Apple Mac Mini Becomes an AI Machine, Hacked to Use Nvidia GPU

Apple’s Mac Mini, which has traditionally not supported external GPUs, has now been paired with an Nvidia GPU by startup TinyCorp. The Mac Mini generally does not support external graphics cards (GPUs). However, the startup named TinyCorp has combined the Mac Mini with an Nvidia GPU, making this compact device far more powerful. TinyCorp has successfully connected the Mac Mini to an Nvidia GPU, likely the GeForce RTX 5060 from the Galax brand. This step is quite significant because Apple devices, including the Mac Mini, are known for their limitations in supporting external GPUs (eGPUs), especially compared to Windows devices. To make the GPU recognised as an internal device, TinyCorp used an ADT-Link adapter that converts the Thunderbolt 4 connection to a PCIe pathway, which is compatible with the GPU. Since macOS does not provide official support for Nvidia GPUs, TinyCorp developed custom software drivers based on Python at the userspace level, allowing the GPU to be used for computation. This approach was taken because macOS does not directly support third-party GPUs like those from Nvidia. The combination of Mac Mini with eGPU is, according to TinyCorp, currently limited to modern consumer-segment GPUs. TinyCorp’s solution opens up new opportunities for the Mac Mini, which was previously limited, to now run larger and more complex AI models locally. Moreover, the popularity of the Mac Mini has recently increased among AI developers, driven by the phenomenon of AI agents like OpenClaw and the surging need for local computation for AI models. The eGPU concept is not new, either on Windows or macOS. However, with the rising trend of edge AI (AI processing directly on devices without the cloud), the need for stronger local computing has also increased. The device will later be equipped with a power management system, allowing the GPU to be turned off when not in use to save energy, as well as a reset feature if errors occur.