Apple Approves Nvidia eGPU Driver for Arm Macs — A Landmark Shift for AI Developers

Apple Finally Allows Nvidia GPUs on Apple Silicon

Apple has approved a third-party driver that enables Nvidia external GPUs (eGPUs) to work with Arm-based Macs — a development once thought impossible.

The Details

The driver, developed by Tiny Corp (the company behind tinygrad), has received Apple's official signing approval. This means users no longer need to disable System Integrity Protection (SIP) to use Nvidia hardware with their Macs.

Key points:

• Not from Nvidia directly — Tiny Corp built the driver independently
• Requires Docker compilation — This isn't plug-and-play yet
• Designed for LLM workloads — The primary use case is running large language models locally
• SIP remains intact — Apple signed the kernel extension, marking a significant policy shift

Why This Matters

Apple's Arm-based chips (M1-M4) use a different architecture than traditional x86 systems, making Nvidia GPU support extremely challenging. This approval opens several possibilities:

1. Local LLM inference — Developers can now leverage Nvidia's CUDA ecosystem on Mac hardware
2. Cost-effective AI development — Combining Mac productivity with Nvidia's GPU performance
3. tinygrad ecosystem growth — Tiny Corp's driver validates their framework's approach to hardware abstraction

Limitations

The driver is still in early stages — users must compile it via Docker, and the setup process requires technical expertise. This is primarily aimed at developers and AI researchers, not casual users.

Industry Impact

This move signals Apple's growing recognition that AI developers need access to Nvidia's GPU ecosystem. While Apple's own Metal framework and the M-series Neural Engine are powerful, the CUDA ecosystem remains dominant in machine learning research.

Sources: The Verge, Tiny Corp (tinygrad)