Almost every UK business now runs on some kind of cloud, but the word covers three quite different things. Public cloud is renting computing from a giant shared provider. Private cloud is the same flexible model kept on hardware reserved for you. Hybrid cloud joins the two so a workload can sit wherever it makes most sense. Get the distinction right and you spend money where it earns its keep; get it wrong and you either overpay for elasticity you never use or trap yourself on infrastructure you cannot scale. This guide explains the three models in plain English, what each is genuinely good at, and how to decide without the marketing fog.
The one idea behind all three
Strip away the branding and cloud means a simple promise: you get computing power, storage and software as a service you can turn up or down, instead of a one-off box you buy and babysit. What changes between public, private and hybrid is not that promise but where the underlying kit physically lives and who else is using it. Once you hold that thought, the three models stop being competing buzzwords and become three answers to one question, which is who runs the hardware and how much of it is shared.
It helps to think about a car. Public cloud is a taxi: someone else owns the fleet, you pay per trip, and you never see the maintenance bill. Private cloud is a company car kept in your own garage: still convenient and on demand for your drivers, but the vehicle and the garage are yours. Hybrid cloud is having both, and being sensible about which one you take for which journey.
Public cloud: rented, shared, instantly elastic
Public cloud is what most people mean when they say cloud. Providers such as AWS, Microsoft Azure and Google Cloud run enormous data centres and rent you a slice on demand. You share the physical hardware with thousands of other tenants, but software keeps each customer walled off from the others. The appeal is that you can go from nothing to a hundred servers in minutes, pay only for what you use, and let the provider worry about power, cooling and hardware failure.
The trade-offs show up in two places. First, cost can creep: pay-as-you-go is brilliant for spiky or unpredictable demand, but a workload that runs flat-out around the clock is often cheaper to own outright. Second, you have less direct control over where data sits and how the underlying platform behaves, which matters in regulated UK sectors. If you are weighing the public providers, our guide to choosing AWS vs Azure for the UK mid-market goes deeper.
- •Best for: variable demand, new projects, global reach, and avoiding upfront capital cost
- •Watch out for: bills that grow with success, egress charges, and reduced control over the stack
- •You manage: your applications and data; the provider manages everything underneath
Private cloud: the cloud model on hardware that is yours
Private cloud takes the same self-service, scale-on-demand experience and runs it on infrastructure dedicated to a single organisation. That hardware can sit in your own server room, in a colocation data centre, or be hosted for you, but the key point is that no other customer shares it. You get the agility of cloud with the control, predictable cost and isolation of owning the kit, which is why it remains popular with finance, healthcare and anyone with strict data-residency rules.
The catch is that someone has to design, run and refresh that platform, whether that is your team or a partner. Modern private clouds are usually built on virtualisation or hyperconverged infrastructure, and we cover the economics of that approach in when HCI beats traditional infrastructure. If you want the dedicated model without owning a building, colocation is the usual home for it.
Hybrid cloud: the pragmatic middle
Hybrid cloud is not a product you buy, it is a strategy where public and private work together as one estate. A common pattern is to keep steady, sensitive or licence-heavy workloads on private infrastructure while bursting into public cloud for spikes, running test environments there, or using it as an off-site copy for disaster recovery. Done well, each workload lives where it is cheapest and safest, and you move it as circumstances change.
Hybrid is the default for most established UK businesses precisely because reality is messy: you rarely start from a blank sheet, and not every system wants the same home. The price of that flexibility is complexity. You need consistent identity, networking and security spanning both worlds, otherwise you simply own two problems instead of one. The aim is one operating model, not two silos that happen to share a logo.
How to choose without overthinking it
Decide per workload, not for the whole company at once. For each system, ask three questions: how predictable is its demand, how sensitive is its data, and how much in-house skill do you have to run it. Spiky and non-sensitive points to public. Steady, regulated or latency-critical points to private. A mix, which is most organisations, points to hybrid. Avoid picking a model because a competitor did; their demand curve and compliance burden are not yours.
Most businesses land on hybrid by accident and then tidy it up on purpose. If you are early in that journey, model the running cost honestly before committing, because the cheapest option on day one is rarely the cheapest over three years. Our cloud vs on-premises total cost of ownership breakdown is the right next read, and a server configuration conversation helps if private capacity is part of the answer.