
The day the email stopped
The problems began shortly after 5 p.m. Norwegian time on Thursday. According to Downdetector, the disruption had spread to 74 countries within an hour, with almost 140,000 problem reports for Outlook. On Friday, the Storting said all members of parliament were without email, and the Ministry of Digitalisation confirmed that the same problem affected several ministries. Microsoft says there is no sign of an attack. It says servers unexpectedly became unreachable, which overloaded the rest of the system. The company has not yet published the exact cause of the fault.
This is the second major Microsoft 365 outage this year, and it was still continuing on Friday.
Three companies, most of the cloud
“The cloud” means servers that companies and governments rent instead of running their own data centres. According to Synergy Research Group, Amazon Web Services is the largest cloud infrastructure provider with a 26 percent share, and Microsoft’s Azure is second with 19 percent. Google Cloud is third with 16 percent. The three largest providers grew strongly in the September quarter because demand for AI is consuming server capacity. According to Synergy, AI services are already about a third of the whole market. Sources: TechInsights, Synergy Research Q3 2026 update; TelecomTV.
Market share does not tell the whole story, because many smaller services run on top of the big clouds. When one of them stumbles, the effect spreads through several layers.
It has not been long since the last time
2025 offered examples of how a small fault spreads:
- AWS, October 20, 2025. A timing error in the DNS automation of the DynamoDB database service cut off many services that depend on it. A fault in one region had global effects. By AWS’s own timeline the event ran from 11:48 p.m. PDT on October 19 to 2:20 p.m. PDT on October 20, about 14 and a half hours. Sources: AWS post-event summary; ThousandEyes analysis.
- Azure, October 29, 2025. A configuration change to the Azure Front Door service caused a global disruption that affected Microsoft 365, among other things.
- Cloudflare, November 18, 2025. A major content delivery network failed and took many well-known sites offline for hours. According to W3Techs figures reported by Botcrawl, Cloudflare serves about 20 percent of all websites. Sources: ISPreview; Botcrawl.
The pattern is clear: the cause was not an attack but a faulty automated change that spread too fast.
Governments are in the same boat
The dependence is not only a business matter. According to Altinget, eight in ten Norwegian organisations use the cloud of Microsoft, Google or Amazon. DFØ, the government’s procurement agency, signed the largest cloud agreement in Norwegian history with four US technology companies. According to NITO, the National Security Authority (NSM) has recommended a separate national cloud service, but none has been built.
Political parties have also raised the risk. The Socialist Left Party (SV) and the Green Party (MDG) demanded action after a survey by the Swiss company Proton claimed that 96 percent of Norwegian companies use American technology. Note that Proton is itself a competitor, and the survey was based on identifying email servers, so the figure is indicative only. Altinget on the Proton survey. NSM’s own warning is plainer: technical problems, cyber operations or changed terms at a large provider could have significant consequences for Norwegian organisations that deliver services society depends on. Altinget on NSM’s Risiko 2026 assessment.
A frequently cited example of the risk is Karim Khan, the chief prosecutor of the International Criminal Court. He was cut off from Microsoft services in May 2025 after US President Donald Trump imposed sanctions on him, as Minister of Digitalisation Karianne Tung has recalled. Altinget.
What if all three went down at once?
A simultaneous full outage at all three companies is very unlikely, because their systems are technically separate. The danger would lie in a shared dependency, for example a broad cyberattack, a common software flaw, or a failure in the basic structures of the internet such as DNS and content delivery networks.
The consequences below are an estimate, not a forecast. Banks, payments, logistics and logins would slow down or stop widely. Authorities would switch to backup systems whose condition is not publicly known. In this week’s outage, some companies were already forced back to the fax machine.
The answers are still coming
In May, Telenor announced it would set up Telenor Sovereign Cloud, a company whose services are stored in Norway and governed only by Norwegian law. A pilot with selected customers comes first, and a commercial launch is planned for the first half of 2027. Telenor press release. According to NITO, the German and Austrian armed forces are replacing Microsoft software with open solutions, and Denmark’s Aarhus is moving to European ones. In Norway, the municipality of Larvik says it has dropped Microsoft and Google for case management and saved millions. On the other hand, the digitalisation minister has called complete sovereignty a utopia, and some business representatives take dependence more calmly and stress the quality of American services.
So the question is not whether the cloud should be used, but how much damage a single mistake is allowed to cause.
Sources: TV 2, CNN, Al Jazeera, Euronews, The Conversation, TelecomTV (Synergy Research), Tom’s Guide, ThousandEyes, Parametrix, IncidentHub, Altinget, NITO, International Finance, Stortinget.