Who needs a modified Mac Pro when you can turn multiple Mac Studios into one integrated computing system? With the upcoming macOS Tahoe 26.2 release, Apple is introducing a new low-latency feature that lets you connect multiple Macs together using Thunderbolt 5. For developers and researchers, this is a potentially useful way to build powerful AI supercomputers that can run local models at large scales. This allows four Mac Studios, each running up to 512GB of integrated memory, to run the 1 trillion parameter km-k2-thinking model far more efficiently than a PC with a power-hungry GPU.
While we’ve seen Thunderbolt Mac clusters before, they were limited by slow Thunderbolt speeds, especially if they required a hub (which could reduce speeds to 10 GB/s). Apple’s new feature allows for full Thunderbolt 5 connectivity at up to 80Gb/s. The clustering capability isn’t limited to the expensive Mac Studio, it will also work with the M4 Pro Mac mini and M4 Pro/Max MacBook Pro. Developers won’t need any special hardware to create a cluster, just a standard Thunderbolt 5 cable and a compatible Mac.
In one demo, I watched a group of four Mac studios load and run that huge KM-K2-thinking model in an early version of ExoLabs’ EXO 1.0. Notably, the cluster uses less than 500 watts of power, which is about 10 times less than a typical GPU cluster (NVIDIA’s RTX 5090 is rated for 575W, but it can be even more demanding).
macOS Tahoe 26.2 will also give Apple’s open source MLX project full access to the neural accelerator on the M5 chip, which should dramatically speed up AI inference. However, ironically, the only M5 Mac available today – the 14-inch MacBook Pro – only supports Thunderbolt 4. This means it won’t be able to take advantage of the new Mac clustering capability.
The integrated memory and low power design of Apple silicon already make the Mac a useful choice for demanding AI work, but the ability to cluster multiple systems together over Thunderbolt 5 is potentially even more attractive to anyone working with larger models. Of course, the Mac Studio with 512GB of RAM isn’t cheap — it starts at $9,499 with the M3 Ultra chip — but that’s only the highest-end option. Labs and companies that already have a Mac Studio, Mac Mini, and MacBook Pro can potentially purchase the cluster system already in place.