A multi-hour outage in the Microsoft 365 ecosystem left companies and government agencies around the world without email — and it is merely a rehearsal for a far more serious scenario. The incident exposed a systemic vulnerability in critical infrastructure that is now managed by automation without human oversight.
What happened to Microsoft services
No signs of a hacker attack were detected. Microsoft stated that there were no traces of a breach: the company operates an automated monitoring system that checks the health of servers and disables those that fail the check. This time, many servers simultaneously began failing the check. They were disabled, and the remaining ones could not withstand the increased load.
According to an industry analyst who writes about geopolitics and cybersecurity, the error may have been in the checking program itself: it began disabling healthy servers. As a result, the outage spread rapidly around the world and, by estimates, affected 450 million corporate workstations. Microsoft has not yet named the exact cause.
Services could not be restored for more than nine hours. Servers were restarted, but they shut down again as soon as users returned to them.
Scale of the consequences
Work came to a halt in many sectors, including:
- in payroll services;
- in government institutions.
At the same time, corporate AI services from OpenAI and Anthropic stopped working.
New challenges
Today, AI agents — programs that perform tasks on their own without human involvement — are being connected to banks, power grids, and supply chains. They run on servers belonging to the same three largest cloud companies.
The next error will be more costly. It will no longer be limited to delayed emails: payments could halt within seconds, hospitals could lose access to patient data, and the delivery of goods could stop.
Critical systems have been placed under the control of automation. At the same time, they lack safeguards designed to allow a human to intervene in time.
For investors, this is a signal that too much depends on a few cloud companies. An outage at one of them could also affect payment services. What happened should not be regarded as an ordinary technical glitch.
Cryptalist analysis: The incident demonstrates the concentrated risk of cloud infrastructure: the failure of automated control systems at a single provider can instantly spread to payments and critically important data. As AI agents are integrated into the financial sector and healthcare, such cascading failures are turning from an operational nuisance into a systemic risk for the entire economy. The market needs redundancy and manual control mechanisms, otherwise the next outage will be measured not in hours of email downtime but in real financial losses.