Developing Situation
Evidence current through: 8:54 a.m. Pacific Time (16:54 UTC), Monday, November 16, 2026.
Microsoft’s preliminary post-incident report, released on November 16, dates the impact from 16:20 UTC on November 12 to 16:10 UTC on November 13. Developments after the time above are not reflected, and Microsoft’s final review is due on November 20. Pacific Time is the time at Microsoft’s headquarters in Redmond, Washington. Times in the text are given in UTC.
Research Note In Two MinutesNote synopsis · AI drafted · Analyst reviewed
Drafted by AI from the full research note and reviewed by a Greyhound Research analyst before publication. The analysis and conclusions are Greyhound Research’s own.
On November 12, 2026, Microsoft 365 failed as one system. Mailbox access in Exchange Online failed from 16:17 UTC, a wider incident was opened at 18:26 UTC for every service that depends on it, and security and management services, including Defender, Intune, and Purview, joined the affected list that evening. On November 14 Microsoft said it had restored service to the remaining affected customers and regions. Its preliminary post-incident report, released on November 16, traces the outage to a latent defect from a fix it first deployed in July, exposed by a surge in load while capacity was already out of service. No count of affected customers has been published.
Greyhound Research’s position is that consolidation on one suite remains the right strategy for most enterprises, and that the standard terms around it are no longer enough. Three things decide a buyer’s exposure, and none comes with the license: knowledge of the vendor’s change gate, a place in the order of restoration and notification, and fallbacks the buyer has written down and rehearsed.
Three findings carry the argument. In a review of ten multi-service incidents since January 2023, Microsoft’s own account attributes six to a change it made and two more to its own configuration behind an outside trigger. Australia’s government said it activated a non-binding agreement with Microsoft to secure priority in recovery, which made the order of restoration visible as something a customer may or may not hold in writing. And a scorecard of five suites shows Microsoft has put the most in writing about outages, while no vendor states an order of restoration.
The research note sets out ten moves for chief information officers and chief information security officers, a ten-move Greyhound CXO Runbook for the wider leadership, eight negotiation terms with the lever that reaches each, seven boardroom questions, and ten questions for Microsoft. None of the buyer actions waits on the vendor. The first three are to export the incident record before it ages out after 30 days, to file the credit claim by December 31, 2026, and to write and rehearse one page for each process that cannot stop.
The evidence has limits. Microsoft’s statements are documented, and its cause rests on a preliminary report ahead of a final review due on November 20. The service level terms are Microsoft’s standard edition of October 2026, which a negotiated agreement may amend, and the reading of restoration order rests on one minister’s statement and on the rollout rule in Microsoft’s preliminary report, which does not mention the agreement. Greyhound Research states the conditions that would strengthen or soften its position, and will revise the status objects when Microsoft’s final review is published.
In This Research NoteSection guide · AI drafted · Analyst reviewed
Twenty-three sections in eight parts, each with a short summary. Select a title to go straight to that section. Each section opens with a Greyhound Spotlight, the one fact in it a reader would not expect, and eighteen end with a block named The Deeper Read, which holds the fuller argument and the passages it rests on. Select any line in that block to open its full text.
- Part 1What Happened
- 1.1What Broke: A Suite That Failed As One System
The outage in plain terms: what stopped, who it reached, what kept running, and the standing assumption it tests.
- 1.2Microsoft’s Account: What Was Said, And What Remains Open
Microsoft’s own statements in order, the reversal on whether a change was involved, the cause it named a day later, the credit the company has earned, and how its preliminary report fares on four tests.
- 1.3The Change Record: Ten Incidents, And What Microsoft Said Caused Them
Ten multi-service incidents since January 2023, sorted by the cause Microsoft stated. Eight trace to a change or a configuration, and the gate that should catch them is documented.
- Part 2How The Failure Spread
- 2.1One Estate, One Blast Radius: Why The Services Fell Together
How shared mail, identity, and graph dependencies put the work layer, the control layer, the assistant, and other vendors’ integrations inside one incident.
- 2.2The Security Layer: When Safe Links Cannot Check
What Microsoft’s documentation says a link check does when it cannot complete, which part of that is the customer’s own setting, and the evidence a security team should pull before it expires.
- 2.3One Process, Hour By Hour: A Supplier Payment Through The Outage
An illustrative case built from the documented timeline: one payment approval in three organizations that differ only in preparation.
- Part 3Who Is Restored First, And Who Must Be Told
- 3.1Restoration Order: The Queue Australia Made Visible
The clause Australia invoked, what it commits and what it does not, the rule Microsoft says set the order of recovery, and the three questions any enterprise can ask.
- 3.2The Regulators’ View: Who Microsoft Must Now Tell
Jurisdiction by jurisdiction: where the provider itself is supervised, where the duty sits with the dependent firm, the legal status and dates of each instrument, and the gap left for everyone outside finance.
- Part 4What The Contract Gives
- 4.1Service Credits: A Refund On A License
How the credit is calculated and claimed, what it does and does not measure, and why a monthly clock cannot see four incidents in one year.
- 4.2The Greyhound Outage Terms Scorecard: Five Suites, Twelve Criteria
A proposed analytical device: what Microsoft 365, Google Workspace, Slack, Zoom, and Salesforce have put in writing about outages, cell by cell, and the one row that is blank for all five.
- 4.3Dependency And Exit: What Leaving Would Actually Take
What each vendor documents about getting data out, why leaving for good and leaving for a day are different capabilities, and why a second suite in reserve is the wrong answer.
- Part 5The Risks And The Counter-Argument
- 5.1Risks To Watch: Where The Next Outage Does Its Damage
Six risks that begin when the service comes back, each with an early warning, a remedy the buyer controls, and an owner.
- 5.2The Strongest Objection: Concentration Still Wins
The best case against this research note, taken limb by limb. Most of it stands, and the one claim that falls is the one the standard contract depends on.
- Part 6What To Do, By Role
- 6.1CIOs And CISOs: Ten Moves Before The Next Outage
Five moves for the chief information officer and five for the chief information security officer, ordered by when they start, none of which waits on Microsoft.
- 6.2The Greyhound CXO Runbook: Ten Moves For The Wider Leadership
Ten decisions that belong outside technology and security, each with the executive who owns it, starting with the list every other move rests on.
- 6.3The Greyhound Negotiation Lens: Eight Terms, Clause By Clause
Eight terms a buyer can ask for, the lever that reaches each, the concession it is likely to cost, and what to do when the answer is no.
- 6.4Boardroom Questions: What Directors Will Need Answered
Seven questions a director can put to management, why each matters, and the evidence to ask for.
- 6.5IT Services Firms: Caught Between The Platform And The Client
Why an IT services firm met the outage as tenant, operator, and claimant at once, what the night looked like on India time, and what providers can build from it.
- 6.6Questions For Microsoft: What Buyers Will Ask Next
Ten questions only Microsoft can answer, each with what the answer would settle and what Greyhound Research would update on receiving it.
- Part 7The Final Greyhound Standpoint
- 7.1Concentration And Restoration: The Final Greyhound Standpoint
The judgment in full, the state of the evidence behind it, and the conditions that would strengthen or soften it.
- Part 8Appendix, Sources, And Disclosures
- 8.1Appendix: The Evidence Register, Source By Source
Every source behind this research note, with when it was read, what was taken from it word for word, and where that material is used, in one drop-down for each class of source. Passages already quoted in The Deeper Read are not repeated.
- 8.2Sources And References: Every Document Behind Every Claim
One hundred and eleven sources in seven classes, from Microsoft’s own reviews, documentation, and terms to other vendors’ terms, regulators’ instruments, incident timelines, and news reporting, each also linked in the sentence that carries it.
- 8.3Disclosures: Independence, Financial, Intended Use, Methodology, Authoring, And Copyright
Independence first, including the statement that Microsoft was not offered a pre-publication factual review, then the Financial Disclosure, intended use, research method and its limits, authoring, and copyright terms.
Where To Start, By RoleReader guide · AI drafted · Analyst reviewed
Find your role and read its sections in the order shown. Each line takes you straight to the section.
Chief information officer
Chief information security officer
Chief financial officer
General counsel
Chief risk officer
Head of procurement
Head of business continuity
Board member
IT services leader
Public-sector technology leader
Part 1What Happened
What Broke: A Suite That Failed As One System
Greyhound SpotlightThe Bank of England, Services Australia, and NHS England all reported that their core services kept running. What stopped was the mail and the meetings that carry approvals and handovers between those systems.
On November 12, 2026, Microsoft 365 stopped behaving like a set of products and failed like one system. Microsoft’s service health updates, as compiled in published incident timelines, record a Teams incident in the Americas from 14:45 to 15:53 UTC and mailbox access failing in Exchange Online from 16:17 UTC under incident EX1550543. A second incident, MO1550821, was opened at 18:26 UTC for every service that depends on Exchange Online. At 17:57 UTC on November 13, more than 25 hours after the mailbox failures began, Microsoft reported “broad recovery and stability” and named a cause. By two reports, it said at 06:09 UTC on November 14 that it had “restored service to the remaining impacted customers and regions,” almost 38 hours after the mailbox failures began. Its preliminary post-incident report, released on November 16, traces the outage to a latent defect in Microsoft’s own gateway servers.
Read More "Microsoft 365 Outage – When The Suite Goes Dark, Restoration Order Becomes The Control"













