Guide
On-prem AI appliance: the complete guide
What an AI appliance includes, how it's deployed, and what to check before you buy one.
By James Drayson
In short
An on-prem AI appliance is a self-contained system, hardware, an AI model, and software, that runs AI inside your own environment instead of in the cloud. It delivers owned, private AI at a fixed cost, deployable on-prem, in a private cloud, or air-gapped.
What an AI appliance includes
- GPU hardware: Sized to your model, from a single GPU server to multi-GPU nodes.
- The model: Ideally one you own, trained on your domain rather than a generic endpoint.
- Serving & apps: An API, chat UI, and usage platform so teams can use it day one.
- Management: Monitoring, updates, and retraining handled without sending data out.
Deployment models
- On-premise: Runs entirely in your own data centre or office.
- Private cloud: Deployed inside your own cloud tenant and region.
- Sovereign cloud: Hosted in a UK sovereign cloud where you still own the model.
- Air-gapped: Fully isolated for classified or highly sensitive work.
Buyer's checklist
- Ownership: Do you get the model weights, or just access?
- Air-gap support: Can it run with no external connectivity?
- Domain training: Will it be trained on your data, not just hosted?
- Total cost: Compare the fixed cost against years of per-token API billing.
What this looks like with Locai
Running AI locally means assembling hardware, a model and a serving stack that work together. Here is what it looks like when that arrives as one product.
Locai Labs builds Locai One, an on-prem AI appliance. It is one machine that arrives with everything already in it: the hardware, our open-weight Locai Juno models, and Locai OS, the operating system that serves the models and handles users, access and monitoring. You plug it into a mains socket and your network, and your team is working in about 15 minutes. No cloud account, no per-token bill, and nothing leaving the building.
The reason a data-centre-class model fits in a box on your floor is SPACE, our compression algorithm. Instead of asking how much of a model can be cut while keeping it broadly similar, SPACE asks what the model needs to be good at, preserves the subnetworks behind those capabilities and strips back the rest. The result is a smaller specialist rather than a shrunken generalist, tuned to the exact hardware it ships on.
Locai One starts at £29,950 for a team, and Locai One Pro at £49,950 for an organisation, bought once and owned outright. Both run air-cooled on standard mains power and work fully air-gapped. Any compatible open-weight model runs alongside Juno, and if you need a model trained on your own proprietary data we can post-train one and deploy it on the same machine.
Frequently asked questions
What is an AI appliance?
A ready-to-run system bundling AI hardware, a model, and software so you can run AI inside your own environment without building the stack.
On-prem vs cloud appliance, what's the difference?
An on-prem appliance runs in your data centre with data never leaving; a cloud appliance runs in a provider's environment. On-prem gives the strongest residency and control.
Can an AI appliance run air-gapped?
Yes. Because the model and hardware are local, an appliance can run fully isolated with no internet connection.
How is an AI appliance priced?
As a fixed-cost asset rather than per-token usage. Pricing is configured to your deployment; see our on-prem AI cost guide to compare against your current API spend.
Book a sovereign AI briefing
A 30-minute session on owning your model: deployment options, the data path, and a clear cost range for your use case.