Public, private and hybrid cloud are the three main deployment models for cloud computing. They describe who the underlying infrastructure is dedicated to and how it is combined, rather than what you run on it. Public cloud runs on a provider’s shared infrastructure, used by many customers. Private cloud runs on infrastructure dedicated to a single organization, whether in its own data center, in colocation or hosted by a provider. Hybrid cloud connects private infrastructure and public cloud so workloads, data and management can work across both.
At a glance
- Public cloud: shared, provider-owned infrastructure, rented on demand and usually billed by usage.
- Private cloud: infrastructure dedicated to one organization, with cloud-style self-service and automation, on-premises or hosted.
- Hybrid cloud: private and public environments connected and managed together.
- The choice is usually made workload by workload, based on cost, performance, compliance and how much control you need.
- Hybrid cloud is different from multi-cloud, which means using more than one public cloud provider.
What problem it solves
“Move to the cloud” is not a single decision. Some applications benefit from public cloud’s elasticity and managed services. Others run around the clock at steady load, depend on specialized hardware or licensing, or are subject to rules about data location and control, and may be cheaper or simpler on dedicated infrastructure. Treating every workload the same leads to either expensive public cloud bills or an aging data center that slows the business down.
The three models give buyers a vocabulary for placing each workload deliberately and for planning how the environments connect, rather than ending up with disconnected islands.
How it works
Public cloud. Providers such as AWS, Microsoft Azure and Google Cloud run large data centers and rent capacity from shared pools: virtual machines, storage, databases, analytics, AI services and more. You provision resources through a portal or API and typically pay for what you use, with discounts for commitments. The provider secures the infrastructure; you are responsible for configuring your accounts, identities, networks and data. See our Public Cloud solutions page.
Private cloud. Infrastructure is dedicated to one organization and runs a cloud platform, often built on VMware, Microsoft, Nutanix or OpenStack technology, sometimes on hyper-converged infrastructure. Teams get self-service provisioning and automation, while the organization keeps control of hardware, location and configuration. It can be run by your own staff or delivered as a managed, hosted service. See our Private Cloud solutions page.
Hybrid cloud. Private and public environments are linked through private network connections or VPNs, a shared identity system, and common management, monitoring and security tools. Some providers sell hardware that runs their public cloud services in your data center, and some private cloud platforms extend into public clouds, which can make moving workloads easier. In practice, hybrid usually means placing each application in one environment and integrating them, rather than constantly moving workloads back and forth.
When it matters for buyers
- During a data center refresh or exit. Hardware end of life is the natural point to decide which workloads go to public cloud, which stay private and where private infrastructure should live.
- When public cloud costs outgrow expectations. Steady workloads are sometimes moved to private cloud, bare metal or colocation.
- When compliance requirements tighten. Data residency, audit and customer contract terms can favor dedicated infrastructure for some systems.
- When the business needs new capabilities fast. Public cloud services for analytics, AI and global reach can be adopted without building them.
- When licensing changes. Changes to virtualization or software licensing terms can shift the economics between models.
Questions to ask vendors
- For public cloud: what will this workload cost over three years, including data transfer, support and commitment terms?
- For private cloud: who owns and refreshes the hardware, and what self-service and automation do we get?
- For hosted private cloud: is the hardware dedicated to us, and where is it located?
- How do we connect private and public environments, and what does that connectivity cost?
- Which identity, security and monitoring tools work across both environments?
- What does it take to move a workload from one model to another later?
How it differs from multi-cloud
These models are often confused with multi-cloud. Public, private and hybrid describe the type of infrastructure and how it is combined. Multi-cloud describes how many public cloud providers you use. An organization running its own private cloud alongside one public cloud is hybrid but not multi-cloud; one running workloads on AWS and Azure only is multi-cloud but not hybrid; one can also be both.
Problems this helps with
- Not knowing which workloads belong in the public cloud and which don't
- Cloud bills that are higher than the data center they replaced
- Compliance rules about where data is stored and who controls the hardware
- Systems on-premises and in the cloud that don't work well together
