A 12-hour Microsoft outage and Nvidia’s lost CEO show how one supplier, or one person, can stall many businesses
Single provider dependency drove two crises on 12 November. Microsoft 365 failed worldwide, and Nvidia’s CEO resigned. Risk teams have warned about both patterns for years, yet few firms had a tested plan.
One crisis struck an email service used in many countries. The other struck a company whose chips power the AI boom. TF sees a shared lesson. Risk gathers wherever one vendor or one leader carries too much weight.
What’s Happening & Why It Matters
A Pattern in Two Failures
Microsoft 365 began failing on the afternoon of 12 November, and Downdetector logged almost 130,000 Outlook problem reports. The Irish Times put the disruption at more than nine hours, and LazyAdmin counted 12 hours 40 minutes by Friday morning. Microsoft later reported wide recovery and stability across its services.
On the same day, Nvidia announced the resignation of CEO Jensen Huang. CFO Colette Kress is interim CEO, and the board has not named a permanent successor. TF covered both events in its outage report and its Huang report.
The two failures seem unrelated. One is a server fault, and the other is a governance crisis. Both exposed organisations with a critical function in a single place.
How One Provider Reaches Everywhere
Microsoft 365 is a bundle for most customers. Email, calendars, chat and documents share the same service. When Exchange Online stalled, Outlook and Teams users in many countries lost work tools together.
The reach went past offices. AP reports UK police could not take online crime reports, and NHS email problems could delay prescriptions. Government services in Singapore, Japan and the Philippines were disrupted too. OpenAI and Anthropic both reported problems with their Microsoft 365 connectors, so a single dependency spread to other providers.
Microsoft has not confirmed a cause, and no malicious activity is indicated. Australia’s Home Affairs minister said Microsoft compared the disruption to the 2024 CrowdStrike outage, which came from a faulty update. Customers saw the same result either way: no email.
When the Provider Is a Person
Huang co-founded Nvidia and led the company for 33 years. Fox Business quotes Kenwell of eToro, who describes Huang as part of the investment thesis. Luria of DA Davidson and Streeter of Wealth Club make similar points about leadership risk.
The market priced the concentration fast. Nvidia lost about $490 billion (about €436 billion) in value, per Yahoo Finance. Its market cap dropped from $4.5 trillion to $4.0 trillion (about €3.6 trillion). Bernstein cut its price target from $400 to $200 (about €356 to €178). CoreWeave, Nebius and Super Micro fell hardest.
The pattern repeats down the chain. Customers depend on Nvidia for chips, and Nvidia depended on one leader for deals and relationships. Nvidia carries about $530 billion (about €472 billion) in commitments and guarantees. Intel, SpaceX, and CoreWeave are among them, so a leadership gap affects many partners.
What a Dependency Audit Covers
Start with a register of providers by business function: email, identity, payments, chips and cloud. Add the second-order links. A connector, a login service or a software update tool can fail while the main vendor seems healthy.
Next, set a tolerance. A firm should know how many hours operations can run without email, and what the fallback is. A fallback nobody has tested is a plan on paper.
For people, the test is the same. Name a deputy for every critical leader, and rehearse the handover. Investors price a missing successor as a cost, as Nvidia’s fall shows.
The Limits of Diversification
Splitting work across vendors costs money and adds complexity. Two email systems need two sets of rules, and two chip suppliers need two software stacks. For some functions, such as AI accelerators, a credible second source may not exist at scale.
The practical goal is known, measured and tested exposure. A firm with a rehearsed fallback is in a stronger position than one with no risk map.
My take: Boards treat concentration as a cost-saving measure, and 12 November showed the bill. The strongest firms will have a second channel and a named deputy in place before the crisis.
TF Summary: What’s Next
Microsoft’s post-incident report should show whether one change caused the outage. Nvidia reports earnings on 17 November, and Huang has agreed to testify to Congress before Thanksgiving. Boards will ask teams for dependency maps next quarter.
MY FORECAST: By 31 March 2027, one large company or government will publish a review of its reliance on Microsoft 365. Nvidia will name a permanent CEO by 30 June 2027.
Related Stories
- Microsoft 365 Outage Hits Outlook and Teams Worldwide
- Jensen Huang Resigns as Nvidia CEO After 33 Years
- Nvidia Stock Falls as Board Opens Probe Into CEO Jensen Huang
- France Blames Iran’s Revolutionary Guards for European Fridge Hacks
- Underwater Data Centres: DeepGreen Advances Maine and Alaska Sites