Microsoft Outage Pushes Japan's Financial Market Back to Fax Machines

Japan's Financial Market Turns Back to Fax

A global Microsoft outage pushed parts of Japan's financial sector back to fax machines on Friday, after the email service market participants use to confirm trade details stopped working. The disruption began Thursday and spread across Microsoft 365 business apps, leaving Outlook, Teams, SharePoint and Copilot offline or partially broken worldwide. In the United States, Downdetector logged as many as 130,000 Outlook fault reports at the peak. Japan's money market felt the disruption directly: one broker, whose order flow depends on Microsoft's email system, asked a domestic asset manager to stop placing orders while the outage ran. Investment trusts delayed publishing their net asset values. The government's shared IT system, overseen by the Digital Agency, also went down, Deputy Chief Cabinet Secretary Masanao Ozaki confirmed. Australia saw the same workaround, and RSM Australia partner Ashwin Pal put the cost to local businesses in the billions of Australian dollars.

Overloaded Servers, No Restoration Timetable

The root cause sits in server overload rather than an attack, going by the company's own status communications. Servers failed automated health checks, and a core process that establishes secure connections became overwhelmed, leaving affected machines unable to process requests. Infrastructure brought back online often failed again once traffic returned. In a 7:25 am UTC update on Friday, Microsoft said a patch had improved stability on some servers and that it was working to deploy the fix more widely, without committing to a restoration timetable. Its Japanese unit said it had found no facts pointing to a cyberattack, and a report covering the detailed cause and preventive measures is planned after full recovery. The blast radius is large by design: the company reported more than 450 million paid commercial seats for Microsoft 365 in an October filing, and Exchange Online hosts the mailboxes and calendars those organizations run on.

Courts, Ministries and Universities Hit

Public institutions in three countries reported email failures. The US Court of Appeals for the Tenth Circuit said its clerk's office could neither send nor receive email for much of Thursday. Japan's Finance Ministry said it was unable to receive email on Friday. Singapore's Government Technology Agency said government email was among the systems affected. In Japan the reach went beyond ministries: the Statistics Center of Japan said from 10:30 am that its email and website contact forms were unavailable, and Kyoto University reported outages across its mail service, Teams and OneDrive. Inside the suite itself, Microsoft Copilot responses, Defender management functions and data-loss-prevention features were listed as affected, and Teams users had trouble scheduling meetings and sending invitations, though most calls and chats kept working.

AI Connectors Fail Alongside Email

The outage reached into third-party AI tools through Microsoft 365 connectors, which let assistants read work email and calendars when users link their accounts. OpenAI reported errors in ChatGPT's Microsoft 365 connector. Anthropic said most email, calendar and search requests through Claude's connector failed during the incident. The episode follows a string of failures at central technology providers: in September, a separate outage took down ChatGPT, Claude and Grok, and in October 2025 a configuration fault on Azure produced a worldwide disruption, though Microsoft has confirmed no link between that fault and this one.

Single Point of Failure

Our reading: the episode matters less for Microsoft's own books than for the redundancy, or lack of it, in financial plumbing. The fax fallback worked because Japan still manufactures the machines, with Brother, Canon and Ricoh continuing to sell them, a legacy nobody designed as disaster recovery. The stock barely reacted: Microsoft shares closed slightly higher on Thursday despite the outage, then slipped in overnight trading, with the name already under pressure after overtaking Nvidia in market value earlier this week. A cooling failure at a CME data center last year made the same point about single points of failure freezing key systems. For crypto, the incident feeds the Web3 centralization debate rather than any direct price effect, since no crypto market mechanism was reported impaired.