Costs of Non-Upgradeable RAM & SSDs

Interesting video discussing profit-margins and non-upgradeable RAM and SSDs:

The point in the video about the economics of upgrading over time is a good one. I’m happy with my M1 MacBook Air, but I know that I am going to need another general purpose laptop soon, and it will be a Windows laptop primarily because I don’t want to spend what Apple is asking for the amount of RAM and storage that I would like. I can buy a decent Windows laptop for well under $1k today and upgrade it to much higher RAM and storage in a couple of years at a very reasonable price. We used to be able to do that with Macs like my beloved mid-2012 MBP, but not any more.


This is the logical outcome of “thinner and lighter” (which along with “no wires” represent the Apple DNA that Steve Jobs and Jony Ives were striving for). I first ran into it when I purchased a 2014 21-inch iMac for video streaming production in a church environment. It needed to be simple, fast enough to handle codecs on the fly, and have enough storage capacity to capture archival video and stock pieces. Because the 21-inch model did not have the nifty RAM door and slots, it meant we had to purchase RAM and storage with the computer at Apple prices. It wasn’t nearly as attractive as the “starting price” for the base model.

Since then, thinner and lighter has meant soldering and integrating parts that would be discrete and replaceable in thicker and heavier devices.

I picked up my old Tom Bihn backpack last week and was stunned to find it weighed over 20 pounds. Inside was my 2012 15-inch MBPro, which I upgraded at least once with an SSD and larger RAM. By itself it weighed close to 3 pounds, and I thought nothing of it when I was using it. Now it feels bloated compared with the 2017 13-inch MBPro that replaced it. But I can’t upgrade anything on that machine, and am thankful that I insisted my employer at the time overspec the processor and RAM when it was purchased.

1 Like

There’s thinner and lighter, yes, but those aren’t the only benefits of soldering components on the logicboard, or integrating them onto the SoC. It lowers costs and improves reliability, and putting the RAM on the SoC is part of the secret sauce of Apple Silicon. You cannot do that and preserve upgradeability. The design has additional benefits that Apple-in-their-infinite-wisdom doesn’t bother to explain to us mere mortals, or their competitors, but which I’m certain is well researched and understood within Apple’s hardware teams.

So, yes, we lose some upgradeability for our devices. That means we pay Apple’s prices for storage and RAM, and yes, it’s more expensive. But it’s not without its upside, and it’s worth keeping that in mind when deciding whether to stick with Macs or not. (To say nothing of the cost and effort of switching.)

Tech product design involves trade-offs. If the trade-offs that Apple chooses to make aren’t to your taste or preference, that sucks, because there’s no alternative. (Other than to switch platforms, which is a zero probability for many (most?) of us here.) But I don’t think it’s fair to presents Apple’s decisions as without benefit, other than to Apple’s bottom line.


Well, yes. And, “thinner, lighter, wireless” are all huge benefits, which I had hoped I made clear in my comment. An iPhone or a MBPro whose components do not wriggle loose and fail is more reliable than a device that falls apart because of an improperly spec’d socket.

The OP sees only the costs, and compares that unfavorably with a platform that can have new parts swapped in. I would argue that the “decent Windows laptop” that needs more RAM and storage after a couple of years was probably under-spec’d in the first place.

His main point—that current Macs are pretty much as-is machines out of the box—is true, and that’s what my response reflected. But what neither you nor I has said is that the same 2017 MBPro I used as an example is not only still useful, but feels like a current model. My most recent workplace went through three different Windows machines over the two years I was there, trying to keep up with their weekly streaming demand. They all felt fine out of the box, but could not keep up with the task.

I think it’s healthy to churn through this upgradability vs. reliability/performance discussion periodically. The very fact a Mac will still be churning away 10 or 15 years after initial purchase means that many of us face replacement decisions infrequently. The first Mac I actually owned, an SE I purchased in 1988, was in daily use until 1994, and I had done everything I possibly could to that “closed” machine including adding a Mobius portrait monitor and internal display card to it. The first Mac laser printer I owned, a LaserWriter Plus that I purchased used in 1989, lasted nearly 10 years and it endured a motherboard replacement that effectively sped it up and doubled its vertical DPI density.

I did not want to ever replace either of those machines. I got full value from them. I paid for new Apple hardware (a Quadra 640, as I recall) and it was a quantum leap. Each new machine over these 35 or so years has been a similar experience, and it has lasted a long, long time. (I can’t say the same for the HP laser printer that replaced the LW+, which had RAM slots but terribly cheap mechanical components that failed.)

Each of my succeeding Macs has actually taken away an upgradibility component or two, but has preserved the longevity and functionality factors that make Macs less expensive over time.

Those are benefits that can seem invisible when it’s time to retire a Mac and we’re faced with a choice between a $2,500 MBPro (for those of us who need the capabilities) and a $1,000 something-else that has ports and slots out the wazoo.

My argument has always been that I’d rather spend that money once and use a machine for 10 years, rather than spend it on three or four machines over that same period. I’d also rather spend my time accomplishing my creative work than “working on my computer.”


I used to think that way and did it for most of the 90s. But then I realized that since Macs have such a high resale value, if you time the upgrade and sell it after a 2-3 years at its peak, you can get a substantial amount of your initial purchase money back and put it toward a new machine. For example, say year 1 you spend $2500 on a new Mac, and three years later sell it for $1500 and buy a new model for $2500 ($1000 extra). That newer one is 2x as fast with more features (bigger drive, more RAM, better screen, lighter, whatever).

Overall you still will spend more total dollars over 10 years, but you’ll also be getting updated machines regularly, that are more powerful, more productive, have less problems (for the most part), are more secure, and are more compatible with Apple’s operating system updates, and require less maintenance. I used to work quite hard trying to make my 7+year-old Macs still run (including buying upgrades, software, external drives, etc.) and still running into limitations. Since I’ve been replacing my Macs every 2-4 years during the past 10 years, I’ve found things to be a lot smoother. (I still do a few older Macs I keep around, but just for certain tasks or nostalgia.)

BTW, there’s nothing wrong with either approach. A lot depends on the type of work you do, your computing proficiency, and what you can afford. I’ve gotten so I don’t like tinkering as much and just want my machines to work, so my new approach of upgrading more often is best for me.


I’m so sorry if it seemed like I was criticizing anyone here! I certainly largely agree with both of your posts. I was mostly reacting to the click-bait-y title/thumbnail of LTT’s video.

If the cost of a new PC today is the primary consideration, then a Windows laptop might very well make sense. It wouldn’t to me, but if that’s your thing, do what you want to do.

If surfing the wave of the latest models is something you don’t mind doing (I find reselling my old laptops traumatic, too much stress from fraudsters, etc.), then that might be a reasonable approach.

My own approach is, when I need a new Mac, buy the max spec’ed version of the model that fits me best, and then use it until it no longer meets my needs. I’ve got a first-gen M1 MacBook Pro 13", and it’s more than fine (though I’m getting nervous about storage, now that I’m below 40% free space on the 2 TB SSD). I’m currently planning to wait until the MBPro 14" gets the M3 Pro/Max refresh, and then I’ll buy that. We’ll see what the cost is, but probably max spec’ed out, so it lasts and lasts and lasts…

1 Like

Also a very good way to go. Secondary market buyers and sellers like Other World Computing validate that approach at both ends. A colleague of mine bought a 2016 MBPro from them during the pandemic so she could produce media for her employer; it cost her about the same as a new mid-range Windows laptop would have, but she was productive from day one and still finds it very usable.

I didn’t feel that way at all! Like you, I reacted to the mindless premise of the link, which has been a fairly reliable annoyance since the Mac/PC wars dribbled into the sand. We’re pretty much on the same wavelength here.

Until 2019 my approach had been to go mid range on the working theory that the starting configuration was too underpowered and the high-end configuration too rich for my wallet. It happened that the last two Macs I acquired (the 2017 MBPro and a late-2019 27 inch iMac) were purchased for me by my employers, and both employers had high expectations for what I would produce with them. So those both have the best available CPU and GPU, and I added 128GB RAM to the iMac out of the box.

So, I’ll be on Intel chips until Apple stops supporting them, at which time I’ll likely go back to a single mid range configuration (and by that time, what? An M5 complete computer on a chip? :slight_smile: ).

You definitely need to watch the video. Linus discusses the pros and cons of soldering down RAM and storage.

But that’s not the biggest problem. The biggest problem is that Apple charges several times what others charge for the same components. DRAM chips that cost less than $30 on a DIMM cost more than $200 when Apple solders them to a Mac motherboard. Even when you look at top-spec memory chips, there’s still a huge difference.

And the arguments that work for RAM don’t make any sense for storage. Especially when Apple’s high-end systems have replaceable socketed modules, but uses cryptographic pairing in order to make sure you can’t ever upgrade them (not even from Apple, let alone with third-party parts).

Nothing we don’t already know, but Linus presents the topic very well.


To be pedantic, I thought with Apple Silicon there are no discrete RAM chips soldered on to a board - RAM is now part of the M series package in the SoC. (Same with iDevice A series processors, etc.) Which, of course, is why it can no longer be upgraded.

Again, watch the video. Linus addresses this. (Time index 6:50)

What Apple is doing with M* SoCs is no different from what they’ve been doing with the A* SoCs. Although the DRAM is in the same package, it is not on the same die. It is a discrete component that can be replaced by someone with suitable microsoldering skills (and there are many YouTube videos from people who have upgraded the DRAM in iPhones!).

His point is not that there’s somthing inherently wrong with this level of integration, but the fact that Apple is charging far more than what anybody would consider reasonable. He cites the fact that Apple charges $200 per 8GB. He then looks at top-end SO-DIMMs which cost $60 per 8GB, retail.

Yes, integrated DRAM chips are not the same as an SO-DIMM, but it’s going to be a really hard argument to claim that their chips are so much better or that integration is so much more expensive that it requires a markup of >3x.

And storage is even worse, with Apple charging $200 for a flash chip that typically sells for about $16, retail. Even if Apple is buying something lightyears better than the competition (which benchmarks indicate that they are not), that’s still an insanely huge markup.


Total cost of ownership (TOC) is always less with Mac. This comes from using small machines since 1980, Macintosh since 1984, Windows since its intro, plus experience installing and maintaining small biz networks. TOC considers all expenses divided by useful years of service. TOC considers purchase cost, usability, ease of use, maintainability, reliability, interactivity among programs, serviceability… When given a choice, I always go with Apple.

I have to be honest - I’m really not going to take the time to watch the video because I don’t really care all that much. To me, computers are no longer built of modular, replaceable components, so I buy what I think I will need for the next 5 years. If I ever underestimate, there is always trade in and buy new or refurbished.

Maybe it’s a shame that computers are no longer modular as they once were, but I think they’ve become incredibly reliable as well. If I can afford to pay Apple’s prices for the upgraded specs I need, that’s what I’ll continue to do, so long as it provides the expected reliability and value to me.

I’m definitely not going to microsolder anything.

I bought my first real computer in 1985, I built my first in 1991 (after doing it often at my job.) I don’t miss those days at all.

And I suppose it happens with vehicles as well. We bemoan the upgrade pricing because the components likely cost the manufacturer far less than they charge, but I still get the vehicle with the components I want. I don’t try to add on later. It’s what the car costs, not the combined value of the cost of the components.

This is the new reality - if you want Apple OS devices, you buy the RAM and storage you need from the start.


I think both sides make a reasonable point.

I recognize there’s significant advantage to integrating these components. So I guess not being able to exchange or add additional modules on a MacBook after the fact is just the price we have to pay. But that doesn’t excuse Apple for making it impossible for an end-user to swap flash modules on systems where it’s socketed (Mac Studio and Mac Pro) or to charge ludicrous mark-up for commodities. I wouldn’t mind them making a profit, but a 3-fold overcharge is just outright rude.

That said, I can’t watch this guy’s videos. I’d need a transcript. He has such an annoying delivery with the over the top voice and all the exaggerated facial expressions. Too much circus for me. I suppose that’s what these “influencer” types think they need to do to get likes and what not.

Apples design has addressed every possible component that slows the process. Direct access Ram is faster than going through a wire or hardware interface. Next step might be direct access storage, but that is probably not soon. I buy pretty maxed out and keep them 5-6 years.

I am fine with choosing the amount of RAM and Disk at time of purchase because it will be LPDDR high speed RAM, mounted directly next to the SoC and interfaced with a much faster interconnect to the SoC. There is no bus to get in the way and the GPU in the SoC shares the RAM. Because the RAM is so fast and the GPU is in the SoC means removing a not insignificant inefficient GPU with it’s own RAM sending and receiving data over a PCIe bus. Why drive your data down the road and round a city block to be processed by the GPU before it can travel back home to the CPU? Soldered components take up far less space than socketed components leaving room for more battery cells. I gotta say the 18+ hours of real world battery life on the 16" M2 MacBook Pro is amazing! At the end of the day, I still have 60% battery left and that’s without sleeping and without any reduction in performance throughout the day.

As to storage, I just need enough to hold what I am actively working on. So enough room to hold all my applications and enough left over to hold a reasonable amount of my data. I can certainly archive older data that I wish to keep to an external disk, NAS, or cloud storage. This depends on your workflow and the sort of data and files you work with on a regular basis. The average person? Well 256GB to 512GB of storage is the current sweet spot. It’s what we deploy to tens of thousands of enterprise employees. They have 4TB of OneDrive storage to offload / archive. Very few run out of disk space.

Granted, this makes the most sense with laptops. Desktops are different and I won’t be buying the new Mac Pro for that reason. It’s designed for pros who need expansion cards but you can’t upgrade it otherwise.

I will just get a Mac Studio, when it is time to replace my M1 Mini. I have an AMD Ryzen 9 7950X3D (16c / 32t) @ 4.200GHz with 128GB RAM and 16TB’s of NVME SSDs and an Nvidia RTX 4090 running Linux for the real heavy lifting work. Dual boots to Win11 only for gaming sessions. At least until I can get my games to run on Linux (which works better than expected but requires considerable effort). I mostly game on consoles anyway. Meaning I rarely boot into Windows. Linux interfaces very nicely with macOS systems. Linux auto-detects AirPrint printers and they just work. I didn’t even need to install a driver. The printer was automagically detected and installed. I don’t need 365 on the Linux desktop but I could use the web version quite effectively if the need arises.

99% of individuals never ever upgrade their computers. They don’t add RAM they don’t add storage. Sure the nerds, geeks, and techies upgrade their systems but most average consumers do not.

1 Like

One other thought that occurred to me: if your Venn diagram is two circles labeled “Apple Hardware” and “Apple Services,” then the RAM and storage limits are addressed by Apple offering to off-load lightly used resources from your device until you need them.

My previous iPads (original and Air 2) were purchased with maximum RAM available at the time, because I knew they had a set configuration and relatively few ways to deal with offline storage.

My current iPad (just purchased) is 64GB, but offloads apps and data to iCloud based on usage. They come back within moments when I need them thanks to a fast fiber Internet connection.

I’m unexpectedly more comfortable with that than I thought I would be.

If I were traveling more with this device, I’d have considered more storage, but as it is, it’s an example of how Apple’s ecosystem works to mitigate perceived hardware deficits.

1 Like

One thing not mentioned here (perhaps it’s in the video?) is the ultimate cost of the non-upgradable SSDs. An SSD will wear out… and the smaller the SSD, the faster it’s going to wear out. With Apple Silicon, that means your computer is now dead. You can’t even boot off an external drive once he internal SSD is no longer usable.

I purchased an m1 mini, 16GB, 1TB SSD used. When I used DriveDX to check the status of the drive, it reported that the internal SSD had 79% lifetime left; somehow over the course of 18 months the prior owner had written 692TB to the drive. I asked what he had used the machine for? running endless benchmarks? He said he’d just used it for normal stuff, photos and videos. I suspect he decided to somehow override the warnings and did a whole drive secure erase with multiple passes, which is completely unnecessary on an SSD (especially if you use FileVault).

Throw away that key and reformat the drive (no zeroing or secure erase necessary) and because of both the encryption and wear leveling on the SSD, it would be a challenge for anyone to ever recover anything off the drive. I’m actually running everything off of a Thunderbolt 3 external fast NVME 2TB drive. While I might sacrifice some speed, it’s still plenty fast and it gives me more piece of mind that this computer will not suddenly seize up and become completely useless in the future.

I’m far more OK with soldered RAM than I am totally non-replaceable SSD. Even if it was just something that only Apple could do - to replace a faulty SSD or one that had been worn out from use. It would seem far more ecologically sensitive to do that instead of having to recycle the entire (now useless) computer.

1 Like

Yes, but this may not be a practical concern for most people. Apple’s been shipping Macs with SSDs for quite some time (and iPhones for a long time before that) and I haven’t read about anyone with dead SSDs.

The wear-leveling and write cycles of the SSDs Apple is using may be large enough that most people will upgrade their Macs before it dies.

FWIW, my 2011 MacBook Air is still running its original 128GB Apple SSD. DriveDX reports 70% life remaining. I realize that I’m not a heavy user on this computer (just casual web surfing and occasional MS Office use) and that this SSD has some pre-failure indicators, but the computer is still working fine. For a 12-year-old system (far longer than most people keep a laptop in service), I think it’s doing just fine.

If I wasn’t planning to replace this computer some time next year, I’d probably replace that SSD now, but since I plan on replacing it, I’ll take my chances.

WRT the Apple Silicon storage, the nature of the failure will determine if the Mac remains usable or if it gets bricked. When an SSD reaches its wear limit, the controller typically switches it to be read-only. Assuming Apple Silicon does something similar, then the ISC and Preboot containers may remain usable even after the rest of the SSD can no longer boot macOS. If so, you may still be able to boot from an external drive (preferably a Thunderbolt SSD for best performance).

1 Like

With Apple still only including 8GB of RAM in the vast majority of Apple Silicone computers, anyone who just browses the web heavily (how many people do you know who have a couple of dozen browser tabs open all the time?) are going to be paging to disk. We’ve been told that the setup of Apple Silicon makes this fast, and indeed it is. But that paging does impact the SSD, making the 256GB SSD an even faster ticking time bomb. The best spec’d 256GB SSDs have a TBW range between 150-300. I’m assuming Apple’s components are going to be at the top of that range. If you’re a basic user, you might never go that high.

I owned a refurb MacBook Pro 13" retina 2013, 8GB 256GB from roughly 2014-2020 and I can’t remember the TBW figure but I was down to about 20% life left. I was able to use an OWC upgrade to swap in a replace meant 480GB SSD. In 2020, I picked up a used top of the line 2015 Retina 15", 16/512, and upon my purchase it had about 10% SSD used, as the previous owner had barely used it. from mid 2020-mid 2022, I used about 12% of the SSD life. And now, for about 15 months of use with my external Thunderbolt 2TB boot drive, I’ve used about 8%, or 97TBW of this drive. I do use my computer a lot more for a lot of I/O heavy stuff - pictures, videos, etc. but I don’t think I’m that unusual of a user. Most people I know use theirs just as much or more.

Also, courtesy of Tidbits: An M1 Mac Can’t Boot from an External Drive If Its Internal Drive Is Dead

I think it’s good advice to suggest going for the 16GB/512GB config if at all possible. That should provide much better longevity and better performance too!

1 Like

As a nontechnical person here, what is the difference in RAM memory and SSD memory? Does RAM also have a limited number of times it can be written to?