Apple just dropped new Mac mini and Mac Studio models with fresh chips—the M6 (Apple's first 2nm processor for Macs) and the M5 Ultra (their most powerful chip yet). On paper, it's just a specs bump. No revolutionary features. But what's telling is how hard Apple is pushing these machines for local AI inference and development work. The company clearly sees this as the future for these desktops, even though it wasn't part of the original design philosophy. The shift happened after macOS 26.2 rolled out last December. That update unlocked low-latency Thunderbolt 5 communication between machines, which opened the door for distributed AI inference using MLX—an open source framework that lets the M-series chips' unified memory architecture do heavy lifting for machine learning. Before that, these Macs weren't really positioned as AI workhorses. Now hobbyists and professional researchers are literally chaining multiple Mac minis or Studios together to run local large language models that would be way too big for any single consumer device. This setup gives people a real alternative to buying specialized hardware loaded with Nvidia GPUs. You can build a distributed system from off-the-shelf Macs and get serious inference performance without dropping six figures on enterprise-grade accelerators. That's why these machines have become surprisingly popular in ML circles. Apple's new refresh isn't about flashy innovations—it's about acknowledging what developers have already figured out and giving them better silicon to do it with.