Yes, I will absolutely trust them again, despite their abandoning me last time. Totally.
Apple returning to the server market would be quite the plot twist: after years of people building unofficial Mac Studio AI clusters, Apple may simply productize the idea. Dual or quad M8 Ultra systems, potentially connected via Nvidia NVLink Fusion, could make local AI infrastructure far more mainstream. The Mac Pro might finally have found its next job.
Well maybe this will make hosting CI runners for Mac and iOS development less of a joke.
They probably saw an opportunity after data centers started buying up all their Mac mini’s and Mac Studios.
I can’t imagine anyone running server farms is particularly hyped about this
Didn’t a load of them get burned pretty badly by apple killing off the last server line and Steve Jobs just told them to cry more when challenged about it
They already did in 2024
Why would anyone ever buy a server from apple? Isn’t it a consumer brand? Feels like opposite of “enterprise friendly priced hardware”
Isn’t it a consumer brand?
Trust me, consumers aren’t the ones buying 5k€ displays, nor the nearly 20k€ maxed out Mac Studios.
Feels like opposite of “enterprise friendly priced hardware”
We don’t know what the pricing will be.
Something to consider: All those nVidia GPU servers have discrete GPUs, meaning they need to have both RAM and VRAM, and any time there’s need to transfer something between RAM and VRAM, that’s overhead. Apple runs everything in a shared pool of memory - which at present is significantly slower than the VRAM on those nVidia GPUs since it’s DDR rather than HBM, but if they do what they did with the Ultra line of chips and go even further, e.g glue together 8 chips instead of 2, they might make up a bit of the difference in memory bandwidth. Or they could add HBM to these chips I guess.
And a single nVidia DGX B300 with 2.1 TB VRAM is several hundred thousand. And a single one of those is not enough to run the biggest models. By offering less powerful chips and cheaper memory, they might theoretically be able to offer tons of memory and acceptable, though lower, performance for much less money. Allows security-conscious companies to run high-end LLMs locally. And since they deal in shared memory, you might be able to use CPU instead of GPU if that’s more efficient in some specific part of inference, without transferring things between different memory spaces.
Well OpenAI allegedly bought a boatload of mac minis. And I dont think it was for brushing up their resumes and watching porn.
So, building Codex for minis?
The macOS license agreement specifies you may only run it on Apple hardware. That means Xserve. So if you’re developing for macOS or iOS and want to use Xcode and target Mac platforms, you need this.
Or if you want to run server-size virtualization. You can install ESXi or another hypervisor on an Xserve and be compliant.
Apple hardware is one of the best platforms for low scale AI workloads. A simple M4 Pro MacBook Pro with 64GB RAM can comfortably run the best “at-home” model currently available (Qwen3-Coder-Next) around 30-40tkps.
A comparably specced AMD Ryzen AI 395+ can do about half of that.
If Apple delivered that in server-scale, easily scalable packages, there would definitely be buyers.
Server, enterprise
Mac mini
I don’t know man, anything that becomes a full-sized server/hardware has different pricing than mini-PCs. We are talking about a collaboration with Nvidia, a company that manages to overcharge everyone.
Of course the pricing will be different. My point here was that if Apple can jam that much compute into something as small as a Mac Mini, imagine what they can deliver at 4-5U scale.
Apple and Nvidia hardware combined might just be the solution to more scalable AI compute that uses less power.
i am assuming the performance comes from the soldered components improving signal integrity? the same reason the framework desktop has soldered memory? if true, then there would not be much customization, i imagine.
Introducing 1 the Apple Xtra Xtreme Xserver, exclusively with subscriptions 2 available in Pro, Pro Plus, Xecutive One and Pro Xecutive One Plus! (wheels $799.99 extra 3)
1 purchase of hardware does not imply ownership
2 subscribing to service also agrees all data stored on or passing through will be used to train AI
3 because fuck you that’s whyedit: cry more Apple fanboys
