Explainer
How cloud computing works: data centres, virtual machines and SaaS
Cloud computing means using computing resources, such as servers, storage and software, over a network on demand, typically paying for what you use; NIST defines it by five characteristics, three service models and four deployment models.[1][2][3] Providers pool computing resources among many customers, split physical servers into virtual machines and spread them across separate locations so that one failure does not take everything down.[2][4][5] AWS, Microsoft Cloud and Google Cloud each reported more than $30 billion in revenue in the third quarter of 2026.[6][7][8]
Revenue figures were checked on 7 November 2026 and reflect the July to September 2026 quarter. This page uses AWS documentation for its examples of regions, virtualization and security.
What “the cloud” means
The US National Institute of Standards and Technology (NIST) published a standard definition in 2011.[1] Cloud computing is a model for convenient, on-demand network access to a shared pool of configurable computing resources, such as networks, servers, storage, applications and services, that can be rapidly provisioned and released with minimal management effort.[1] NIST’s model has five essential characteristics, three service models and four deployment models.[3]
The five essential characteristics are:
- On-demand self-service: customers provision capacity themselves, automatically, without talking to anyone at the provider.[9]
- Broad network access: services are reached over the network through standard mechanisms from phones, tablets, laptops and workstations.[10]
- Resource pooling: the provider’s resources serve many customers through a multi-tenant model, reassigned as demand shifts.[2]
- Rapid elasticity: capacity scales out and back in with demand, and often appears unlimited to the customer.[9]
- Measured service: usage is metered and reported, typically on a pay-per-use basis.[2]
Service models: IaaS, PaaS and SaaS
Think of how much of the stack you manage yourself. With software as a service (SaaS), such as web-based email, you just use the provider’s application.[11] With platform as a service (PaaS), you bring your own application and the provider runs everything underneath.[11] With infrastructure as a service (IaaS), you rent the raw building blocks, processing, storage and networks, and install whatever you like, including the operating system.[11]
NIST distinguishes the models by the line of control. SaaS consumers use the provider’s applications and manage at most limited user settings; PaaS consumers deploy applications onto the provider’s platform; IaaS consumers provision processing, storage, networks and other fundamental resources and can run arbitrary software, including operating systems.[11] That line also sets the security split: under AWS’s shared responsibility model, a customer running an EC2 virtual server manages the guest operating system, its patches, installed software and firewall configuration, while AWS protects the underlying infrastructure.[12]
Deployment models: public, private, community, hybrid
NIST lists four deployment models. A public cloud is provisioned for open use by the general public; a private cloud serves a single organization; a community cloud serves organizations with shared concerns; and a hybrid cloud binds two or more of these together while they remain distinct.[13]
Inside the provider: regions, zones and virtual machines
Behind the abstraction are physical data centres. AWS groups its data centres into Regions, geographic locations each made of at least three physically separate Availability Zones; Regions are isolated so that a service failure stays within one Region.[14] An Availability Zone is one or more data centres with independent, redundant power, networking and connectivity.[5] Zones in a Region are up to about 100 km apart, far enough to avoid shared disasters but close enough to copy data between them in milliseconds, and AWS runs more than 100 of them.[5] Its Regions are linked by a private network of more than 9 million kilometres of land and undersea fibre-optic cable.[15]
Pooled resources can be physical or virtual.[2] In AWS’s Nitro System, a hypervisor takes commands to start and stop virtual machines and divides a server’s memory and processor capacity between them using the processor’s built-in virtualization features.[4] With infrastructure as a service, customers can then install their own operating system and software on those virtual machines; see how operating systems work.[11]
The business of cloud
In the third quarter of 2026, Amazon Web Services sales were $46.0 billion, up 39% from a year earlier.[6] Microsoft Cloud revenue was $63.9 billion in the quarter to 30 September 2026, with Azure growing 48%.[7] Google Cloud revenue was $30.7 billion in the same quarter.[8] The three figures are not directly comparable, because each company groups its cloud businesses differently; Microsoft’s Agents and Infra segment, for example, includes both Azure and Microsoft 365 cloud services.[16]
The providers are spending heavily on the infrastructure behind these services. Amazon’s cash capital expenditures were $172.2 billion in the first nine months of 2026, mainly for technology infrastructure, most of it supporting AWS, and for fulfillment capacity.[17] Microsoft says it will keep investing to support growth in its cloud offerings and its AI training and other infrastructure.[18] Why AI chips for these data centres are concentrated among a few suppliers is analysed in AI chip supply concentration.
Read next
How CPUs work: instructions, cores, caches and clock speed
How a computer's central processor runs software: the fetch-decode-execute cycle, registers and caches, cores and threads, instruction sets and RISC.
Questions readers ask
What is cloud computing in simple terms?
NIST defines it as on-demand network access to a shared pool of configurable computing resources, such as networks, servers, storage, applications and services, that can be provisioned and released quickly with minimal management effort.[1]
What is the difference between IaaS, PaaS and SaaS?
With software as a service you use the provider's applications; with platform as a service you deploy your own applications onto the provider's platform; with infrastructure as a service you rent processing, storage and networks and can run any software, including your own operating system.[11]
What is a public, private or hybrid cloud?
A public cloud is open to the general public; a private cloud serves a single organization; a community cloud serves a group of organizations with shared concerns; and a hybrid cloud combines two or more of these.[13]
What is an availability zone?
In AWS's terms, an Availability Zone is one or more data centres with independent power, networking and connectivity. Each public AWS Region has at least three, placed up to about 100 km apart so a local disaster does not hit them all.[5][14]
Who is responsible for security in the cloud?
Under AWS's shared responsibility model, AWS protects the infrastructure that runs its services, while customers who run virtual servers are responsible for their guest operating system, applications, firewall settings and data.[12]
Sources
· 18 claims from 6 publishers
Each numbered claim is a statement we checked against the sources listed with it. Tap a number in the text to see its sources in place.
- [1]
NIST defines cloud computing as a model for convenient, on-demand network access to a shared pool of configurable computing resources, such as networks, servers, storage, applications and services, that can be rapidly provisioned and released with minimal management effort. confirmedas of 2011-09-28
- The NIST Definition of Cloud Computing, SP 800-145 · National Institute of Standards and Technology · Primary source · 2011-09-28 · SP 800-145, Section 2, PDF p. 6 (retrieved 2026-11-07)
- [2]
NIST's resource pooling characteristic means a provider's resources serve many customers through a multi-tenant model, with physical and virtual resources assigned and reassigned according to demand, and measured service means usage is metered, monitored and reported, typically on a pay-per-use basis. confirmedas of 2011-09-28
- The NIST Definition of Cloud Computing, SP 800-145 · National Institute of Standards and Technology · Primary source · 2011-09-28 · SP 800-145, Essential Characteristics, Resource pooling (retrieved 2026-11-07)
- The NIST Definition of Cloud Computing, SP 800-145 · National Institute of Standards and Technology · Primary source · 2011-09-28 · SP 800-145, footnote 1 (retrieved 2026-11-07)
- [3]
NIST's cloud model has five essential characteristics, three service models and four deployment models. confirmedas of 2011-09-28
- The NIST Definition of Cloud Computing, SP 800-145 · National Institute of Standards and Technology · Primary source · 2011-09-28 · SP 800-145, Section 2, PDF p. 6 (retrieved 2026-11-07)
- [4]
In AWS's Nitro System, the hypervisor receives virtual machine commands such as start and stop and partitions memory and CPU resources between virtual machines using the hardware virtualization features of the server processor. confirmedas of 2026-11-07
- The components of the Nitro System · Amazon Web Services · Primary source · The Security Design of the AWS Nitro System, The Nitro Hypervisor (retrieved 2026-11-07)
- [5]
An AWS Availability Zone is one or more discrete data centers with independent, redundant power, networking and connectivity; zones in a Region are up to about 100 km apart to avoid correlated failures but close enough for synchronous replication, and AWS operates more than 100 of them. confirmedas of 2026-11-07
- Availability Zones (AWS Fault Isolation Boundaries) · Amazon Web Services · Primary source · AWS Fault Isolation Boundaries, Availability Zones (retrieved 2026-11-07)
- Availability Zones (AWS Fault Isolation Boundaries) · Amazon Web Services · Primary source · AWS Fault Isolation Boundaries, Availability Zones (retrieved 2026-11-07)
- Availability Zones (AWS Fault Isolation Boundaries) · Amazon Web Services · Primary source · AWS Fault Isolation Boundaries, Availability Zones (retrieved 2026-11-07)
- [6]
Amazon's third-quarter 2026 net sales rose 12% to $201.9 billion; North America sales were $112.9 billion, International $43.1 billion and AWS $46.0 billion, up 39%. confirmedas of 2026-10-29
- Amazon.com Announces Third Quarter Results (Exhibit 99.1) · Amazon (SEC filing) · Primary source · 2026-10-29 · Summary bullets (retrieved 2026-11-07)
- Amazon.com Announces Third Quarter Results (Exhibit 99.1) · Amazon (SEC filing) · Primary source · 2026-10-29 · Summary bullets (retrieved 2026-11-07)
- Amazon.com Announces Third Quarter Results (Exhibit 99.1) · Amazon (SEC filing) · Primary source · 2026-10-29 · Summary bullets (retrieved 2026-11-07)
- [7]
Microsoft Cloud revenue was $63.9 billion in the quarter ended September 30, 2026, up 30%, and Azure revenue grew 48%. confirmedas of 2026-10-28
- Microsoft FY27 first quarter earnings press release · Microsoft Investor Relations · Primary source · 2026-10-28 (retrieved 2026-11-07)
- Microsoft FY27 first quarter earnings press release · Microsoft Investor Relations · Primary source · 2026-10-28 (retrieved 2026-11-07)
- [8]
In the third quarter of 2026, Google Search & other revenue was $63.4 billion, YouTube ads $11.2 billion, Google Network $7.5 billion, Google subscriptions, platforms and devices $14.3 billion, Google Cloud $30.7 billion and Other Bets $372 million. confirmedas of 2026-10-28
- Alphabet announces third quarter 2026 results (Form 8-K, Exhibit 99.1) · Alphabet (SEC filing) · Primary source · 2026-10-28 · Q3 2026 Supplemental Information ($ millions; Q3 2025, Q3 2026) (retrieved 2026-11-07)
- Alphabet announces third quarter 2026 results (Form 8-K, Exhibit 99.1) · Alphabet (SEC filing) · Primary source · 2026-10-28 · Q3 2026 Supplemental Information, Google Search & other row (retrieved 2026-11-07)
- [9]
Among NIST's essential characteristics, on-demand self-service lets a customer provision computing capabilities automatically without human interaction with the provider, and rapid elasticity lets capabilities scale outward and inward with demand, often appearing unlimited to the customer. confirmedas of 2011-09-28
- The NIST Definition of Cloud Computing, SP 800-145 · National Institute of Standards and Technology · Primary source · 2011-09-28 · SP 800-145, Essential Characteristics, On-demand self-service (retrieved 2026-11-07)
- The NIST Definition of Cloud Computing, SP 800-145 · National Institute of Standards and Technology · Primary source · 2011-09-28 · SP 800-145, Essential Characteristics, Rapid elasticity (retrieved 2026-11-07)
- [10]
NIST's broad network access characteristic means cloud capabilities are available over the network through standard mechanisms usable from many kinds of devices, such as phones, tablets, laptops and workstations. confirmedas of 2011-09-28
- The NIST Definition of Cloud Computing, SP 800-145 · National Institute of Standards and Technology · Primary source · 2011-09-28 · SP 800-145, Essential Characteristics, Broad network access (retrieved 2026-11-07)
- [11]
NIST's three service models are software as a service (SaaS), where customers use the provider's applications; platform as a service (PaaS), where they deploy their own applications onto the provider's platform; and infrastructure as a service (IaaS), where they provision processing, storage and networks and can run arbitrary software, including operating systems. confirmedas of 2011-09-28
- The NIST Definition of Cloud Computing, SP 800-145 · National Institute of Standards and Technology · Primary source · 2011-09-28 · SP 800-145, Service Models, SaaS (retrieved 2026-11-07)
- The NIST Definition of Cloud Computing, SP 800-145 · National Institute of Standards and Technology · Primary source · 2011-09-28 · SP 800-145, Service Models, SaaS (retrieved 2026-11-07)
- The NIST Definition of Cloud Computing, SP 800-145 · National Institute of Standards and Technology · Primary source · 2011-09-28 · SP 800-145, Service Models, SaaS (retrieved 2026-11-07)
- The NIST Definition of Cloud Computing, SP 800-145 · National Institute of Standards and Technology · Primary source · 2011-09-28 · SP 800-145, Service Models, PaaS (retrieved 2026-11-07)
- The NIST Definition of Cloud Computing, SP 800-145 · National Institute of Standards and Technology · Primary source · 2011-09-28 · SP 800-145, Service Models, IaaS (retrieved 2026-11-07)
- [13]
NIST's four deployment models are private cloud, for one organization; community cloud, for a group of organizations with shared concerns; public cloud, open to the general public; and hybrid cloud, a combination of two or more distinct cloud infrastructures. confirmedas of 2011-09-28
- The NIST Definition of Cloud Computing, SP 800-145 · National Institute of Standards and Technology · Primary source · 2011-09-28 · SP 800-145, Deployment Models, Public cloud (retrieved 2026-11-07)
- The NIST Definition of Cloud Computing, SP 800-145 · National Institute of Standards and Technology · Primary source · 2011-09-28 · SP 800-145, Deployment Models, Hybrid cloud (retrieved 2026-11-07)
- [14]
AWS describes its Regions as geographical locations where it clusters data centers; each public Region has at least three physically separate Availability Zones, and Regions are isolated so that service failures are limited to one Region. confirmedas of 2026-11-07
- Regions (AWS Fault Isolation Boundaries) · Amazon Web Services · Primary source · AWS Fault Isolation Boundaries, Regions (retrieved 2026-11-07)
- Regions (AWS Fault Isolation Boundaries) · Amazon Web Services · Primary source · AWS Fault Isolation Boundaries, Regions (retrieved 2026-11-07)
- [15]
AWS says its Regions are linked by a private global network of more than 9 million kilometres of land and subsea fiber-optic cable. confirmedas of 2026-11-07
- Regions (AWS Fault Isolation Boundaries) · Amazon Web Services · Primary source · AWS Fault Isolation Boundaries, Regions (retrieved 2026-11-07)
- [16]
Microsoft's Agents and Infra segment includes Azure and Microsoft 365 cloud, which covers Microsoft 365 Copilot, GitHub Copilot and security offerings such as Security Copilot. confirmedas of 2026-10-28
- Microsoft Corporation Form 10-Q for the quarter ended September 30, 2026 · Microsoft (SEC filing) · Primary source · 2026-10-28 · Note 16 (segment information) (retrieved 2026-11-07)
- [17]
Amazon's cash capital expenditures were $75.9 billion in the third quarter of 2026 and $172.2 billion in the first nine months of the year, mainly for technology infrastructure (most of it to support AWS) and fulfillment capacity. confirmedas of 2026-10-30
- Amazon.com, Inc. Form 10-Q for the quarter ended September 30, 2026 · Amazon (SEC filing) · Primary source · 2026-10-30 · Item 2, Liquidity and Capital Resources (retrieved 2026-11-07)
- Amazon.com, Inc. Form 10-Q for the quarter ended September 30, 2026 · Amazon (SEC filing) · Primary source · 2026-10-30 · Item 2, Liquidity and Capital Resources (retrieved 2026-11-07)
- [18]
Microsoft says it will keep investing in capital expenditures to support growth in its cloud offerings and its investments in AI training and other infrastructure, and it also leases datacenters and related infrastructure. confirmedas of 2026-10-28
- Microsoft Corporation Form 10-Q for the quarter ended September 30, 2026 · Microsoft (SEC filing) · Primary source · 2026-10-28 · Item 2. Other Planned Uses of Capital (retrieved 2026-11-07)
- Microsoft Corporation Form 10-Q for the quarter ended September 30, 2026 · Microsoft (SEC filing) · Primary source · 2026-10-28 · Note 12 – Leases (retrieved 2026-11-07)
Revision history (1)
- Page created.
Created Nov 7, 2026. Last reviewed by an editor on Nov 8, 2026. Next scheduled review: Feb 7, 2027.
Cite this page
"How cloud computing works: data centres, virtual machines and SaaS." ContentLora, updated Nov 7, 2026. https://contentlora.com/explain/how-cloud-computing-works
Spotted an error? Suggest a correction or emailcorrections@contentlora.com.