Your PC can help build AI, and earn while it does.
Install the HyperCluster software and choose how much of your graphics card, AI chip and disk to share. The network uses it to train and run other members’ AI, read pictures for apps that use HyperAI, and store files. The work it really does earns HYPEC.
For Windows PCs with an NVIDIA graphics card, and Linux. Members’ AI jobs need an NVIDIA RTX 3060 or newer with at least 8 GB of memory. Download it from your Passport, under Devices.

Before you start
A quick check before you install, so that your PC can take training jobs as well as storage.
- Windows with an NVIDIA graphics card, or Linux
- An NVIDIA graphics card: RTX 3060 or newer, with at least 8 GB of memoryMembers’ AI jobs (training, and running their models) need one. Laptop cards count too, if they have 8 GB or more. So do workstation cards from the RTX A4000 and data-centre cards (A100, L40, H100…). Not the GTX 10 or 16 series, the RTX 20 series (the 2080 Ti too), the RTX 3050 or the RTX A2000.
- Virtualization switched on (Windows)Training runs in Linux inside Windows, which needs it. Open Task Manager → Performance → CPU: it should say “Virtualization: Enabled”. If it doesn’t, restart the PC, open its BIOS (usually Del or F2 while it starts), switch on “SVM Mode” (AMD) or “Intel Virtualization Technology” (Intel), then save and restart. The software tells you too, when you install it.
- Windows Subsystem for Linux: set up for youThe software switches on the Windows Subsystem for Linux (WSL) and sets up its own Linux and Docker in it, so there is nothing else to install. If Windows needs a restart to finish, training sets itself up after the next one.
- Signed in to Windows for trainingTraining jobs run in the Windows account of whoever is signed in, so they run while someone is signed in.
How it works
- 1
Install the software
Download it from your Passport, under Devices. It finds your disk, AI chip and graphics card.
- 2
Link it to your Passport
Open the file you downloaded with it, or type your HyperCluster ID. Adding many PCs? One file links them all.
- 3
Choose your share
For example half of the graphics card and a fifth of the free disk. Change it any time.
What your share is used for
HyperAI
AI chips look at pictures for apps that use the network.
LiveTraining and running AI
Members train their own AI on a shared graphics card, and run it, each job locked inside a container.
LiveStorage
Your spare disk keeps copies of files for the network. They are encrypted, so you can’t read them.
LiveDevelopers
Apps call the network’s AI through an API, and pay for it by the picture or by the second.
LiveWhat you get
Every day, the work your PC really did is checked and counted: photos looked at, training and AI run, files kept. Each kind of work earns HYPEC at the rate HyperCluster sets for it, and the day’s work is written to our blockchain with its proof, where you can see it.
Graphics-card jobs: a payout rate for each card
When your graphics card runs a member’s training job, a run of their own model or a chat with a chat model they trained, it earns its payout rate for each hour the job runs, counted by the second. Each graphics card has its own payout rate. Members pay 20 % more than the payout rate, and that 20 % is HyperCluster’s share, so you get about 83 % of what the member pays.
Payout rates for common graphics cards| Graphics card | $ an hour | HYPEC an hour |
|---|---|---|
| RTX 3060 · 12 GB | $0.057 | 5.7 |
| RTX 4070 · 12 GB | $0.119 | 11.9 |
| RTX 3090 · 24 GB | $0.177 | 17.7 |
| RTX 4090 · 24 GB | $0.408 | 40.8 |
| A100 PCIE · 40 GB | $0.488 | 48.8 |
| RTX 5090 · 32 GB | $0.550 | 55.0 |
Your PC is checked first: a short automatic test before it gets any work, and spot checks after that. Work that fails a check earns nothing.