A GPU server should not sit idle
A company buys a GPU server for its own work: training during the day, a render queue, a research project. Then the nights, the weekends and the weeks between projects come, and the most expensive machine in the building does nothing.
From today, companies that buy their GPU servers from Server Factory can put those hours on ParalonCloud, where developers and AI teams rent GPUs by the minute, and earn from every minute someone else uses the machine.
Who Server Factory is
Server Factory is a Supermicro Authorised Partner, also supplying ASUS, MSI and ASRock Rack, founded in 2010. From offices and warehouses in the UK and the USA, it specifies, builds and ships GPU servers, HPC and storage systems worldwide. Its UK headquarters in High Wycombe is ISO 9001 certified, and its United States office is in Des Moines, Iowa.
Its GPU range goes from a single card to an eight-GPU rack node, for training, inference and rendering on NVIDIA GPUs, configured to the model the customer runs and validated before it ships. That includes Supermicro's GPU servers and Gold Series systems and server cards such as the RTX PRO 6000 Blackwell Server Edition. Its GPU servers can be listed on ParalonCloud as soon as they are installed.
How it works
- Buy the server from Server Factory, specified for your own work.
- Create a ParalonCloud account and add your company under Settings → Company details, so you are paid by bank.
- Run one Docker command on the server, from Add a node. The connection is outbound only: no public IP and no open ports.
- We benchmark every GPU (memory bandwidth and compute, measured) and the server can be rented about three minutes later.
- Choose when it is available. Switch it on for the market when you are done for the day and off when you need it, from the dashboard or automatically through the Provider API.
- Get paid. You keep 80% of every rental, credited every minute it runs. Companies are paid monthly by bank transfer in US dollars, with the invoice issued for them.
What a rented hour earns
Every card has a default price per GPU hour, which owners can keep or change. At the defaults of 12 November 2026, one hour of rental earns the owner:
| Server | Default price | The owner keeps (80%) |
|---|---|---|
| 8× H200 | 8 × $3.70 an hour | $23.68 per rented hour |
| 8× RTX PRO 6000 Blackwell Server Edition | 8 × $1.50 an hour | $9.60 per rented hour |
| 4× RTX 5090 | 4 × $0.56 an hour | $1.79 per rented hour |
Earnings depend on how many hours the server is actually rented, which no one can promise. Current prices are always on our pricing page: H200, RTX PRO 6000 Blackwell Server Edition, RTX 5090.
What renters get
The people who rent these servers get a full machine through JupyterLab or SSH, billed per minute, with no egress fees, and an app such as ComfyUI or Open WebUI ready when they want one. Every server shows its measured GPU benchmark and a reputation built from its finished rentals, so a well-run machine is easy to pick.
Get started
- Need a GPU server? Talk to Server Factory about the configuration.
- Already have one? Add it to ParalonCloud. The requirements are on Hardware requirements.
Frequently asked questions
Who is Server Factory?
A Supermicro Authorised Partner that also supplies ASUS, MSI and ASRock Rack servers, founded in 2010. From offices and warehouses in the UK and the USA it specifies, builds and ships GPU servers, HPC and storage systems worldwide.
How do I list a Server Factory GPU server on ParalonCloud?
Install the NVIDIA driver (CUDA 12.4 or newer), Docker and the NVIDIA Container Toolkit, then run the one Docker command from Add a node. We benchmark every GPU and the server can be rented about three minutes later. The connection is outbound only.
How much of each rental does the owner keep?
80% of what the renter pays, credited every minute the server is rented. Companies are paid monthly by bank transfer in US dollars, with the invoice issued for them.
Can I still use the server myself?
Yes. You choose when it is on the market: switch it off when you need it and back on when you are done, from the dashboard or automatically through the Provider API.
