Reserve Now Portal
Introducing the maccloud Studio

Your own AI.
Nothing shared.

Mac Studio with M5 Ultra, dedicated entirely to you — full GPU and Neural Engine for your own local LLMs.

Choose your memory Learn more ↓
36-core
CPU
80-core
GPU
32-core
Neural Engine
M5 Ultra chip
4TB storage
Every configuration runs the same M5 Ultra — not Max, not a previous generation. Only unified memory changes between plans.
Why M5 Ultra?

One chip. No neighbors.

Every machine is single-tenant — the full chip, the full memory bandwidth, and the full Neural Engine are yours alone, whether you're serving a 70B model or fine-tuning your own.

Always M5 Ultra — never Max, never a previous generation. We standardize on Apple's newest, most capable chip because unified memory capacity and bandwidth are what actually determine what you can run — not which chip tier you happened to get. Every maccloud plan runs the identical M5 Ultra; only the memory changes.
One dedicated node, apart from shared nodes
Included with every unit

Real access. Not a black box.

You get root-level SSH, an out-of-band console, remote power reboot, and switched PDU control on every machine — manage it the way you'd manage your own rack, not the way a locked-down managed platform lets you.

SSH access
Out-of-band console access
Remote power reboot
Switched PDU power control
SSH, console, and remote power access
Fit check

Is maccloud right for you?

Good fit if you…
Run local LLMs (7B–400B+) or fine-tune your own models and need the full chip and memory bandwidth to yourself
Need more unified memory than any off-the-shelf cloud Mac offers — AWS's largest EC2 Mac instance tops out at 32GB
Want real SSH, console, and power control — not a locked-down managed API
Can plan around a 16–24 week launch window in exchange for locking in pre-launch pricing today
Probably not if you…
Need compute running today — the launch window won't fit an urgent deadline
Your workload is iOS/macOS CI, Xcode builds, or general Mac hosting rather than AI inference
Your models comfortably fit in 16–32GB and you don't need the extra headroom yet
You want a fully managed inference API with no server administration at all
Configure

Choose your memory.

Pre-launch pricing — reserve before general availability
Mac Studio · 256GB · M5 Ultra
Estimated launch: 16–18 weeks
$1,499/mo retail$1,200/mo pre-launch
≈$1.64/hr equivalent
You won't be billed until your Mac Studio launches — reserving just locks in today's pre-launch price.
Hourly figures are a 730-hr/month equivalent, shown for comparison against hourly-billed cloud instances.