# A private AI server: your prompts and documents stay on it.

> A private AI server is a dedicated GPU server for your team, running open-weight models you choose behind a private, authenticated endpoint. We set up model serving, monitor it and keep it updated, for a fixed monthly price by GPU type, quoted before setup.

Source: https://wharfcloud.co.uk/private-ai-server · Last updated 21 October 2026 · WharfCloud

## At a glance

- **Price:** Fixed monthly, by GPU type, quoted first
- **Models:** Open-weight models of your choice
- **Access:** Private, authenticated API endpoint
- **Your data:** Not used to train anything
- **Managed:** Setup, monitoring and updates by us
- **How to order:** Start with a short conversation

## Why run your own model at all

Public AI assistants are useful, but many firms cannot paste client files, payroll data or unpublished accounts into them. A private server runs the model on hardware dedicated to you. Prompts and documents are processed there and are not shared with a model provider.

Open-weight models have become good enough for summarising, drafting, extracting data from documents and answering questions over your own files. Our guide [Private AI and UK GDPR](https://wharfcloud.co.uk/guides/private-ai-uk-gdpr) covers what to check before you start.

## What we set up and run

Table: Private AI server: who does what

| Task | WharfCloud | You |
| --- | --- | --- |
| Choosing a GPU size for the models and users you have | Yes, with you | Tell us the use case |
| Model serving behind an authenticated endpoint | Yes |  |
| Monitoring, restarts and security updates | Yes |  |
| Upgrading to newer model versions | On request | Decide when |
| Your applications and prompts | We help when asked | Yes |

We agree the hosting location with you before setup and record it in your order. If you only need GPUs occasionally, [hourly GPU](https://wharfcloud.co.uk/gpu) from a prepaid balance is cheaper.

## How it works

1. **Tell us the use case.** Which tasks, how many users, which documents. We suggest a model and a GPU size.
2. **Fixed quote.** You get a fixed monthly price in writing before anything is set up.
3. **Setup.** After payment we build the server, deploy the model and give you the endpoint and keys.
4. **We run it.** Monitoring, updates and help in UK business hours, with critical incidents answered by a person within 1 hour, 07:00 to 23:00 every day.

## Price

- **Managed:** Fixed monthly price by GPU type, quoted before setup.

Prices in GBP. No VAT is added: we are not VAT registered. Talk to us first: https://wharfcloud.co.uk/contact?topic=private-ai-server

## Questions we get asked

### How much does a Private AI server cost?

It is a fixed monthly price that depends on the GPU type. We quote it in writing after a short conversation, before anything is set up. No VAT is added.

### Which models can it run?

Open-weight models whose licence allows your use, sized to fit the GPU. We will tell you if a model you want needs a bigger GPU or is not licensed for commercial use.

### Do you use our data to train models?

No. Your prompts and documents are processed on your server for your requests only.

### Is it UK GDPR compliant?

We act as your processor under a [data processing agreement](https://wharfcloud.co.uk/legal/dpa), and agree the hosting location with you. Compliance also depends on how you use it, which our [guide](https://wharfcloud.co.uk/guides/private-ai-uk-gdpr) explains.

---

WharfCloud is a trading name of Sunucu Internet Services Consultancy Ltd, registered in England and Wales, company number 12963678. Not VAT registered.
Support: engineers Monday to Friday, 08:00 to 18:00 UK time. Critical incidents: A person responds to critical incidents within 1 hour (Every day, 07:00 to 23:00 UK time). Automated monitoring and automated recovery overnight and at weekends.
Contact: hello@wharfcloud.co.uk · https://wharfcloud.co.uk/contact