Knowledge Commons
HomeTopicsSeriesGuidesPopularAboutContact

There Was No Attack, There Was a Monopoly

The Microsoft 365 Outage, the "Global Cyber Crisis" Headline, and the Class Anatomy of the Cloud

19 min readAuthor: Oğuz Demirkapı
There Was No Attack, There Was a Monopoly

Yesterday evening, at 19:17 Turkey time, millions of people around the world lived through the same thing at the same moment: the mailbox would not open, the Teams meeting would not start, the calendar looked empty. Within a few hours the same headline fell onto screens: "GLOBAL CYBER CRISIS: Microsoft 365 failures reported in 74 countries, outage reports hit AWS and other giant platforms at the same time."

Part of the headline is true, part of it is false, and the most important part was never written. In this essay we will first summarise briefly what happened, then go into the detail: what happened, what did not, what the language of "cyber crisis" covers over, and what this outage tells us about whose hands hold the digital infrastructure.


First, the Summary

  1. What happened: On Thursday 12 November, at 16:17 UTC (19:17 Turkey time), Microsoft's email infrastructure, Exchange Online, collapsed. Outlook, the Teams calendar, SharePoint, Copilot, Defender and dozens of services chained to it stopped in turn. According to Downdetector the reports spread to 74 countries and regions in about an hour; for Outlook alone the peak was close to 140,000 reports. On the morning of 13 November, when this essay was written, the services had not yet fully returned; the outage had passed twelve hours.
  2. What did not happen: This was not a cyberattack. Microsoft executive Charles Lamanna explicitly ruled an attack out; the company said "there was no increase in traffic, a group of servers unexpectedly went down." The claim that AWS was "hit at the same time" is also largely an illusion: the reports for AWS (about 1,350) are very small beside Outlook's (over 100,000); AWS's own board showed only a local problem in Tokyo.
  3. What was not written: The real crisis is not "cyber", it is structural. A fault in a single component of one company pushed the Japanese government onto paper, stopped New Zealand's visa processing, sent investment bankers in Hong Kong onto WhatsApp, and drove a hospital in Toronto to a "Code Grey" alarm. This is the natural result of handing society's general communications infrastructure over to a handful of private monopolies.
  4. Who pays the price: Microsoft's contract promises 99.9% monthly availability. What the outage is worth is a "credit" of only 25–50% of the licence share of the affected service; and that is not automatic. It goes to whoever applies by 31 December, and is deducted from a future bill. The lost working day, the overtime, the delayed visa, the postponed surgery appointment sit outside the account.
  5. Our position: The fear of a "cyberattack" serves the security state and the security industry; it makes the real problem, the ownership of infrastructure, invisible. The answer is not more firewalls. It is a public, auditable, plural and common digital infrastructure.

Now let us go into the detail.


What Happened? A Collapse, Hour by Hour

We can follow the course of the outage from the incident records Microsoft published on its admin centre (EX1550543 and MO1550821) and from the technical press that tracked them. We give the times in Turkey time:

Time (Turkey)Event
12 November, 19:17The Exchange Online outage begins; users cannot open their mailboxes
Around 20:25Reports on Downdetector have spread to 74 countries and regions
Around 21:00Outlook reports peak: about 140,000
21:26A second incident record (MO1550821) is opened for dependent services
Around 22:06Microsoft stops all deployments and configuration changes
13 November, around 00:08Some servers are brought back by a different "health check" method
03:00The servers that came back stop responding again once they take traffic
06:07Microsoft says "the change is not working as expected"
07:22Partial recovery in some regions, but "not permanent"

Sources: LazyAdmin's incident roundup, Security Affairs, Downdetector's post on X.

Microsoft's own statements also changed along the way. First it was said that there was "high load in multiple regions." Then it was said that "the load does not come from an increase in traffic, but from some servers unexpectedly going down." Then it was explained that a process establishing secure connections on the servers had become overloaded, and that the servers could no longer handle requests. Through the night a recent network configuration change began to be rolled back; but the company later said it had not found a change that coincided with the start of the outage. So the root cause is not yet known. The definite answer will be in the post-incident report Microsoft publishes.

We note this uncertainty, because honesty requires it: every piece of information we have is the company's own account of the fault it lived through. There is no independent audit. That is itself a class fact; we will come back to it below.


Who Was Affected? One Company's Fault, a Planet's Paralysis

The list of the impact shows how deeply the digital monopoly is woven into the social fabric (Streamline Feed roundup, Wire):

  • States: Japan's Digital Minister ordered a switch to an emergency text-message backup, and officials printed documents onto paper. New Zealand's Ministry of Foreign Affairs announced that embassies would not be able to receive email, and Immigration New Zealand stopped visa processing. The Supreme Court of the Philippines reported an email disruption. Government systems in Singapore, intelligence and federal police email in Australia, and the US House of Representatives were affected. In the United Kingdom, units such as South Wales Police directed the public to the telephone for non-emergency calls.
  • Finance: In Hong Kong, bankers at Morgan Stanley and Goldman Sachs could not reach email or Teams. In Singapore, CMC Markets dealers moved to the telephone and WhatsApp. A publicly listed company (Workhorse Group) postponed its quarterly earnings call.
  • Health and transport: Unity Health in Toronto issued a "Code Grey" email-outage alarm. Air India and IndiGo redirected their customer-service email to other channels.
  • Security: Defender's Safe Links feature could no longer verify links. The "security" service was out of action at the moment it was needed most.
  • Artificial intelligence: Copilot could not reach users' mailboxes; OpenAI and Anthropic also reported problems in their Microsoft 365 connectors. In the world where we are told "agents will do our work", the agents' hands and feet are tied to the same monopoly.

In Turkey too, since the pandemic, many universities, schools, public institutions and private companies have built their communications on Microsoft 365 and Teams. Last night's collapse hit us directly as well: the white-collar worker on the evening shift, the programmer working remotely, the student handing in homework.


Why We Reject the "Global Cyber Crisis" Headline

The headline makes two claims: that this was a "cyber crisis", and that AWS and "the other giant platforms" were hit at the same time. Both are misleading.

There was no cyberattack. The picture Microsoft relayed is not the picture of an attack. A denial-of-service attack appears as extra traffic; here the traffic did not increase, capacity was lost. The company says it found no trace of malicious activity (Ransom News analysis). Caution is required until the root cause is explained. But the data in hand does not support a "cyber war" scenario.

AWS did not collapse; people complained from somewhere else. Downdetector is a site of user reports; it counts human complaints, not a technical measurement. When Outlook would not open, people also reported Gmail, AWS and Slack to ask "is it working?" Indeed about 1,350 reports came in for AWS and about 1,000 for Gmail; more than 100,000 for Outlook. AWS's health board showed only a local problem in Tokyo, and Google's cloud board reported no problem. In the first hours, publications such as Macworld wrote that problems were also seen in services such as Slack, Reddit and Cloudflare; but there is no verified information tying these to a common cause.

Why does this matter? Because language produces politics. When one says "cyber crisis", what comes to mind is an external enemy, an attacker, a war. This frame has three functions:

  1. It moves the responsible party outside. Once the fault is loaded onto a ghost attacker, the company that gathered the infrastructure into one pair of hands and then suffered the outage takes the place of the "victim".
  2. It grows the security market. Every "cyber crisis" story means more security products, more surveillance, more budget. We discussed who this benefits most in The Panopticon's Doorkeeper.
  3. It offers the security state a pretext. As in Turkey, the discourse of "cyber threat" often prepares the ground for internet restrictions, access blocks and new legal powers.

Yet there is no attacker. What there is, is a monopoly that runs society's communications arteries according to its own profit account, and that monopoly's internal fault. The crisis is not cyber. It is a crisis of ownership.


There Is No Such Thing as the Cloud, There Is Someone Else's Computer

There is an old joke among computing workers: "There is no such thing as the cloud, there is only someone else's computer." This joke is the shortest analysis one can make of the digital form of what Marx called commodity fetishism.

The "cloud" metaphor evaporates the infrastructure. It presents it as something light, everywhere, belonging to no one. The cloud, though, is concrete buildings, cooling systems and server racks, in particular countries, belonging to particular companies, tied to particular power grids, kept standing on the night shift by particular workers. Last night, the one "restarting the servers", "changing the health-check method", "scaling the virtual machines" was not Microsoft. It was engineers whose names we will never know. The outage made visible, for a moment, what the cloud actually is.

It is known that the three big cloud companies (Amazon, Microsoft, Google) hold close to two thirds of the global cloud infrastructure market. In corporate email and office software, Microsoft's user base of more than 430 million has in effect become a standard. We discussed this picture through Microsoft's internal labour processes in Forty Percent of the Code, a Quarter of the Staff, and through the ownership of communication in Whose Is the Mailbox?. What happened last night is the field evidence of those arguments.


When the General Conditions of Production Become Private Property

In the Grundrisse, Marx names infrastructures such as roads, canals and means of communication, which are the product of no single capitalist and yet which all of them need, "the general conditions of production". These conditions are social: the whole of society uses them, the whole of society depends on them. In the nineteenth century the railway was such a condition, and in the twentieth the telephone network. In the twenty-first, cloud and communications software have taken that place.

There is a basic contradiction here:

  • Use is social: A government, a hospital, a school, a union, an investment bank use the same infrastructure.
  • Ownership is private: A single company's board decides how the infrastructure will be built, where it will stand, when it will be updated, and who will be given priority in a fault.
  • Risk is social, gain is private: The cost of the fault spreads to everyone; the subscription revenue collects in a single balance sheet.

This contradiction is not solved by "better engineering". Microsoft's engineers are very likely among the most capable in the world. The problem is not capability. It is that the single point of failure is produced by the relation of production itself. The law of competition drives companies to concentrate for scale, to shared components for cost, and to constant change for speed. This is Microsoft's second large "capacity loss" event of 2026 (the first was in North America in January, The Register). In the last two years we also lived through July 2024, when a CrowdStrike update stopped the world, and October 2025, when AWS's US-East region collapsed and Azure Front Door failed. These are not exceptions. They are a pattern.


The Class Ledger of an Outage

Let us look at who paid the price of the outage. When we set the two sides side by side, the picture clears:

The Monopoly's ColumnThe Column of Labour and Society
A promise of 99.9% monthly availability (a right to about 43 minutes of outage a month)An outage of more than 12 hours, a lost working day
In return for the outage, a credit of only 25–50% of the licence share of the affected serviceOvertime, night duty, postponed deliveries, a cut performance score
The credit is not automatic; it goes to whoever applies by 31 December, deducted from a future billA delayed visa, a postponed appointment, a police force and a court that cannot be reached
The root cause will be explained in the company's own report, on the company's own calendarNo independent audit, no duty to account to the public
Subscription revenue continues to flow without interruptionThe burden of moving to WhatsApp, to a personal phone, to paper falls on the worker's back
Reputation is protected by the statement "there was no attack"New security products and budgets are sold on the fear of a "cyber crisis"

Source: LazyAdmin for the credits in Microsoft's service-level agreement.

Look at the account: when the company breaks its own contract, the price it pays is a small discount on the next bill. Behind that bill, though, is the labour of millions of people at work. The time spent by the worker who carries the meeting on over WhatsApp when Teams will not open, who calls the customer by telephone when the email will not go, who checks at midnight "has the system come back", is in no SLA table. The monopoly's fault is made good by the worker's unpaid overtime.

There are unseen workers too: the site-reliability engineers who chased the outage through the night, the call-centre workers on the support line, the IT staff in institutions trying to answer the angry user. They are the ones who "solve" the outage; but it is the company that tells the story, makes the statement and protects the reputation.


In the Age of Artificial Intelligence the Single Point of Failure Grows

This outage has a new side: the artificial-intelligence layer. Microsoft 365 Copilot, and the Microsoft connectors of OpenAI and Anthropic, did not work last night. Because these tools do not work on their own. They work by reaching the same mailboxes, the same calendars, the same files.

The technology companies tell us "agents will do your work". What happens when everything the agent will reach sits on a single company's server? The answer was seen last night: dependence compounds. On one side the cloud monopoly, on the other the model monopoly; as the two bind to each other, the reach of a single point of failure grows. We took up this relation from the side of capital in Bubble or Debt?, and from the side of labour in The Human in the Loop. Last night "the human in the loop" was on duty again: the human who, when the system collapsed, carried the work by telephone, by paper, by hand.


A Separate Question for Turkey: Where Is Our Data, Whose Is the Key?

In Turkey there is in fact a framework for this in the public sector. The 2019 Circular on Information and Communication Security Measures foresees that critical public data be kept inside the country, and that recourse to foreign cloud services be limited. The KVKK also binds the transfer of data abroad to a rule. In practice, though, from schools to universities, from municipalities to the private sector, a large part of communication flows through the clouds of foreign monopolies.

Here we have to stay clear of two wrong roads:

  • The nationalist answer: The discourse of a "domestic and national cloud" carries the problem from ownership to the flag. Putting a domestic monopoly, or a company close to power, in place of the foreign monopoly does not end the dependence. It only changes hands. And this time the surveillance comes from closer in. We told of this danger through digital identity in The Wallet Is Ours, Whose Is the Key?.
  • The liberal answer: The advice to "use more than one vendor" reduces the problem to a corporate purchasing preference. Moving from one company to three is walking around inside the oligopoly.

Our question is different: Whose hands hold this infrastructure, under whose control, for whose benefit? As long as the answer to these three questions does not change, it does not matter under which country's flag the outage is lived.


What, Then, Is to Be Done? A Public, Plural, Common Infrastructure

The lesson to be drawn from the outage is not "Microsoft is bad, let us move to another company." The lesson is that the general conditions of production of society must be taken under social control. There are concrete layers of this that can be spoken of today:

  • Open protocols and interoperability: Email is already an open protocol; SMTP and IMAP belong to no company. The problem is not in the protocol. It is in the closed ecosystems built on top of the protocol. Open standards should be made obligatory, and institutions should be kept from being locked to a single supplier.
  • Public and federated services: Universities, municipalities, professional chambers and unions can set up communications services based on free software, running on their own servers and bound to one another in a federated way. Many public institutions in Europe are taking steps in this direction; in Turkey too, the universities have the experience of a public network such as ULAKBİM.
  • Independent audit and accounting to the public: The outages of digital services that perform a public function should be examined not by the company's own blog post, but by independent, public incident reports. The NIS2 and DORA regulations in Europe point to the risk of supplier concentration; but regulation stays limited so long as it is left to capital auditing itself.
  • The organising of the computing worker: We are the ones who build this infrastructure, who run it, and who get it back on its feet in an outage. A say in how the infrastructure will be built passes through the organisation of the labour that keeps it standing. The Computing Worker's Handbook was written for this reason.

These are not built in one night. But what an outage reminds us of is this: if one company's internal fault can stop a planet, that company is no longer a private company. It is a social infrastructure taken under private ownership. And what is social must be taken under social control. We set this down as a principle in A Common Product Is Not Common Without Common Ownership; last night we saw the practice.


Concrete Tasks

For computing workers:

  • Record the overtime and the night work done through the outage; ask for the return on this labour. The monopoly's fault should not be your unpaid overtime.
  • Raise, in the institution where you work, the dependence on a "single supplier" and the outage plan. A backup communications channel that is "everyone's personal WhatsApp" is not a plan. It is a transfer of the burden.
  • Make small trials at your workplace of free-software alternatives (federated email, Matrix, Nextcloud, Jitsi and the like), and share the experience.

For unions, professional chambers and democratic organisations:

  • Do not entrust your own communications infrastructure to a monopoly's free package. Ask whose server the membership list, the correspondence and the archive sit on. You can read Whose Hands Hold the Membership List? again with this in mind.
  • Take the demand for a public and federated communications infrastructure into your programme; open for discussion the idea of a shared infrastructure with universities, municipalities and chambers.

For everyone:

  • When you meet headlines of "cyber crisis", "digital attack", "invisible enemy", stop and ask: Who is the attacker? What is the evidence? Whose interest does this fear serve?
  • When a service collapses, make it a habit to ask "whose hands, whose control, whose benefit".

Dear Young Comrades

Last night millions of people looked at the same empty screen at the same moment. That empty screen told us something: the things we use every day, and take to be as natural as a cloud in the air, are in fact someone's property. When the owner of that property makes a mistake, states go onto paper, bankers onto the telephone, hospitals onto the alarm. But the owner of the property does not change, the bill does not change, the power does not change.

If we read this outage as a "cyber crisis", we learn nothing but fear. If we read it as a crisis of ownership, we learn where we have to look. The machine is new, the chain is old. But it is labour that forges the links of the chain, and labour that keeps the machine standing. We build the infrastructure, we run it, we get it back on its feet. We will also learn to govern it together, and for everyone.

Knowledge belongs to everyone. Infrastructure must belong to everyone too.


Sources

Related essays from Knowledge Commons: Whose Is the Mailbox? · Forty Percent of the Code, a Quarter of the Staff · Bubble or Debt? · The Human in the Loop · The Wallet Is Ours, Whose Is the Key? · The Panopticon's Doorkeeper · A Common Product Is Not Common Without Common Ownership · The Computing Worker's Handbook · Organising 2.0


This essay is shared under the CC BY-SA 4.0 licence. It is a living document; it will be updated when Microsoft publishes its post-incident report.

Related Posts

Newsletter