I do agree they’re built to order but I bet Apple sees the demand for this specific server with 512GB of RAM and they could build a lot ready to throw in the SSD chips. For AI alone, they should sell well compared to prior models.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Mac Studio With Massive 512GB RAM Facing Two-Week Delivery Delay
- Thread starter MacRumors
- Start date
- Sort by reaction score
This is a niche machine. It's no wonder that not many people will find a use for it. 95% of people will find a better deal with the Mac Mini and Mac Studio with the M4 MAX.
I would really like someone to test this 😄
It's only $9,500 for 512 GB RAM and a 1 TB SSD (for reference, that's the same price as the Mac 512k, when adjusting for inflation). Don't max out the SSD (use external storage) and that gives people an extra $4,000 to spend on gas for that Lambo.
Last edited:
Those same professionals would buy multiple Mac minis and rackmount them. They work in parallel, not serial. They don't put all their money in one machine. A $14k computer is excessive for anyone.
Its the hottest prosumer AI/LLM computer right now. An equivalent GPU with 512GB VRAM costs 5 figures
From what I've read here apparently science, math, and lot and lot of open safari tabs. Or just a few Chrome tabs.
Not super surprising - the top-spec M4 Pro Mac mini's had a month or two delay at launch and have, to my knowledge, never been available for immediate delivery.
Last edited:
Agreed. Plus, since deepseek R1 is mixture of experts, the 810GB/S bandwidth isn't THAT huge of a bottleneck.
High end configs have always had some delivery times…
Well, that is some exaggeration, isn’t it?
It is still a niche item... and sure maybe you could argue 512gb of ram is going to attract folks, is that with a 1 tb, 2 tb, 4tb, 8 tb or 16 tb storage? Makes zero sense to have that kind of money tied up in inventory waiting for a buyer when there are 5 options on storage.
Money can buy an ultra-fast Mac, but it cannot buy ultra-fast delivery, apparently.
Apple's new Mac Studio with M4 Max and M3 Ultra chip options began arriving to customers today, but the highest-end model is facing delays.
If you order a maxed-out Mac Studio with a 32-core M3 Ultra chip, 512GB of unified RAM, and 16TB of SSD storage on Apple's online store today, the mind-boggling $14,099 configuration will not arrive until late March. Unsurprisingly, it is the massive 512GB of RAM upgrade option that seems to be causing the longer wait here.
If ordered today, that configuration is estimated for delivery between March 26 and March 31, so there is currently a two- to three-week wait.
Thanks, Paul!
Article Link: Mac Studio With Massive 512GB RAM Facing Two-Week Delivery Delay
This is a nothing burger. ANY BTO Mac Studio ultra is facing a week to week and a half delay right from the start. So shocker for the most expensive one, there is another whole week delay.
Do wonder just how many folks are affected by this. I bet it's fewer than the number that own an AVP, and we all know how that of repeated bad joke goes about the 2 people that own them.
I really feel like apple saw the LLM wave coming as part of their choice to make M chips have unified memory. Its a massive advantage now
For anyone or for you? I could absolutely use the maxed out Mac Studio for my neuroscience research (although I'd save money and not max out the SSD -- I'm not sure why so many people are fixated on the $14,000 -- you can buy it with 512 GB of RAM for less than $10,000). I have deep learning processes that can use up to 64 GB of RAM per CPU core (a 32 core machine with 512 GB of RAM is 16 GB per core) -- see the attached screenshot. Also, the fact that this is high speed shared RAM with the GPU is a major benefit to some people, especially if they want to run or develop large LLMs locally.
I do most of my work on a high performance cluster, but being able to run some things locally would be easier much of the time. So no, it would not be excessive for me. I'm not buying one because I have cheap access to my university's high performance cluster, but otherwise, I'd be strongly considering this. Although, "cheap" would still cost me about $13,000 over 5 years for similar specs to the fully maxed out Studio Pro (less dedicated RAM, but ability to access a lot more; that would also be much less VRAM).
Plus, with the education discount, it’s only $9000 for the top end machine with a 2 TB SSD and 512 GB of RAM. That’s a lot of money but for a small, quiet computer with tons of power and RAM that also works as VRAM? Quite reasonable.
Attachments
Last edited:
For a small-medium business, and even for many self-employed pros, $14K in assets is peanuts.
Definitely not people who can afford a Lambo in most cases. I could get one for my company and I drive a Fiat Panda. Still an Italian car though
This is the Studio version of the $1000 monitor stand or those ridiculous over priced wheels from years past. Apple knows nobody needs this, but they also know a group of dopes with too much money are going to buy it and claim they need it. As an Apple stockholder, I fully support this strategy.
Other than LLMs, what does one need 512gb of ram for?!
I just pulled up the remote desktop on a server here at work that handles pharmacy operations for several hospital campuses, and it's only 64gb.
I just pulled up the remote desktop on a server here at work that handles pharmacy operations for several hospital campuses, and it's only 64gb.
Last edited:
Spot-on. Apple is a master at keeping JIT (just in time) fulfillment running smoothly, wanting to keep inventories at a minimum.
...well I thought it was funny
People here were telling me no one would buy high end Mac’s with lots of memory.
They have no idea how big the personal LLM development market is..
They have no idea how big the personal LLM development market is..
I just want to try one that has that much ram, I don't even know why. I have a M1 Max MBP with 32gb and I've never run into a ram limitation.
There are those that need a lot of RAM. There are those who need a lot of processing. There are even those who need both. And then there are those who need neither.