Nvidia released a tool that connects your computers into a home AI! It's free and supports macOS too

Do you have a gaming PC with an Nvidia graphics card at home, plus a MacBook, and maybe another machine just sitting there? Nvidia PAIR is a free tool that turns them into a single computational unit for running artificial intelligence — without a single extra cable. The software has been released in a beta version for Windows, Linux, and macOS.

How it works

The principle is simple. PAIR automatically finds compatible machines on your network, groups them together, and then offers applications a single access point. When tasks start to accumulate, it distributes them among devices that are currently idle — regardless of the operating system they are running. From a software perspective, the tool works with popular extensions like Ollama and LM Studio, so you don’t have to change your habits.

One important detail that Nvidia admits

The documentation contains a sentence that is good to read before you start dreaming of a home supercomputer. PAIR does not combine graphics cards into a single virtual one — individual machines remain separate and process tasks concurrently. So, you won’t get more memory for one large model, but the ability to process more things at once.

What you need

The requirements are surprisingly mild. Any GeForce RTX graphics card from the 20 series upwards, a DGX Spark system, or a Mac with an M4 chip or newer, along with 8 GB of RAM and a recommended 20 GB of free space, will suffice. Internet is only needed to download models; the operation itself does not require it.

Privacy is the main selling point. Prompts, files, and agent context do not leave your home network and are not sent anywhere — which is an increasingly valuable feature for tools of this type. The version is currently labeled as beta 0.1.1, so expect minor glitches.

Do you run artificial intelligence locally, or do cloud services suit you?

Source: Nvidia