Clarifying the Lack of Mac Studio SSD Upgradability
As soon as early purchasers of the Mac Studio received their units, some took the opportunity to tear Apple’s newest Mac apart to see what was inside. Many were stunned to find that the Mac Studio’s SSDs were physically replaceable, only to be disappointed when their Mac Studios wouldn’t boot after swapping SSDs. Some accused Apple of locking the drives in software or even bricking the computers.
As Andrew Cunningham of Ars Technica explains, that isn’t the case. Every SSD has two components: the controller that acts as an interface to the machine and the flash storage itself. Apple now builds the SSD controller into its own chips: the entire M1 family and the T2 security chip used in Intel-based Macs. The SSDs inside the Mac Studio are merely raw storage without a controller. iFixit has since gotten a Mac Studio drive swap to work, which means the storage in the Mac Studio is repairable. If matching Apple’s SSD hardware design is all that’s necessary, we expect that vendors like OWC will eventually offer compatible Mac Studio SSDs. However, there could also be Apple-proprietary chips involved that may prevent such products.
We got two base model Mac Studio units to accept each other’s drives, so storage swaps are possible! Both drives in one unit isn’t working just yet…wish us luck. pic.twitter.com/i2wcbppluf
— iFixit (@iFixit) March 22, 2022
One of the things that annoys me greatly about the right-to-repair movement is how quick many of its proponents are to attribute anything that makes DIY repairs and/or upgrades less convenient to direct, intentional malice. The video cited in the referenced Ars Technica article follows that script to a T.
I sympathize with those who want to hardware tinker and also run macOS. My computer history starts in the mid-'70s, when tinkering was absolutely required. There was a time when I could strip down and reassemble a fully loaded Apple ][+ with my eyes closed. I still have one of the old Apple II security-screw drivers in my toolkit, and a handful of Apple hex keys in a drawer. (I worked a lot on school machines.) I believe in the right to repair as a fundamental principle.
But these people are ignoring one simple fact: This isn’t about them.
No matter how evil one may think Apple is, the bottom line is…well, the bottom line. Apple makes decisions that are intended to support its financial goals. It’s a business. That’s what businesses do.
If Apple thought that thwarting DIY’ers was such an important goal in and of itself, or even just a primary component of making the device secure, you wouldn’t be able to get into the machine at all without at least seriously damaging it.
DIY’ers aren’t the intended buyers of the Mac Studio. They’re not the intended buyers of any Apple devices.
Apple doesn’t make devices with the intent of tinkering. They make devices that are intended to be secure and reliable—a goal that frequently is at odds with tinkering.
There’s no simple way to get everything you might want. (Though getting Apple to license a Mac-like windowing system for a Linux distro might come close for some people.) This doesn’t mean they’re being malicious. It just means that there are more factors involved than mere tinkerability.
Until I bought my iMac in 2018 I have always “tinkered” with my Macs. The prices for hardware upgrades from Apple always were ridiculously bad. Swapping out hard drives or installing new RAM when needed was simple.
It should be much more about the customer. Theoretically, the computers are being made for the customer. Getting all the information we need to operate, maintain, and possibly upgrade an Apple computer shouldn’t feel like pulling teeth. We shouldn’t have to rely on investigative geniuses like Howard Oakley so much for basic things that Apple could have made public. I suspect that half the time that details are labeled “proprietary” by manufacturers (of any type of product), there is no actual concern that those details will be of value to competitors. So proprietary just as often means, “We don’t want to talk about it” - for a host of reasons that probably involve hoping to generate more profit. The profit goal is good but too often it’s made with genuine disrespect to the customer.
The commenters to this article so far (3 comments) seem to miss the point of the architecture of the Mac Studio with regards to storage, and that this follows from the overall architecture of Apple Silicon (SOCs). Apple is making monolithic system on chips (SOCs) that have direct access to not only CPU cores, GPU cores (even shared memory for more bandwidth and faster access for CPU and GPU), the other co-processors, and finally the bulk storage. The storage controllers are built in to the SOC, making storage far closer to the processors in terms of bandwidth and lower latency, offering considerable speed increases over typical SSDs that have bulk NAND storage on the same board as the SSD controllers, AND an interconnect to the system bus (usually PCIe nowadays.)
Are they doing this to thwart user upgrades? I’d say not. Are they doing this because it’s cheaper? I’d guess not, because they have created a custom implementation that does not take advantage of industry standard economy of scale benefits found with generic SSD modules. I say they are for sure doing it this way to get the bulk NAND storage as close to the SOC as possible, increasing bandwidth, decreasing latency, making everything faster. Is the way the Apple Studio doing it with removable NAND storage module less expensive than soldering NAND directly to the logic board? Maybe not, until you speculate that this architecture may be shared with upcoming Mac Pros, and possibly 27" iMac Pros (as Rene Ritchey points out, these may not really be completely dead yet.) Also, because of the larger enclosures of these kinds of machines, there isn’t so much necessity of soldering NAND to the logic board.
Thinking about this even more, this whole architecture is the same with Macbook Air, Mac Mini, even iPad and iPhone, it’s just that we’ve always seen NAND soldered to the logic board, and immediately decried the lack of upgradability. Now that we see a Mac with socketed, removable NAND, everyone jumps to the conclusion that it should be upgradable as easily as swapping a PCIe-based SSD, because that’s the model they know. But now we have seen a physical model for socketed NAND separate from the SOC controller, and everyone has jumped to conclusions that are not accurate.
There’s a “rock and hard place” discussion buried somewhere in here…
Is it in the customer’s best interests to have a high performing, less complex, more reliable, secure platform? One that just works without needing to fiddle around with the innards, compatibility of components, drivers? Apple thinks yes, as evidenced by the the architecture and design decisions that they made (especially with Apple Silicon). The downsides are that there may not be off-the-shelf components that tinkering consumers can swap in and out.
Is it in the consumer’s best interests to have components that enable them to swap out components? Yes as it allows them choice. But the downsides are that performance may not be better, the platform is more complex (T2 chip on Macs and TPM requirement on Windows 11), lower reliability and may not be as secure?
You make the decision which way you prefer. Neither is “good” or “bad”, and it all comes down to what you value.
And everyone has gotten used to thinking of a PC as an Intel/AMD CPU with socketed/removable components. Apple Silicon Macs turned that expectation on its side. It’s hard to break 30-40 years of “this is what a computer should look like”.
It’s also architecturally very similar to the use of the T2 chip on Intel Macs. The T2 is the SSD controller, connecting directly to raw flash modules (of one form or another). And we have seen it perform better than standard NVMe SSD modules.
The reality is Apple is steering more toward locked-down hardware, which is very much in the company’s DNA. Jobs didn’t want anyone monkeying with the Mac.
My frustration is that every other vendor tends to copy whatever Apple does, whether it’s a good idea or not, and even non-Apple hardware is much more locked down than it used to be.
A lot of good points made in this thread, and I think it’s important to remember how many people actually tinkered with their computers, vs never changed anything other than peripherals after the purchase. 5%? Apple likes things to just work.
That being said, I am very glad Apple has the SSDs in sockets vs soldered down. Replacing a whole motherboard (ie the whole computer except for the case basically) is very wasteful environmentally, and replacing only one failed part, even if it requires some crypto re-pairing, is a lot better, cheaper, and faster I would assume.
I would argue that motherboards of the size that have socketed memory, NVMe slots and PCI slots actually need to have components that can be replaced. You have lots of real estate and soldering for those connectors as well as the components the connectors are made of. Not to forget the components on the cards that have to interface on both sides of the card and bus.
It’s very remarkable to see the small size of Mac motherboards these days. It would be an interesting case study on the environmental impact of a small motherboard, how often it’s components fail that necessitate a replacement and overall recyclability vs a more traditional socketed build.
Given my speculation on the costs of a self-contained SSD vs. a controller integrated into the SOC, user KTMGLEN over at Arstechnica says:
"ASIC/chip/SoC designer here. Any time we can pull an external component off the BOM and integrate it into the SoC, it saves a massive amount of money. Replacing a $1 or $2 component with a few cents of silicon area is always worth it. It also comes with better integration, better performance, better reliability, and better security. If you already have the IP or plan to develop the IP, it’s really not even a decision worth thinking about. (note: talking mostly about digital functions here, integrating things like flash, eeprom, data converters isn’t always possible or economical)
“In this specific case, my opinion is that adding a separate normal NVMe or PCIe or SATA drive interface would have increased the design and verification work for the chip since that is not how any of Apple’s existing Ax/Mx chips work. It also would have increased the BOM cost while potentially decreasing performance, security, reliability, etc.”
So I guess we see an alignement between increased performance, encrypted data privacy, AND BOM savings.
I suspect the answer may be more straightforward: With Apple’s penchant for security, it wouldn’t make sense to permit any ssd removed from its original Studio slot to be installed anywhere else where its data could be accessed. I would expect that to include “the other” slot in the same machine or any other machine. Of course, Apple would have some method to make it work but only when the original machine and SSD were brought into an Apple Service Center (presumably a genuine Apple Store or depot) along with proper documentation to prove ownership. Then the “good” SSD could be installed into a working Studio and it would be billed to either Warranty or the customer.
This is the case for the removable SSDs for a Mac Pro or iMac Pro. If you replace the SSD module (or move one from another computer), you can pair it with the computer’s T2 chip using Configurator. And the act of doing so wipes the content, because the encryption keys have all been changed.
But there is no way to move an SSD with its data intact to a new Mac. Even Apple can’t do that with proof of ownership, because it is cryptographically bound to the T2 chip that it was used with.
I fully expect there to be a similar procedure for the Mac Studio. And Apple may even open it up to the public (as they did with the T2-based Macs) in the future.
Apple USED to care about its users, but as you say, they care now care EVEN MORE about their PROFITS.
You are confusing the RIGHT TO REPAIR with Build It Yourself people. I DON’T WANT to build my own Macintosh, but I’d like to do repairs and minor upgrades like replacing a failed power supply (I had to do it on my iMac G5), upgrading the RAM from the minuscule amount (WITHOUT paying the exorbitant Apple prices), adding larger storage, etc.
Yes, right to repair is a different topic than tinkering. There are numerous instances ( see Louis Rossaman’s videos and WSJ’s Joanna Stern’s article ) where independent techs don’t have the specs to enable them to repair machines. In some cases Apple overcharges ( by replacing say a board instead of the minor component that could be replaced ) ( a quote from Stern’s article" Apple said it would cost $999 to fix a MacBook Pro. An independent repair store did it for $325"). Apple also withholds specs from independents even though many of these folks are fully qualified ( Rossaman is one example ) to perform repairs
In addition, Apple makes design decisions that are deliberately intended to make repair difficult. I’m not talking about using special screwdrivers and glue to hold the cases shut, but the fact that they contract with chipmakers for custom (and exclusive) versions of generic chips so nobody else can get them.
Louis Rossmann has mentioned this many times. One of his examples is the USB power management chip. When Apple introduced USB charging to their laptops, they used generic off-the-shelf USB power management chips. Anybody could buy these chips from any supply company (like Mouser or DigiKey).
But today, they replaced the chip with a custom version of the same chip. It’s made by the same company and (as far as anybody can tell) performs the exact same function as the generic one, but the pins are arranged differently, so they’re not interchangeable. And Apple paid for an exclusive distribution contract so nobody else can buy this chip for any reason whatsoever.
The result is that independent repair shops need to scavenge these chips from old Macs, since there is no other way to get them.
I’m sure Apple will claim that there is something proprietary in this chips, but we’re talking about a USB power delivery controller chip, not an M1. If this chip is doing anything different from the dozen or so generic USB PD chips, we’d all love to learn what it is.
I’m not sure if there should be a law against doing this, but it is definitely a sleazy decision that serves no logical purpose other than to make life difficult for independent repair shops.
…and to support higher Apple repair prices because users effectively have no option( in some repair situations ). Such practices often attract anti-trust investigations/violations, although I’m not aware of any FTC/DOJ action ( and haven’t looked for any ).
I get that, but I’d prefer a machine that doesn’t need to be repaired in the first place and that is structurally tight. The current Apple machines are so much better than they were 20 years ago, particularly the laptops.*
The profits thing is a red herring. The extra profits they might make forcing people to get Apple to repair something aren’t even a drop in the bucket overall.
I don’t think Apple cares one way or another about third-party repairability – they prioritize pretty much everything else. That’s largely always been true (I remember getting a special tool to crack open a Mac Plus and having to snip something to make it recognize the RAM upgrade).
*with some exceptions, like the butterfly keyboard.
Join the discussion in the TidBITS Discourse forum