AMD and Nvidia eGPU drivers for Mac have finally received Apple’s official approval, marking the first time third-party graphics processors can run legitimately on Apple Silicon without hacking around system security. Tiny Corp, the company behind the driver, announced the approval around April 1, 2026. But before you dust off your gaming rig, understand this: these drivers are built for artificial intelligence researchers and compute tasks, not for the gaming and graphics acceleration most users expect from external GPUs.
Key Takeaways
- Tiny Corp’s drivers for AMD RDNA3+ and Nvidia Ampere+ GPUs now work on Apple Silicon Macs via Thunderbolt 3/4 without disabling security features
- Drivers support AI and compute workloads through TinyCorp runtime libraries, not graphics acceleration or gaming
- Requires macOS 12.1 or newer, a compatible GPU in a powered enclosure, and available USB4 or Thunderbolt port
- Apple Silicon eGPUs remain officially unsupported for graphics; only Intel Macs have native eGPU graphics support
- Thunderbolt bandwidth limitations throttle performance compared to desktop GPU setups
What Changed With Apple’s Approval
Previously, running Nvidia or AMD eGPUs on Apple Silicon required users to disable System Integrity Protection (SIP), Apple’s core security feature that prevents unauthorized code from modifying the system. That workaround was clunky and risky. Tiny Corp’s signed driver eliminates the need for that hack entirely, letting the GPU run with full system-wide access without compromising security. This is genuinely new territory—Nvidia has historically resisted supporting modern Macs, and AMD’s official eGPU support remains limited to Intel machines.
The approval itself is surprising given Apple’s walled-garden approach. Yet the company clearly sees value in enabling AI compute on Apple Silicon without forcing researchers to buy expensive external hardware or rely on hacked drivers. Apple’s own MLX framework handles machine learning on native Apple Silicon, but it lacks integration with third-party GPUs. This driver fills that gap for researchers who need extra compute muscle.
AMD and Nvidia eGPU Drivers for Mac: AI Only, No Games
Here’s the critical limitation: these drivers are not built for graphics acceleration or gaming. You cannot use them to play games at higher frame rates or better graphics settings. You cannot use full Nvidia libraries or tools. The drivers work exclusively with TinyCorp’s runtime libraries, which means you’re locked into AI and compute workflows—primarily large language models and similar workloads. If you were hoping to finally play AAA games on your Mac with an external Nvidia RTX card, this is not your solution.
Apple’s official stance on eGPUs for graphics remains unchanged: only Intel Macs with Thunderbolt 3 support graphics acceleration through external AMD cards, specifically the Radeon RX 6800, 6800 XT, and 6900 XT models in compatible enclosures like the Sonnet eGFX Breakaway Box or Razer Core X. Apple Silicon Macs have no official graphics eGPU support whatsoever. Tiny Corp’s drivers sidestep this limitation for compute, but they do not enable graphics acceleration on Apple Silicon.
Hardware Requirements and Setup
Getting this working requires specific hardware. You need an Apple Silicon Mac running macOS 12.1 or newer, a compatible GPU (minimum AMD RDNA3+ or Nvidia Ampere+), and an eGPU enclosure with adequate power delivery. The connection runs through USB4 or Thunderbolt 3/4 ports. Thunderbolt bandwidth becomes a bottleneck here—these connections max out at 40 Gbps, which is substantially slower than a desktop PCIe 4.0 slot. Performance will be noticeably throttled compared to internal GPUs or desktop setups.
Setup involves connecting the powered enclosure, installing the TinyCorp driver, and configuring your AI runtime to use the external GPU. TinyCorp provides documentation through its TinyGPU project for detailed installation steps. This is not a plug-and-play experience like plugging in a USB device—you are working with developer-level tools and expecting some technical configuration.
Why This Matters, and Why It Falls Short
For AI researchers and machine learning engineers working on Apple Silicon Macs, this is meaningful progress. You can now accelerate compute-heavy workloads without buying a separate Linux machine or relying on cloud services. The security improvement—no more disabling SIP—is also significant.
But for the broader Mac user base, this driver solves almost nothing. Gaming remains impossible. Graphics acceleration remains unavailable. The Thunderbolt bandwidth bottleneck means even AI workloads will be slower than a desktop setup. And Apple’s refusal to officially support graphics eGPUs on Apple Silicon means this remains a workaround, not a feature Apple stands behind. Future macOS updates could break compatibility or restrict unsigned drivers further, leaving users stranded.
Compared to Apple’s native MLX framework for on-device AI, these drivers add external compute flexibility but sacrifice the simplicity and guaranteed compatibility of native solutions. For users who need that extra power, they represent a rare concession from Apple. For everyone else, they are a reminder that Apple’s ecosystem still has significant constraints.
Can I use these drivers for gaming on my Mac?
No. These AMD and Nvidia eGPU drivers for Mac are designed exclusively for AI and compute tasks through TinyCorp’s runtime libraries. They provide no graphics acceleration or gaming support. Apple Silicon Macs have no official graphics eGPU support at all, and these drivers do not change that limitation.
What GPUs work with the new drivers?
Minimum requirements are AMD RDNA3+ models and Nvidia Ampere+ models. Older cards are not supported. You will also need a powered eGPU enclosure with adequate power delivery and a Thunderbolt 3, Thunderbolt 4, or USB4 port on your Mac.
Do I need to disable System Integrity Protection anymore?
No. The major advantage of Apple’s approval is that Tiny Corp’s signed driver no longer requires disabling SIP. You can run the driver with full security features enabled, eliminating the previous workaround that made many users uncomfortable.
Apple’s approval of AMD and Nvidia eGPU drivers for Mac is a small but meaningful crack in the company’s walled garden—specifically for researchers and AI engineers who need external compute. For everyone else, it remains a niche solution that does not solve the gaming or graphics acceleration problems Mac users have faced for years. The real story here is not what these drivers enable, but what they still refuse to allow.
Edited by the All Things Geek team.
Source: Tom's Hardware
