesday’s open at 08:22 GMT. On 13 November, it reached $513.82 intraday and closed at $523.88.

From the 12 November close through Monday, EODHD data show Microsoft fell 0.96% while the QQQ Nasdaq-100 tracker fell 3.84%, a difference of about 2.9 percentage points. Across 11-16 November, Microsoft slipped 0.60% while QQQ lost 4.15%, a 3.55-point outperformance. On an index set to 100 at the November 11 close, Microsoft ended at 99.4 and QQQ at 95.9.

The stock remained 7.4% below its $559.15 November 4 high and $13 above its $504.96 November 2 low. The November 13 outage dip held above that low.

It sat above FinanceFeeds’ $505 bear reference, set after the $545.18 close on 9 November. It remained below the $520.19 bear case in FinanceFeeds’ 24 October prediction.

What happens next?

Microsoft faces an ex-dividend date of 19 November for a $0.98 payout, a final report on 20 November, and expected earnings around 26 January, when Azure growth and capital spending will test investor confidence. Azure revenue grew 48% in the September quarter. Microsoft spent $41 billion on capital expenditure in the June quarter.

Microsoft’s added capacity and new safeguards now face their next reliability test. So far, investors have treated the incident as an operations problem with limited insured cost. Evidence of recurrence could strengthen the case for customers to diversify, while a clean final report would weaken that concern.

Source: financefeeds.com

Share this article:

More from Grace Kim

Grace Kim

Show full bio

Market editor covering people, launches and day-to-day practice in Semiconductors, chip manufacturing, hardware, AI compute at Chip Dispatch.

1,045 articles

Related articles

« Previous article