Short answer: there is no single right model. Shared cloud is the right call for a smaller company still testing the value of AI on low-sensitivity data. Single-tenant makes sense when your production data cannot share an environment with anyone else, when a NIS2 audit is in play, or when you need hard control over isolation. The rest of this piece is a map for working out which of the two situations is yours.
We're leaving on-prem aside here, since owning hardware is a separate conversation. This is about a choice inside the cloud: a shared instance versus a dedicated, isolated instance that's yours alone.
What actually separates shared from single-tenant
Both run in the cloud. The difference is who you share the environment with.
- Shared cloud (multi-tenant): one instance serves many customers at once. Your data is logically separated, but it physically shares the same infrastructure, the same compute, and often the same model. The isolation is software, not hardware.
- Single-tenant (dedicated instance): you get a separate, isolated environment for yourself. No one else shares your resources. The data still physically lives with the infrastructure provider, but the environment is yours and only yours.
That distinction matters mostly in three places: when you answer an auditor's question, at the boundary of data you're not allowed to release, and in control over what changes in the environment and when.
When shared cloud is the right call
Despite what some vendors imply, shared cloud isn't the "lite" version. For many companies it's exactly where you should start.
- You're still testing whether AI helps at all. If you're at one workflow and one team, shared cloud starts fastest and doesn't tie up budget in a dedicated environment.
- The data is low-sensitivity. Marketing content, general documentation, material that already is or could be public. Here hardware isolation buys you no real protection, just cost.
- You're not under a hard audit. If you're neither an essential nor an important entity under NIS2 and no contract forces a dedicated environment, shared cloud meets the bar without overbuilding.
- Entry barrier and speed matter. Smaller team, limited budget, results needed in weeks. Shared cloud gives you the lowest barrier to start.
For a company under roughly 50 people testing a first use case on low-sensitivity data, shared cloud isn't a compromise. It's the right decision.
When single-tenant starts to make sense
Single-tenant comes into play when several of these signals show up together, not just one.
- Production data that can't be shared. Technical drawings, formulations, process documentation, data under NDAs. If the mere fact of sharing an environment breaches a contract or policy, isolation stops being optional.
- A NIS2 audit asking about the data boundary. The auditor will ask exactly where data is processed and who can reach it. "A dedicated environment, only ours, with documented isolation" is far easier to defend than "logical separation inside a shared instance."
- You need control over the change cycle. In shared cloud, provider-side changes hit everyone at once. Single-tenant lets you control when and what changes in your environment.
- Volume and horizon already justify it. Once AI becomes a daily tool rather than an experiment, a dedicated environment stops being overbuild and starts being predictability.
How they look side by side
The simplest way is to compare the two on the dimensions that actually drive the decision.
- Isolation: shared is logical, software-level. Single-tenant is full, an environment that's yours alone.
- Entry barrier: shared is low, single-tenant higher.
- Ease of defending to an auditor: shared needs you to explain the separation mechanism. Single-tenant answers in one sentence.
- Control over changes: shared sits with the provider. Single-tenant sits with you.
- Who it's for: shared for smaller teams, testing, with low-sensitivity data. Single-tenant for the regulated, with data that can't share an environment.
A four-step way to decide
- Map the data sensitivity in this one workflow. Not the whole company, just what actually goes into the model. If it's data that could be public, shared is enough.
- Check your regulatory status. Being an essential or important entity under NIS2 raises the weight of dedicated isolation.
- Ask what you'd tell the auditor. If "logical separation" satisfies you, shared is fine. If you need a hard boundary, that's a signal for single-tenant.
- Weigh the horizon. A quarter-long experiment versus a tool for years. The longer and heavier the use, the stronger the case for a dedicated environment.
For many mid-sized manufacturers the honest path is: start with shared cloud on a single low-sensitivity workflow, and reach for single-tenant once data, regulation and scale point the same way. For NIS2-covered entities working on documentation that can't be shared, single-tenant can be the right call from the start.
Not sure which model fits your production data? Book a 30-minute call with the founder, no pitch, just your case.
Fryderyk, CortexMine. We write about private AI for NIS2-covered manufacturers, based on our own deployments and tests.