Last Updated on July 9, 2026 by Karl Thompson
We no longer primarily own the computers that power our lives—we increasingly rent computing instead. This chapter explores how that transformation happened, why it matters, and what it reveals about power and hidden infrastructure in the twenty-first century.
What is Cloud Computing?
When most of us think about computers, we still imagine the machine sitting on our desk or in our pocket. Our laptop stores our documents, our phone stores our photographs, and our tablet runs our apps. At least, that’s how it appears.
But the reality is very different.
Check your email over breakfast and the messages aren’t sitting on your phone. Open Microsoft Teams at work and the meeting isn’t being hosted by your laptop. Stream a film on Netflix and it isn’t being played from your television. Ask ChatGPT a question and the answer isn’t being generated by your device. Even asking your smart speaker for tomorrow’s weather sends your request hundreds or even thousands of miles away before an answer returns a fraction of a second later.
Increasingly, our devices have become little more than windows onto vast networks of remote computers. The real computing power—the storage, processing and software—exists somewhere else.
Over the past two decades we’ve undergone one of the biggest technological shifts since the invention of the personal computer. We no longer primarily own computing power—we increasingly rent it. Every time we log into Microsoft 365, stream a film, save photographs to the cloud or use an AI chatbot, we are borrowing computing resources owned by somebody else. Computing has quietly become a utility, consumed on demand much like electricity or running water.
That “somebody else” is what we call the cloud.
The name itself is misleading. There is nothing ethereal about cloud computing. There is no giant digital cloud floating invisibly above our heads. Every photograph uploaded to Google Photos, every online bank transfer, every streamed television programme and every AI conversation ultimately depends on physical computers housed inside enormous data centres, linked together by thousands of kilometres of fibre-optic cable and powered by vast quantities of electricity. The cloud is not an alternative to physical infrastructure; it is simply a different way of organising it. The National Institute of Standards and Technology (NIST) defines cloud computing as on-demand access to shared computing resources delivered over a network, allowing users to access computing capacity whenever they need it rather than owning the infrastructure themselves.
This represents a remarkable change in how society uses technology. Twenty years ago, most organisations bought their own servers, installed software on individual computers and employed teams of technicians to keep everything running. Expanding meant purchasing more hardware, finding more office space and investing large sums of money before any benefits were realised.
Today, organisations increasingly rent computing capacity from a handful of global cloud providers. Need twice as much computing power? It can be available within minutes. Need less next month? Simply stop paying for it. The cloud has transformed computing from a product that organisations owned into a service they consume when required.
This transformation is now so commonplace that most people rarely think about it. Cloud computing has become one of the hidden infrastructures of everyday life—largely invisible precisely because it usually works so well. As sociologist Susan Leigh Star argued, infrastructure tends to disappear into the background of everyday life, only becoming visible when it breaks down. Few people think about cloud computing until an airline’s booking system crashes, online banking stops working, or a major cloud outage takes thousands of websites offline at once.
Cloud computing therefore illustrates one of the central themes of this book. The technologies we rely upon every day increasingly exist beyond our homes, workplaces and even our countries. Instead, modern life depends upon immense shared computing infrastructures owned by a remarkably small number of corporations. Understanding why society has moved from owning computers to renting computing—and what that means for power, dependence and control—is one of the defining sociological questions of the twenty-first century.
The World’s Largest Computer
Cloud computing has grown so rapidly that it is now difficult to imagine modern society functioning without it. What began as a way for technology companies to rent spare computing capacity has evolved into one of the world’s most important infrastructures, underpinning everything from online banking and supermarket logistics to government services, healthcare records and artificial intelligence.
The scale of this transformation is extraordinary. According to research company Gartner, organisations worldwide are expected to spend around US$723 billion on public cloud services in 2025, with spending continuing to rise as businesses migrate more of their operations online and invest heavily in artificial intelligence. Cloud computing has shifted from being a niche technology used by internet companies to becoming the default way that organisations buy computing power. Today, many new businesses never purchase their own servers at all, instead building their entire operations on rented cloud infrastructure (Gartner, Forecasts Worldwide Public Cloud End-User Spending to Total $723 Billion in 2025: ).
This vast market is dominated by surprisingly few companies. Industry analysts at Synergy Research Group estimate that three firms—Amazon Web Services (AWS), Microsoft Azure and Google Cloud—collectively account for around two-thirds of the global cloud infrastructure market. Although hundreds of companies provide specialist cloud services, most ultimately rely upon the same handful of hyperscale providers whose data centres span dozens of countries and millions of servers (Synergy Research Group, Cloud Market Share Trends).
Behind these figures lies an even larger physical infrastructure. Each of the major cloud providers operates networks of enormous data centres organised into Regions—geographical clusters of facilities—and Availability Zones, which are designed so that if one data centre experiences a failure, another can immediately take over. Amazon Web Services alone operates more than one hundred Availability Zones across dozens of Regions worldwide, with further expansion continuing as demand grows (AWS Global Infrastructure).
For most users, this infrastructure is completely invisible. Opening a spreadsheet stored in Microsoft 365, uploading photographs to Google Photos or asking an AI assistant a question takes only seconds. Yet each of these simple actions may involve computers spread across multiple buildings, cities or even countries, communicating continuously to deliver what appears to be an effortless experience.
This immense scale helps explain why cloud computing has become such a powerful hidden infrastructure. Few organisations could ever afford to build systems of this size for themselves. Instead, millions of businesses, governments and individuals now share the same underlying computing platforms, renting tiny fractions of infrastructures so large that they have become global utilities in their own right.
Cloud computing has therefore created one of the defining paradoxes of the digital age. Computing has become more personal than ever—we each carry powerful devices in our pockets—yet the computing that makes those devices useful has become increasingly centralised. Behind billions of individual screens lies a remarkably small number of shared infrastructures serving much of the world’s digital economy.
Case Study — Amazon Web Services: Renting the World’s Computers
If cloud computing has a birthplace, it is Amazon Web Services (AWS).
That may seem surprising. After all, Amazon is best known as an online retailer rather than a technology company. Yet today AWS has become the world’s largest cloud provider, generating tens of billions of dollars in annual revenue and supplying computing infrastructure to millions of customers, from start-ups and universities to banks, hospitals and governments.
The origins of AWS lie in a problem Amazon created for itself. During the early 2000s, the company was expanding rapidly and repeatedly faced the same challenge. Every time demand increased, engineers had to buy more servers, install more software and build more computing capacity. Much of this expensive equipment then sat idle outside busy shopping periods such as Christmas or Black Friday.
Amazon realised that computing power did not have to be tied to a single organisation. If its own servers were underused for much of the year, why not rent that spare capacity to other businesses?
In 2006, AWS launched a service called Elastic Compute Cloud (EC2), allowing customers to rent virtual computers over the internet instead of buying physical machines. Businesses could increase or reduce their computing capacity within minutes, paying only for what they actually used. It represented a radical departure from the traditional model of purchasing hardware years in advance.
The idea proved revolutionary. Instead of investing millions of pounds in computer rooms, organisations could treat computing as an operational expense rather than a capital investment. A small start-up suddenly had access to the same type of infrastructure as a multinational corporation. If demand doubled overnight, more computing power could be added almost instantly. If demand fell, it could simply be switched off.
Today, AWS operates one of the largest computing infrastructures ever built. Its global network is organised into Regions, each serving a particular part of the world, and within each Region are multiple Availability Zones—physically separate data centres connected by high-speed fibre-optic networks. This arrangement means that if one facility experiences a power failure, fire or other disruption, workloads can often continue running from another location with little or no interruption. According to AWS, it now operates more than one hundred Availability Zones across dozens of geographic Regions worldwide, with new Regions continually under construction (AWS, Global Infrastructure.)
Although AWS pioneered this model, it is no longer alone. Microsoft Azure and Google Cloud have developed similar global infrastructures, while companies such as Oracle and IBM serve particular sectors and specialist markets. Together, however, the three largest providers now account for around two-thirds of the world’s cloud infrastructure market, giving them enormous influence over the digital economy (Synergy Research Group, Cloud Market Share Trends.)
For most people, none of this is visible. A video conference connects in seconds. A banking app opens almost instantly. A retailer processes thousands of online orders every minute. Behind these ordinary experiences lies a remarkable reality: millions of organisations are no longer running their own computers. Instead, they are renting tiny fractions of one of the largest shared computing systems ever constructed.
Amazon did not simply create another successful technology business. It helped redefine what a computer is. Increasingly, a computer is no longer a machine that sits on your desk. It is a service that can be summoned from almost anywhere in the world, used for as long as it is needed, and then quietly disappears again until the next request.
How Cloud Computing Works – Borrowing Someone Else’s Computer
If cloud computing is simply renting somebody else’s computer, how does that actually work?
Imagine you’ve decided to start a new business twenty years ago. You’d need to buy a server, find somewhere to put it, install software, protect it from hackers, back it up regularly and hope you bought enough computing power for future growth. If your business suddenly became successful, you’d have to buy another server. If it failed, you’d be left with expensive equipment gathering dust.
Cloud computing turns this model on its head.
Instead of buying a computer, you simply rent as much of one as you need.
Of course, you’re not literally renting an entire computer. The breakthrough that made cloud computing possible was virtualisation. Rather than dedicating one physical server to one customer, special software divides a powerful machine into many virtual computers, each behaving as though it were a completely separate machine. One physical server might therefore be serving dozens, or even hundreds, of different organisations simultaneously, each completely isolated from the others.
To the customer, it feels as though they have their own computer. In reality, they are sharing one with many others without ever noticing.
This approach makes cloud computing remarkably efficient. If one customer only needs extra computing power for an hour, they can borrow it and then release it back into the pool for someone else to use. Instead of thousands of organisations each owning underused servers sitting quietly in office cupboards, cloud providers can keep their equipment working much closer to full capacity. Renting computing is therefore not simply cheaper; it also makes much better use of the world’s computing resources.
Cloud providers offer these resources in different ways depending on how much control customers want.
At the simplest level is Software as a Service (SaaS). This is the version most people use every day. Applications such as Microsoft 365, Gmail, Dropbox and Salesforce all run on remote computers. Users simply open a web browser or app and use the software without worrying about where it is running.
A step further is Platform as a Service (PaaS). Here, software developers rent an environment in which they can build and deploy their own applications without having to manage the underlying computers themselves.
Finally, there is Infrastructure as a Service (IaaS). This is the closest equivalent to renting a traditional computer. Customers lease virtual machines, storage and networking while deciding exactly what software they want to install. Amazon’s Elastic Compute Cloud (EC2) was one of the first major examples of this approach.
The National Institute of Standards and Technology (NIST) identifies five characteristics that distinguish cloud computing from traditional IT. Computing resources are available on demand, they can be accessed over a network from almost anywhere, they are drawn from large shared pools of hardware, they can expand or shrink rapidly as demand changes, and customers pay only for the resources they actually use rather than purchasing permanent capacity (NIST, The NIST Definition of Cloud Computing)
Behind the scenes, the process is astonishingly complex. Every time you save a document, stream a film or ask an AI assistant a question, your request is routed through the internet to one of the provider’s data centres. Software decides which servers have spare capacity, allocates the necessary computing power, stores the information across multiple locations and returns the result to your device—often in well under a second. Most users never notice this orchestration because it happens automatically and at extraordinary speed.
This is perhaps the greatest achievement of cloud computing. It has transformed one of the most technically complex infrastructures ever created into something that feels almost effortless. We no longer think about where our software runs or where our files are stored. We simply expect them to be there.
And that is precisely what makes cloud computing such an important hidden infrastructure. Like electricity flowing from a socket or clean water arriving through a tap, it has become a utility that millions of people use every day without ever seeing the vast systems that make it possible.
Material Foundations — The Cloud Is Built on Steel, Silicon and Electricity
The word cloud makes modern computing sound almost weightless. In reality, cloud computing is one of the most resource-intensive infrastructures ever created.
Every document stored online, every streamed film and every AI prompt ultimately depends upon physical materials extracted from the earth, manufactured into sophisticated hardware and powered continuously by vast quantities of electricity.
Cloud computing begins with servers. These highly specialised computers fill enormous data centres, each containing thousands or even tens of thousands of machines working simultaneously. Rather than serving a single organisation, they are shared between many customers, allowing millions of people to rent tiny fractions of the same physical infrastructure.
Those servers depend upon some of the most advanced semiconductor chips ever manufactured. Modern cloud platforms increasingly rely on specialised graphics processing units (GPUs) to train and operate artificial intelligence systems, creating unprecedented demand for cutting-edge chips. According to NVIDIA, today’s AI data centres require tightly integrated systems combining GPUs, high-speed networking and advanced cooling technologies to deliver the enormous computational power demanded by modern AI applications.
Keeping these facilities running requires extraordinary amounts of energy.
Unlike offices, cloud data centres never close. They operate twenty-four hours a day, every day of the year. Electricity therefore becomes as important as computing itself. The International Energy Agency (IEA) estimates that data centres already account for a significant and rapidly growing share of global electricity demand, with artificial intelligence expected to accelerate that growth substantially over the coming decade.
Energy is only part of the challenge. Computers generate heat, and facilities containing tens of thousands of servers generate an enormous amount of it. Operators increasingly use liquid cooling systems, carefully engineered airflow and, in some locations, large quantities of water to prevent equipment from overheating. Designing efficient cooling systems has become one of the defining engineering challenges of the cloud era.
Cloud computing also depends upon the infrastructures explored earlier in this chapter. Every cloud platform requires data centres to house its equipment, fibre-optic networks to move information between facilities and reliable electricity grids capable of supplying uninterrupted power. Remove any one of these foundations and the cloud simply ceases to exist.
The physical demands of cloud computing help explain why building a global cloud platform is so expensive. It requires billions of pounds of investment in land, buildings, power infrastructure, networking equipment and advanced semiconductors before a single customer can upload a file or launch a website.
The cloud therefore represents an important paradox. It is often described using the language of weightlessness and virtuality, yet it depends upon one of the largest concentrations of physical infrastructure ever assembled. The digital economy has not escaped the material world; it has become more dependent upon it than ever before.
6. Ownership and Control — Who Owns the Cloud?
One of the defining features of cloud computing is that millions of organisations appear to have their own computing infrastructure, when in reality much of it is owned by remarkably few companies.
A local authority storing planning applications, a hospital managing patient records, a university delivering online teaching and a supermarket processing online orders may all appear to operate completely independent computer systems. Behind the scenes, however, many are renting computing capacity from exactly the same cloud provider.
This represents one of the greatest concentrations of infrastructure ownership in modern history.
Three companies—Amazon Web Services (AWS), Microsoft Azure and Google Cloud—dominate the global cloud market. Together they provide the digital infrastructure upon which governments, businesses and public services increasingly depend. Other firms, including Oracle, IBM and Alibaba Cloud, play important roles, but the overwhelming majority of cloud computing is concentrated within this small group of hyperscale providers.
Their dominance is no accident.
Building a global cloud platform requires extraordinary investment. Providers must acquire land, construct data centres, purchase millions of servers, build high-speed fibre connections, secure reliable electricity supplies and continually replace hardware as technology advances. The costs run into tens of billions of pounds every year, creating enormous barriers for potential competitors.
Size also creates further advantages.
Large cloud providers can spread their costs across millions of customers, allowing them to offer lower prices while investing continuously in new technologies such as artificial intelligence, cybersecurity and advanced networking. The larger they become, the more attractive they become to new customers—a cycle that reinforces their market position.
For customers, moving away is often much harder than moving in.
Many organisations redesign their software specifically for one cloud provider, integrate dozens of specialist services and store vast quantities of data within proprietary systems. Switching providers may therefore require rebuilding applications, transferring enormous datasets and retraining staff, making migration both expensive and risky. Economists describe these as switching costs, while businesses often refer to the problem as vendor lock-in.
These concerns have attracted increasing attention from competition authorities.
In 2025, the Competition and Markets Authority (CMA) launched a full market investigation into UK cloud services following concerns that technical and commercial practices may be making it unnecessarily difficult for customers to switch providers. The investigation is examining issues including software licensing, data transfer charges and barriers to effective competition.
Similar concerns have emerged internationally. The European Commission has encouraged the development of common cloud standards and data portability initiatives to reduce dependence on individual providers and make it easier for organisations to move between competing cloud platforms. These efforts reflect growing recognition that cloud computing has become part of Europe’s critical digital infrastructure.
From a sociological perspective, ownership matters because infrastructure creates power.
Railways shaped the Industrial Revolution because those who controlled transport influenced trade and economic development. Electricity companies transformed twentieth-century society because they controlled access to power. Today, cloud providers increasingly occupy a similar position within the digital economy. They do not simply provide computing services; they own the infrastructure upon which thousands of other organisations now depend.
Sociologist Nick Srnicek argues that many of today’s technology giants have evolved into platforms—businesses that derive their influence by controlling the infrastructure used by others rather than by producing goods themselves. Cloud computing is perhaps the clearest example of this model. It has enabled Amazon, Microsoft and Google to become not merely software companies but providers of essential digital infrastructure.
This represents a profound shift in the organisation of modern society.
Businesses may believe they own their digital operations, yet increasingly they rent the infrastructure that makes those operations possible. Ownership has become concentrated even as access has become widespread.
Cloud computing has therefore changed more than the economics of information technology.
It has reshaped where power resides in the digital age.
Geography — The Cloud Has an Address
Cloud computing is often described as though it exists everywhere at once. In reality, it is concentrated in a surprisingly small number of places.
Every cloud service has a physical location. Every photograph uploaded to the cloud, every online payment and every AI prompt is processed inside a real data centre, connected to real fibre-optic networks and supplied by a real electricity grid. The cloud may feel borderless, but it has a postcode.
The world’s largest concentration of cloud infrastructure is Northern Virginia in the United States, an area widely known as Data Center Alley. Home to hundreds of commercial data centres, it hosts cloud infrastructure operated by Amazon Web Services, Microsoft Azure, Google Cloud and many other providers. The region developed because it combined excellent fibre connectivity, abundant electricity, relatively inexpensive land and close proximity to Washington, D.C., creating the ideal conditions for hyperscale computing.
Europe’s leading cloud hub is Frankfurt. The city sits at the heart of the continent’s digital infrastructure thanks to DE-CIX, one of the world’s largest internet exchanges. Its central location, excellent connectivity and role as a major financial centre have made Frankfurt a natural location for cloud providers serving European customers.
In the United Kingdom, cloud infrastructure is concentrated around London and Slough. London’s role as a global financial centre creates enormous demand for cloud services, while Slough has developed into Britain’s largest cluster of commercial data centres. Together they form the UK’s principal cloud region, serving businesses, government departments and public services across the country.
Other major cloud hubs include Dublin, Amsterdam, Paris, Singapore, Tokyo and Sydney. Although they are separated by thousands of kilometres, these locations share remarkably similar characteristics. They all possess excellent international connectivity, reliable electricity supplies, highly skilled workforces and favourable investment conditions. Geography still matters—it has simply become digital geography.
Cloud providers carefully select these locations for another reason: latency. Although information travels through fibre-optic cables at extraordinary speeds, it is not instantaneous. The further data must travel, the longer the delay. Locating cloud infrastructure close to major cities, businesses and internet exchanges reduces these delays, ensuring that websites load quickly, video calls remain smooth and financial transactions are processed almost instantly.
Geography also determines resilience.
Rather than concentrating all their infrastructure in one place, providers divide it into Regions, each containing multiple Availability Zones located several kilometres apart. If a flood, fire or power failure affects one data centre, another can often continue operating with little interruption. Physical distance, once viewed as a disadvantage, has become an essential part of cloud resilience.
Increasingly, geography is also becoming political.
Governments are paying closer attention to where data is stored and which legal systems govern it. Concerns over cybersecurity, national security and digital sovereignty have encouraged many countries to require certain categories of information to remain within national borders. This has accelerated investment in regional cloud infrastructure and so-called sovereign clouds, designed specifically to meet national regulatory requirements.
Research by TeleGeography, which tracks the world’s internet infrastructure, shows that hyperscale cloud providers continue to expand their networks of regions and availability zones to meet growing demand while reducing latency for users around the world.
Similarly, Cushman & Wakefield’s Global Data Center Market Comparison demonstrates that investment in cloud infrastructure is increasingly concentrated in cities able to provide plentiful electricity, fibre connectivity, suitable land and supportive planning policies. These factors—not chance—have created today’s geography of the cloud.
Cloud computing therefore challenges one of the most persistent myths of the digital age.
The internet did not abolish geography.
Instead, it created a new geography—one in which a relatively small number of strategically important places have become indispensable to the operation of modern society. Behind the apparent weightlessness of the cloud lies a network of real cities, real infrastructure and real centres of power.
8. Historical Development — How Computing Came Full Circle
Cloud computing may feel like a modern invention, but its origins stretch back more than half a century. In many ways, today’s cloud represents a return to computing’s earliest model—shared resources accessed by many users rather than individual machines owned by individuals.
During the 1950s and 1960s, computers were vast, room-sized mainframes owned by governments, universities and large corporations. They were so expensive that very few organisations could justify purchasing their own. Instead, dozens or even hundreds of users shared a single computer, connecting through simple terminals. Computing was a scarce, shared resource rather than something individuals possessed. IBM’s history of the mainframe charts how these powerful machines became the backbone of business, government and scientific research during the twentieth century.
Everything changed with the arrival of the personal computer. During the late 1970s and 1980s, companies such as Apple and IBM, together with Microsoft’s software, brought computing onto individual desks. Businesses bought their own servers, employees stored files on local hard drives and organisations built dedicated computer rooms to support their operations. Computing became personal—and, most importantly, it became something people owned.
The rapid growth of the internet during the 1990s connected these individual computers together, allowing information to move almost instantly around the world. Yet most organisations still maintained their own IT infrastructure. Every new employee required another computer, more storage and additional software licences. Expanding often meant buying more servers months or even years before they were actually needed.
This model proved both expensive and inefficient.
Most servers spent much of their lives doing very little. Businesses had to purchase enough computing capacity to cope with their busiest days of the year, even if much of that hardware sat idle for the remaining eleven months. Companies carried all the costs of ownership—buying equipment, maintaining it, cooling it and eventually replacing it.
The breakthrough came with virtualisation.
Engineers discovered that a single powerful physical server could be divided into multiple virtual machines, allowing many different organisations to share the same hardware securely. Instead of one server serving one business, one server could support dozens of customers simultaneously, dramatically increasing efficiency while reducing costs. This technological innovation laid the foundations for modern cloud computing.
The decisive turning point came in 2006, when Amazon Web Services (AWS) launched Elastic Compute Cloud (EC2). For the first time, businesses could rent computing power over the internet whenever they needed it, paying only for the resources they actually used rather than investing in expensive hardware of their own. AWS describes EC2 as the beginning of a new model in which computing became available on demand rather than through ownership.
Microsoft, Google and other technology companies rapidly followed. Over the following two decades, cloud computing evolved from a specialist service used mainly by technology firms into the default way organisations purchased computing infrastructure. According to Microsoft’s history of Azure, the shift fundamentally changed how businesses develop software, store information and scale digital services.
Today, cloud computing is entering another phase of development.
Artificial intelligence is driving unprecedented demand for specialist processors and enormous cloud data centres, while technologies such as edge computing are moving some computing resources closer to users to reduce delays. Governments are investing in sovereign cloud infrastructure, and researchers are already exploring how future quantum computers might also be delivered through the cloud rather than installed locally.
Looking back, the history of computing reveals a striking irony.
Society has travelled from shared computing to personal computing and then back to shared computing once again—but on an unimaginable scale. The difference is that today’s shared computers are no longer owned primarily by governments or universities. They are increasingly owned by a handful of global technology corporations.
The history of cloud computing is therefore not simply a story of technological progress.
It is the story of how society moved from owning computers to renting computing—a transformation that has reshaped not only information technology but also the ownership, organisation and power structures of the digital economy.
Failure and Resilience — When the Cloud Stops Working
One of the defining characteristics of infrastructure is that we rarely notice it until it fails.
Most people never think about cloud computing while checking their email, joining a video meeting or paying for shopping online. The cloud remains invisible because it usually works remarkably well. Yet when it fails, millions of people can suddenly discover just how much of modern life depends upon it.
Unlike the computer failures of the past, cloud outages rarely affect a single organisation. Because businesses, governments and public services increasingly rent computing from the same providers, one technical fault can disrupt thousands of organisations simultaneously.
Perhaps the best example occurred in December 2021, when Amazon Web Services (AWS) suffered a major outage centred on its US-East-1 region in Northern Virginia—the world’s largest concentration of cloud infrastructure. A networking failure disrupted thousands of organisations across North America. Streaming platforms, food delivery services, financial companies, smart home devices and Amazon’s own retail operations all experienced problems. Services including Netflix, Disney+, Coinbase, Robinhood, Ring and the Associated Press were affected, revealing that many apparently unrelated organisations depended upon exactly the same hidden infrastructure (TechTarget, Preparing for Cloud Provider Outages).
A second reminder came in July 2024, when Microsoft Azure experienced a major global outage following a distributed denial-of-service (DDoS) cyberattack. Problems within Microsoft’s own defensive systems amplified the disruption, affecting Azure, Microsoft 365, Microsoft Teams and LinkedIn. Around the world, businesses temporarily lost access to email, collaboration software and cloud-hosted applications, bringing parts of the working day to a standstill (Microsoft, Preliminary Post Incident Review – July 2024 Azure Incident).
Google Cloud has experienced similar problems. In May 2024, a networking issue affected multiple Google Cloud services across several regions, disrupting applications hosted on Google’s infrastructure. Although resolved relatively quickly, the incident demonstrated that even the world’s most sophisticated cloud platforms remain vulnerable to technical failures.
Fortunately, outages of this scale remain uncommon, and most are measured in hours rather than days. Their significance lies not in their duration but in what they reveal.
Banks, hospitals, airlines, retailers, universities, government departments and media organisations may appear to operate independently, yet many rely upon the same underlying cloud platforms. A fault affecting one hyperscale provider therefore becomes more than a technical problem. It becomes a disruption to everyday social and economic life.
Cloud computing also depends upon other hidden infrastructures. It requires uninterrupted electricity supplies, resilient fibre-optic networks, undersea cables carrying international data traffic and a steady flow of advanced semiconductor chips. Failure in one system can quickly cascade into another. A power outage can threaten data centres. Damage to an undersea cable can slow international communications. Geopolitical disruption affecting semiconductor production can delay the expansion of cloud infrastructure for years.
This is why resilience has become one of the defining principles of cloud engineering.
Critical information is replicated across multiple Availability Zones and often across different Regions, ensuring that services can continue operating even if an individual data centre fails. Cloud providers continuously monitor hardware, automatically moving workloads away from failing equipment before customers notice a problem. Backup power systems, duplicate fibre connections and sophisticated cybersecurity defences all exist for one reason: to prevent local failures becoming global disruptions.
Yet complete resilience is impossible.
As society has shifted from owning computers to renting computing, it has also shifted from isolated failures to shared risks. A company server crashing once affected a single business. Today, a fault inside a hyperscale cloud provider can affect thousands of organisations and millions of people within minutes.
Sociologist Susan Leigh Star observed that infrastructure becomes visible only when it breaks down. Cloud computing illustrates this perfectly. We rarely think about the cloud while it is working. We notice it only when a payment fails, a meeting disconnects, a booking system crashes or a favourite website suddenly disappears.
The major cloud outages of recent years reveal a profound sociological lesson.
The convenience of modern life rests upon shared infrastructures that most people never see. Those infrastructures have made computing cheaper, more reliable and more accessible than ever before. At the same time, they have created new forms of collective dependence. When we chose to rent computing instead of owning it, we also chose to share its vulnerabilities.
That is the hidden trade-off at the heart of the cloud.
Sociological Significance — From Owning Computers to Renting Computing
Cloud computing is often described as a technological revolution. It has undoubtedly transformed how organisations store information, develop software and deliver digital services. Yet its greatest significance is not technical.
It is social.
Throughout this chapter, we have seen how cloud computing has quietly changed one of the most fundamental relationships in modern society. For much of the twentieth century, individuals and organisations owned the computers that powered their lives. Today, they increasingly rent computing instead.
That shift has profound sociological consequences.
At first glance, cloud computing appears to decentralise power. Almost anyone can launch a business, store unlimited data or access artificial intelligence without investing millions of pounds in computer hardware. Small organisations can now use computing resources that only the world’s largest corporations could once afford. In many respects, cloud computing has democratised access to technology.
Yet beneath this apparent decentralisation lies an opposite trend.
The infrastructure itself has become increasingly concentrated. Millions of organisations appear to operate independent digital systems, but many rely upon exactly the same cloud providers, the same data centres and the same fibre-optic networks. Computing has become more accessible than ever before, while ownership of that computing has become concentrated in fewer hands.
This is a pattern that extends far beyond cloud computing.
Modern societies increasingly rely upon infrastructures that are shared rather than individually owned. Most of us do not generate our own electricity, operate our own mobile phone networks or build our own roads. Instead, we depend upon large systems managed by specialist organisations. Cloud computing represents the latest stage in this broader historical process. Computing has become another essential utility—one that flows to us when required but remains largely invisible while doing so.
Sociologist Manuel Castells argued that modern societies are increasingly organised through networks rather than isolated institutions. Cloud computing provides a striking example of this transformation. Schools, hospitals, banks, retailers, governments and families may appear independent, yet they are increasingly connected through shared digital infrastructures spanning continents. These networks enable extraordinary levels of communication and economic activity, but they also create new forms of dependence and vulnerability.
At the same time, Nick Srnicek argues that many of today’s largest technology companies have evolved into platforms. Their power lies not simply in producing software but in owning the infrastructure upon which countless other organisations depend. Cloud computing illustrates this perfectly. Amazon, Microsoft and Google do not merely provide online services; they increasingly provide the digital foundations on which much of modern society operates.
This helps explain why debates about cloud computing are no longer simply technical.
Questions about competition, regulation, cybersecurity, artificial intelligence and national sovereignty are all, at their heart, questions about infrastructure. Who owns it? Who controls it? Who benefits from it? Who bears the risks when it fails?
These are fundamentally sociological questions.
They remind us that technology is never simply about machines. It is about the organisation of society itself.
C. Wright Mills argued that the sociological imagination allows us to connect personal experiences with wider social structures. Cloud computing offers a perfect example. Saving a document to the cloud, joining a Teams meeting or asking an AI assistant a question may feel like individual actions. In reality, they depend upon vast infrastructures built, owned and maintained by organisations operating on a global scale.
Cloud computing therefore reveals one of the defining characteristics of the twenty-first century.
As computing has become more accessible, its ownership has become more concentrated. We have gained unprecedented access to digital resources while becoming increasingly dependent upon infrastructures we neither see nor control.
The cloud is therefore more than a technological innovation.
It is one of the hidden infrastructures through which modern society is organised—and one of the clearest examples of how the sociological imagination helps us see the invisible systems shaping our everyday lives.
11. Looking Ahead — The Next Generation of the Cloud
Cloud computing has fundamentally changed how society organises computing, but its evolution is far from over. Over the coming decade, the cloud is likely to become larger, more intelligent and even more deeply embedded in everyday life. At the same time, it will face growing challenges from artificial intelligence, energy demand, geopolitics and regulation.
The greatest driver of change is artificial intelligence.
Training and operating large AI models requires extraordinary amounts of computing power, far beyond the resources available to most organisations. Instead of buying thousands of specialised processors, businesses increasingly rent AI infrastructure from cloud providers. According to Gartner, worldwide spending on public cloud services continues to grow rapidly, driven in large part by demand for generative AI and cloud-based computing resources.
This growth is transforming the cloud itself.
Traditional data centres were designed to support websites, email and business software. The next generation is increasingly being built around AI, with dense clusters of graphics processing units (GPUs), advanced liquid cooling systems and dedicated high-voltage electricity supplies. As AI becomes integrated into search engines, office software, healthcare, education and scientific research, demand for cloud computing is expected to accelerate further.
At the same time, cloud infrastructure is becoming more geographically distributed.
For two decades, computing has become increasingly centralised within enormous hyperscale data centres. Yet many emerging technologies—including autonomous vehicles, industrial robotics, augmented reality and real-time healthcare—cannot tolerate even tiny delays. This is encouraging the growth of edge computing, where smaller facilities are located much closer to users while remaining connected to larger cloud regions.
Governments are also rethinking the cloud.
Growing concerns over cybersecurity, data sovereignty and geopolitical tensions have encouraged many countries to invest in sovereign cloud infrastructure, ensuring that sensitive public data remains under domestic legal jurisdiction. Rather than relying entirely upon global cloud providers, governments are increasingly seeking greater control over critical digital infrastructure.
Environmental pressures are likely to become equally important.
Cloud computing already consumes significant quantities of electricity, water and advanced semiconductor hardware. As artificial intelligence expands, providers will face increasing pressure to improve energy efficiency, develop more sustainable cooling technologies and secure reliable sources of low-carbon electricity. The future growth of the cloud may depend as much on the availability of power as on advances in computing technology itself.
Another important trend is integration.
Cloud providers are evolving from companies that rent computing into businesses offering complete digital ecosystems, combining artificial intelligence, cybersecurity, communications, data analytics and software development within a single platform. While this makes technology easier for customers to manage, it may also increase dependence on a relatively small number of providers and further strengthen their market position.
Looking further ahead, technologies such as quantum computing may eventually represent the next major transformation. Although still at an early stage of development, quantum computers are unlikely to sit on office desks. Instead, organisations will almost certainly access them through cloud platforms, continuing the long-term trend towards renting increasingly powerful computing rather than owning it.
Taken together, these developments point towards a broader sociological conclusion.
Cloud computing has never been simply about better computers. It has been about changing patterns of ownership, infrastructure and dependence. Each technological advance has made computing more accessible while simultaneously concentrating more of the underlying infrastructure within a relatively small number of organisations.
That process is unlikely to end.
Artificial intelligence, edge computing, sovereign clouds and quantum technologies will undoubtedly reshape how cloud computing operates. Yet they are all likely to reinforce the central theme of this chapter rather than overturn it.
The future of computing will not sit on our desks.
It will continue to exist as a shared infrastructure—globally connected, largely invisible and increasingly essential to the functioning of modern society.
Conclusion — Computing Is Just the Beginning
Cloud computing has quietly transformed one of the world’s most important resources. In little more than two decades, society has moved from owning computers to renting computing. What once sat in office server rooms and under our desks is now delivered on demand from vast networks of remote data centres owned by a handful of global corporations.
For most people, this transformation has been almost invisible.
We simply expect our files to appear on every device, our video calls to connect instantly and our online services to be available whenever we need them. Yet behind these everyday experiences lies an immense physical infrastructure of buildings, fibre-optic networks, electricity grids and semiconductor factories, organised into one of the largest shared computing systems ever constructed.
Cloud computing illustrates a recurring theme throughout this book. Modern societies increasingly depend upon infrastructures that are hidden from view precisely because they work so well. They shape our lives not through dramatic moments of technological change but through the quiet normality of everyday dependence.
But computing is only one part of the story.
Every time we tap a contactless bank card, transfer money through a mobile banking app or buy something online, another hidden infrastructure springs into action. Within seconds, encrypted messages pass between retailers, banks, payment processors and international card networks, authorising transactions worth trillions of pounds every year. Like cloud computing, these systems are largely invisible. Most people use them several times a day without ever considering how they work or who controls them.
If cloud computing has changed where modern society performs its calculations, payment networks determine how value moves through the digital economy.
The next section therefore follows another hidden infrastructure that most of us rely on daily but rarely see. It explores the global payment networks that allow money to flow almost instantaneously around the world—and asks what their hidden architecture reveals about power, trust and the organisation of modern economic life.
Hidden Infrastructures
This post is one in a series which explores hidden global infrastructures… you might also like my previous two posts…