Microsoft suffered one of the longest and most disruptive service outages in its history on Thursday, with Teams, Outlook, Word Online, and several other flagship products experiencing partial or total failure for approximately twenty-four hours. The BBC reported that the company issued a public apology after the disruption affected users across North America, Europe, Asia, and beyond, with the Guardian noting that the company was scrambling to restore services as the incident stretched from Wednesday evening through Thursday. Al Jazeera confirmed that Microsoft acknowledged services were “still returning” hours after the initial incident, a measured phrase that translated in practice to millions of workers unable to send email, attend virtual meetings, or access documents. The BBC further reported that the question of whether emails sent during the outage would be permanently lost remained, for a time, genuinely unanswered — a detail that concentrates the mind.
The received wisdom
The sympathetic reading of Microsoft’s predicament is that large-scale distributed computing systems are intrinsically complex, that even the most carefully engineered architecture will occasionally fail, and that the company’s rapid public communication and eventual resolution demonstrated responsible incident management. Technology journalists who cover the cloud industry will note that the “five nines” availability standard — 99.999% uptime, amounting to roughly five minutes of downtime per year — is an aspiration, not a guarantee, and that sophisticated enterprise customers build multi-cloud redundancy precisely because single-vendor dependency is a known risk.
The progressive policy frame will reach for a monopoly-and-regulation argument: Microsoft’s dominance in enterprise productivity software, cemented by the Microsoft 365 suite, means that no enterprise customer of any scale has a practical alternative when the platform fails, and that this concentration of dependency should attract regulatory attention. This argument also has genuine purchase. The European Commission has long viewed Microsoft’s bundling of Teams with the broader Microsoft 365 suite as an anticompetitive practice — an investigation that produced a formal charge in 2023. If the market were genuinely competitive, customers would have substitutes. That they do not is not simply a consequence of consumer choice; it is a consequence of market structure.
A different read
Both the technical-complexity defence and the antitrust remedy point to real features of the problem, but neither reaches the core of it. The deeper issue is that over the past decade, the world’s businesses, governments, schools, and hospitals have transferred the operational continuity of their institutions to a small number of private technology platforms — Microsoft, Google, Amazon — without meaningfully reckoning with the civilisational implications of that dependency. This is not a story about one company having a bad day. It is a story about how a civilisation organises its critical institutional functions.
Consider what was actually disrupted on Thursday. Hospitals whose clinical staff communicate via Teams. Courts whose e-filing systems run on Azure. Schools conducting examinations through Microsoft’s education platform. Military logistics operations that have adopted Microsoft 365 under government cloud contracts. None of these are hypothetical use cases; all of them are documented deployments of a platform that, for approximately twenty-four hours, could not reliably deliver email. The BBC’s investigation into the outage’s duration asked a simple but telling question: will emails be lost? The fact that this question could not be immediately answered tells you something important about where real institutional memory and operational continuity now reside.
The historical parallel is instructive. After the 1965 Northeast blackout, which left thirty million Americans without power for up to thirteen hours, the United States undertook a systematic review of grid interconnection and created the North American Electric Reliability Corporation to enforce reliability standards on a utility industry that had been allowed to build interdependencies without adequate safeguards. The lesson was not that electricity is dangerous and should be abandoned. The lesson was that when a technology becomes so embedded in civilisational function that its absence constitutes an emergency, its operators are no longer simply private companies serving customers — they are infrastructure providers with obligations that go beyond the terms of their service-level agreements.
The cloud computing industry has not been treated this way. The major cloud providers operate under contracts that explicitly disclaim liability for downtime beyond token service credits. Microsoft’s terms of service, like those of Google and Amazon, make clear that the company’s financial exposure for a twenty-four-hour outage of its enterprise communication platform is essentially nil. That arrangement was negotiated when these platforms were productivity conveniences. It has become untenable now that they are operational necessities.
The right answer is not nationalisation or heavy-handed government control of cloud services. It is a framework — perhaps modelled on the utility reliability standards that govern electricity, natural gas, and water — that imposes minimum resilience and redundancy requirements on platforms that exceed a threshold of systemic criticality. This would likely require international coordination, given that Microsoft’s infrastructure spans jurisdictions. It would certainly be complicated and fiercely resisted by an industry that has benefited enormously from regulatory arbitrage. But the alternative — continuing to treat the email and communications infrastructure of hospitals, courts, and defence establishments as a private product governed by click-wrap terms of service — is not a defensible public policy position any longer.
The government cloud contracts that Microsoft has accumulated are perhaps the most direct lever available. If the federal government and its allied governments conditioned large cloud contracts on demonstrable resilience standards, meaningful redundancy requirements, and financial liability provisions proportionate to the actual costs of failure, the industry would adapt. The procurement power of the state is, in this domain, the most credible regulatory instrument available — and it is currently being used without the leverage that the state’s dependency would entitle it to exercise.
What to watch
Watch whether the Microsoft outage triggers a serious congressional hearing on critical infrastructure reliability for cloud services — not the performative variety, but one with subpoena power and genuine interest in the contractual terms under which federal agencies have outsourced their operational continuity. Watch also for the European regulatory response, where the Commission’s existing Microsoft antitrust investigation may be an opportunity to impose resilience requirements alongside any structural remedy. And watch whether major corporate clients begin quietly building multi-cloud redundancy into their procurement requirements — which would be rational but which the economic incentives of the current market structure actively discourage.
— J