Apple has approved a third-party driver from Tiny Corp that lets Apple Silicon Macs recognize external AMD and Nvidia GPUs for compute workloads, a notable change after years without Apple-supported eGPU functionality on ARM-based machines. Reported by multiple outlets as approved on March 31, the driver from George Hotz’s startup is designed for compute tasks such as AI workloads, not for gaming, display output, or traditional graphics acceleration. The move cracks open a long-sealed door in Apple’s hardware ecosystem, though the opening is narrower than many users might hope.
What is verified so far
The core facts are consistent across multiple reports. Apple greenlit the TinyGPU driver on March 31, making it possible for Apple Silicon Macs to recognize and use external Nvidia and AMD graphics cards over Thunderbolt connections. The driver is Tiny Corp’s own creation, not an official product from Nvidia or AMD. George Hotz, the well-known hacker and founder of Tiny Corp, developed the software, and Apple’s approval through its standard driver-signing process gave it legitimacy on macOS. This was widely framed in discussion as effectively ending a years-long lack of eGPU support on Apple Silicon Macs, though Apple has not publicly described it as a policy change.
The driver enables compute-only access. That distinction matters. Users can offload machine learning inference and training tasks to an external GPU, but they cannot use it for screen rendering, gaming, or traditional graphic acceleration. AppleInsider confirmed that while AMD or Nvidia eGPUs can now work on Apple Silicon Macs, graphic acceleration is explicitly excluded from the driver’s capabilities.
Early testing has already produced tangible results. Apple Silicon Macs running the TinyGPU driver have successfully run AI models using external Nvidia and AMD hardware. These tests demonstrate that the driver functions as advertised for its intended purpose: accelerating AI computation rather than rendering frames in a game or driving a display output.
The software’s orientation toward AI is deliberate. Tiny Corp has positioned itself as a company focused on making GPU compute accessible, and its driver reflects that priority. Tom’s Hardware noted the software is designed for AI and not built for gaming, a framing that sets clear expectations for anyone hoping to plug in an RTX card and play titles at high frame rates on a MacBook Pro.
What remains uncertain
Several significant questions remain unanswered. Apple has not issued any public statement confirming the driver approval or explaining why it chose to allow this particular software after years of blocking eGPU support on ARM Macs. The approval came through Apple’s standard code-signing process, which does not require a formal announcement, but the silence leaves open questions about whether this reflects a broader policy shift or a narrow exception for compute-focused tools.
Neither Nvidia nor AMD has commented on compatibility or future support. It is unclear whether either company plans to release its own official drivers for macOS on Apple Silicon, or whether Tiny Corp’s solution will remain the only path for connecting their hardware to modern Macs. The absence of institutional backing from the GPU manufacturers means users are relying entirely on a startup’s software for a workflow that could become central to professional AI development on Apple hardware.
Performance benchmarks are also thin. While reports confirm that AI models have run successfully on external GPUs connected to Apple Silicon Macs, no detailed comparison data exists showing how this setup performs against native Apple GPU compute, against the same cards running on Linux or Windows, or against cloud-based alternatives. Without that data, it is difficult to assess whether the TinyGPU driver offers a meaningful performance advantage or simply provides a functional but suboptimal bridge.
The long-term durability of this arrangement is another open question. As with other third-party, signed kernel-level components, a future macOS update could potentially change what continues to work, or introduce restrictions that limit third-party GPU access. Tiny Corp’s driver works today, but there is no public commitment from Apple ensuring it will continue to work tomorrow. Developers building workflows around external GPU compute on Macs are, for now, operating without that guarantee.
How to read the evidence
The strongest evidence here comes from the driver’s actual approval and the confirmed test results. The fact that Apple signed the driver through its official process is a concrete, verifiable action. The successful AI model runs on external GPUs represent real-world validation of the software’s functionality. These are not speculative claims or leaked roadmaps; they are observable outcomes that anyone with the right hardware can, in principle, reproduce.
Most of the coverage, however, is interpretive rather than primary. No official Apple documentation, press release, or developer blog post accompanies this approval. The reporting relies on Tiny Corp’s own communications, forum discussions, and tech press analysis. The Verge clarified that the driver belongs to Tiny Corp rather than Nvidia, an important distinction that prevents readers from assuming official Nvidia support exists for macOS on Apple Silicon.
The framing around this approval also deserves scrutiny. Several outlets have described it as Apple “breaking a barrier” or ending a ban, language that implies Apple actively chose to open its platform. The reality may be less dramatic. Apple’s driver-signing process is not an endorsement; it is a security verification step. Tiny Corp submitted a driver, it met Apple’s technical requirements, and Apple signed it. Reading this as a strategic decision to welcome Nvidia back into the Mac ecosystem requires evidence that does not yet exist in the public record.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
