Monday, August 31st 2026

MacBook Neo Burns Through SSD Cycles At An Alarming Rate

Apple's affordable MacBook Neo was a shock to the system of many a Windows laptop OEM when it launched in March 2026 with enough performance for just about all casual computing tasks and home PC use. However, it seems as though the MacBook Neo's 8 GB of RAM may pose a longevity risk, because of how often the laptop runs out of memory and is forced to use the swap file on the SSD to compensate. In recent SSD longevity tests, the YouTube channel UFD Tech found that the MacBook Neo could burn through a significant portion of its SSD lifespan with Swap usage alone.

Using an SSD burn tool, UFD Tech found that the 256 GB MacBook Neo burned through an estimated 5% of its overall lifespan after 20 TB, projecting a rated lifespan of around 414 TB, while the 512 GB saw a projected rating of around 1521 TB. The Brett, the host of the YouTube channel, then went on to use the MacBook Neo for a day to estimate how much data was written to the Neo's SSD swap file as it ran out of RAM throughout the day. Using the laptop as a glorified web portal, running Chrome for research, reading, and communication via WhatsApp and Discord, the MacBook Neo wrote "nearly 900 GB" of data to its SSD in just three hours.
Assuming the Neo's estimated 400-420 TB endurance, the laptop's SSD is only estimated to last around 1,400 hours, or 58 days, of use as a full-time basic productivity laptop. The good news is that the 512 GB SSD in the MacBook Neo appears to have at 2-3× the durability of the 256 GB model, resulting in around 5,000 hours of constant use before burning through its rated durability.
Of course, this will vary greatly by each individual's use case, and the rated lifespan of an SSD isn't the be-all and end-all of the drive's actual usable lifespan. However, this test does raise alarm bells over Apple's reliance on Swap because of the system's limited RAM. The results are made all the more concerning by the fact that the SSD in the MacBook Neo is non-replaceable, so when it eventually burns through all of its write cycles, it will be a more expensive repair than just buying a new drive, and there will likely be data loss involved.

Even Apple seems to realize that RAM is the MacBook Neo's weakest link, as rumors have been swirling about the hardware maker switching to the A19 Pro before long, upping the memory from 8 GB to 12 GB.
Source: UFD Tech
Add your own comment

55 Comments on MacBook Neo Burns Through SSD Cycles At An Alarming Rate

#26
kondamin
kondaminthose use MLC and the modules can be changed even if not standard m.2
Ah i thought you were talking about mac's
Posted on Reply
#27
Hecate91
DukenukemxThese devices should have removable SSD's, like every other PC in the market. They will fail, but with Macs they will take out the entire machine. Normal use doesn't matter because it can happen due to a number of reasons outside your control. For example, Apple M1's were reported to have excessive SSD wear due to a bug in MacOS. Not to forget that Apple put half as many SSD's in the M2's which having more NAND chips would spread out the SSD wear.
Exactly, the point is it doesn't matter if the writes being made are normal or not.
The point is these laptops are e-waste once the SSD fails, unless you have experience and tools to BGA solder a new SSD in,and can find replacement NAND chips.
And this is why people complained about the neo only having 8GB of RAM, all of the SSD swapping is going to wear the SSD more rapidly.
Then again this is like baby's first Macbook, lots of people buying them might find the limit of 8GB of ram or the phone SoC and trade it in for a better one before the SSD wears out.
Posted on Reply
#28
TheinsanegamerN
DukenukemxThese devices should have removable SSD's, like every other PC in the market. They will fail, but with Macs they will take out the entire machine. Normal use doesn't matter because it can happen due to a number of reasons outside your control. For example, Apple M1's were reported to have excessive SSD wear due to a bug in MacOS. Not to forget that Apple put half as many SSD's in the M2's which having more NAND chips would spread out the SSD wear.


Also yes, Macbook SSD's do fail. There's an entire industry built around repairing Apple SSD's because it requires serious expertise to do something a 12 year old could do with a Windows machine.
But them being removable ignores the point: after 8 years, these devices are, at least 99.99% of them, are still chugging along without issue. By the time the writes become relevant, these devices will be 10+ years old, having long earned their keep.
Posted on Reply
#29
remixedcat
Dr. DroI hope this is supposed to be witty, because it's not applicable to this computer. It's basically an iPhone 16 Pro's SoC, meaning RAM and storage are fixed and cannot be upgraded.
and a nightmare for data recovery.
kondaminthose use MLC and the modules can be changed even if not standard m.2
it's embedded on the SoC so it's non upgradeable/replaceable unless you board swap, and data recovery is a nitemare.
TheinsanegamerNIt's certainly hysterical. 900gb of writes in 3 hours is far from normal usage.
At work our primary machines were all 8/256 laptops. Thousands of machines from 2017, still in use. The number of drive failures we have had can be counted on one hand, none from excessive writes.
OS imaging or drive imaging will do an entire drive write + plus syncing data which needs to be done on new drives and restore operations.
Posted on Reply
#30
Dukenukemx
TheinsanegamerNBut them being removable ignores the point: after 8 years, these devices are, at least 99.99% of them, are still chugging along without issue. By the time the writes become relevant, these devices will be 10+ years old, having long earned their keep.
That's a strange statistic you pulled there, but on average it's 8% to 12% failure after 8 years of normal use. Considering the Neo's aren't exactly normal use, I'd say they'd fail sooner. Another thing to consider is that SSD's could fail due to user error like spilling a drink or getting a surge in your power supply. There is no reason for Apple to not have M.2 slots other than to push customers to pay for overpriced storage upgrades. The repair process for these Apple SSD's is nightmare fuel. Just trying to get working NAND chips that aren't already worn out is stupid. You also need another working Mac to properly write data to the NAND.

Posted on Reply
#31
Let3r
Which is especially bad because it’s non-replaceable
DukenukemxThat's a strange statistic you pulled there, but on average it's 8% to 12% failure after 8 years of normal use. Considering the Neo's aren't exactly normal use, I'd say they'd fail sooner. Another thing to consider is that SSD's could fail due to user error like spilling a drink or getting a surge in your power supply. There is no reason for Apple to not have M.2 slots other than to push customers to pay for overpriced storage upgrades. The repair process for these Apple SSD's is nightmare fuel. Just trying to get working NAND chips that aren't already worn out is stupid. You also need another working Mac to properly write data to the NAND.

Luis Rossman! Soooooo gooooooooood. Totally right. Though I did watch a video where because the MacBook uses iPhone 16 pro max chips they took a 1TB chip from that and put it in the neo.
Posted on Reply
#32
sudothelinuxwizard
DukenukemxAlso yes, Macbook SSD's do fail. There's an entire industry built around repairing Apple SSD's because it requires serious expertise to do something a 12 year old could do with a Windows machine.
Yes they do. Mostly random failures however (such as buck converters shitting themselves and sending the NAND 12V, destroying it), which are very infrequent. You need to keep in mind how many Apple devices are sold per year.

It doesn't excuse them soldering the drive down (especially on the thicker models), but given how worked up the internet gets any time Apple has issues of any sort I am pretty sure someone at Apple checked failure rates and determined they were fine.

I also would like to point out as a youngin myself the average 12 year old can't even use anything that isn't an iOS app (even macOS is too hard for them). I would bet every computer I have that not even 5% of 12- or even 16-year olds could replace a SATA or M.2 drive.
DukenukemxThat's a strange statistic you pulled there, but on average it's 8% to 12% failure after 8 years of normal use. Considering the Neo's aren't exactly normal use, I'd say they'd fail sooner. Another thing to consider is that SSD's could fail due to user error like spilling a drink or getting a surge in your power supply. There is no reason for Apple to not have M.2 slots other than to push customers to pay for overpriced storage upgrades. The repair process for these Apple SSD's is nightmare fuel. Just trying to get working NAND chips that aren't already worn out is stupid. You also need another working Mac to properly write data to the NAND.

Backblaze is a cloud service, it probably writes even harder than when you trash the swap on a neo. I have no idea why you call that "normal use".
Posted on Reply
#33
remixedcat
DukenukemxThat's a strange statistic you pulled there, but on average it's 8% to 12% failure after 8 years of normal use. Considering the Neo's aren't exactly normal use, I'd say they'd fail sooner. Another thing to consider is that SSD's could fail due to user error like spilling a drink or getting a surge in your power supply. There is no reason for Apple to not have M.2 slots other than to push customers to pay for overpriced storage upgrades. The repair process for these Apple SSD's is nightmare fuel. Just trying to get working NAND chips that aren't already worn out is stupid. You also need another working Mac to properly write data to the NAND.

you need to replace both the NAND and controller due to encryption being cpu bound since the SSD controller is built into the SoC. which means a nightmare data recovery + pricey board swap
sudothelinuxwizardBackblaze is a cloud service, it probably writes even harder than when you trash the swap on a neo. I have no idea why you call that "normal use".
backblaze runs as any other cloud drive service like idrive, sync.com, crashplan, etc.

also apple's claims of needing to solder the ssd and have the controller on the SoC for max speed only talk about theoretical max sequential, which is not most of the workloads. Even a SATA SSD is plenty for lower QD workloads for most average joes and janes...for nvmes, however...my FC530 500GB which is a higher end nvme has these speeds which are plenty even for my latency sensitive audio production. So apple's marketing has people brainwashed into thinking it's for speed when it's for maximum ass raping people with expensive recovery and replacement costs.

Note: I ran this test while doing a lot of other stuff on my computer and the drive is 80% full too. it would be higher if it was running by itself and more than 80% full

Posted on Reply
#34
Wirko
OnasiThere is, so far, no “malfunction”. There is a software reading. We do not know if it’s correct. If we’ll have an epidemic of failing Neos - we will talk then.
Samsung 990 Pro SSDs also had a "fast countdown to zero" issue. It was considered serious.

How common are soldered SSDs in laptops other than Apple and Microsoft Surface?
Posted on Reply
#35
R0H1T
MrDweezilThe good news is this can be solved with more RAM or a bigger SSD, both of which are super cheap and trivially replaceable by the customer.
The bad news though is that we're talking about Apple, a company famous for charging 2-4x their competitor's prices for simple DRAM/NAND upgrades so it'll be a cold day in Hell's deepest layers before Apple willingly "solves" this :laugh:
sam_86314This is a problem with the laptop only having 8GB of memory, so it's probably using the swap space just as much as the system RAM.
Nope, have an M2 Mac mini here with 8GB RAM & there's definitely not the same amount of wear/tear on my SSD ~ it's also primarily used for web browsing so the use cases are "almost" identical wrt the OP. I wonder if they have QLC drives though, in the Neo?
Posted on Reply
#36
xorbe
3 hours? 900GB / 180 minutes / 60 seconds = 85MB/s sustained (claimed)
Posted on Reply
#37
remixedcat
xorbe3 hours? 900GB / 180 minutes / 60 seconds = 85MB/s sustained (claimed)
Easy to do that when imaging to the drive or initial setup or tons of updates and/or indexing/fs Journaling
WirkoHow common are soldered SSDs in laptops other than Apple and Microsoft Surface?
Some dell xps had em but I think it was only the smaller screen models of a certain time. Also some Lenovo x1 carbon, however the controller was still its own chip on the board and not integrated into the SoC.
Posted on Reply
#38
R0H1T
remixedcatEasy to do that when imaging to the drive or initial setup or tons of updates and/or indexing/fs Journaling
Even playing endless media/YT videos and/or opening multiple tabs can do that especially if the system is low on memory! In fact browsers today are a good (re)source to test your system responsiveness & yes NAND wear, in part because they interact in so many ways with your system (also) depending on the sites you've opened.
Posted on Reply
#39
Shihab
sudothelinuxwizardI also don't completely get your analogy, it's like asking whether I'd want to replace my SSD with a hard drive.
The analogy is about taking extreme events for frequent ones.

Go through enough of them, and you may find an SDD that lasts to the Zettabyte range. But you wouldn't bet that the average (not even the 1 percentile) would even come close.

It's a probabilities game. If you have an indicator that correlates with said propability, reason dictates you would want to minimise it -or its rate of change- if your goal is reducing the said probability. You wouldn't just take anything that doesn yield you a 100% probability as acceptable.

In other words, you wouldn't just take a random youtuber's SSDs or some light bulb in some American firestation as a metric, you would take a probability distribution generated from a sufficently large sample, which is what manufacterers do to come up with that TBW figure.

Now unfortunately, manufacterers don't share the useful data with us afaik, so prediction based on one's own degree of risk aversion isn't possible. All we have to work with is the fact that this TBW figure is the point after which the failure risk becomes to high for the manufacterer's comfort. And personally, I don't see why the same shouldn't apply to the consumer as well.
Posted on Reply
#40
_roman_
Hecate91The point is these laptops are e-waste once the SSD fails
or DRAM fails or CPU fails because everything is soldered. + the other issues when the board itself has any issues.

80 to 100€ in the past for a replacment asus motherboard for a laptop may have fixed a few e-waste boxes. i am sure quite often plugin dram and cpu moved from one hardware to the next one.

soldered smartphone / tablet is okay because it is mobile based and wlan standards and mobile standards are changing. you need the frequencies and the hardware gets quite slow. but for a desktop box or notebook soldering the expensive components is bad.

-- I state. Lenvo / Apple / DELL whoever else cares only for the Warranty or Service plan period. I checked the serial and saw lenovo service plan, 4 years thats it. after that it is ewaste not offically for the manufacturer. I did not keep those but i got really nice answers over the years from ASUS support. Their fault and my motivation why i went to desktop computing or refurbish mobile notebook computers.

those 4 years are in the best case, 2 years or 1 year is very very common

-- i tested again this year SATA Crucial SSDs vs sata hdd vs PCIE nvme. nope the claims sata ssds are fast enough, they also hang like hell in caja file manager. advanced file systems like btrfs speed up things over sluggish ext4 and ntfs, but sata ssds days are gone. sata ssds are good for cold, read only storage or for retro computing, e.g. dated hardware like intel 3rd generation intel core.

-- chromium browsers sometimes want 2GB or more on certain tabs. www-client can fill up dram and swap very fast, google chrome was known for having a memory hole bug for years. vivaldi still kills itself certain browser tabs in 2026 here even with the previous 2x32gib dram configuration.
Posted on Reply
#41
Dukenukemx
_roman_or DRAM fails or CPU fails because everything is soldered. + the other issues when the board itself has any issues.
When has a CPU ever fail for you? I've had ram fail but that's also why I still want removable ram. I like the idea of CAMM2. Even still, ram failure is often due to a bad batch of ram that no manufacturer will ever tell you. SSD's will fail at some point, we just don't know exactly when.
Posted on Reply
#42
dlgh7
I mean the bread would be in the pudding right? Because it means we would see this exact issue on the iPhone. Not only that but it would date back numerous generations on all of them with 8gb of Ram or less. And they have all failed right?
Posted on Reply
#43
sudothelinuxwizard
ShihabThe analogy is about taking extreme events for frequent ones.

Go through enough of them, and you may find an SDD that lasts to the Zettabyte range. But you wouldn't bet that the average (not even the 1 percentile) would even come close.

It's a probabilities game. If you have an indicator that correlates with said propability, reason dictates you would want to minimise it -or its rate of change- if your goal is reducing the said probability. You wouldn't just take anything that doesn yield you a 100% probability as acceptable.

In other words, you wouldn't just take a random youtuber's SSDs or some light bulb in some American firestation as a metric, you would take a probability distribution generated from a sufficently large sample, which is what manufacterers do to come up with that TBW figure.

Now unfortunately, manufacterers don't share the useful data with us afaik, so prediction based on one's own degree of risk aversion isn't possible. All we have to work with is the fact that this TBW figure is the point after which the failure risk becomes to high for the manufacterer's comfort. And personally, I don't see why the same shouldn't apply to the consumer as well.
Well, I still believe that with Apple's or even a PC manufacturer's drives will not fail before the rest of the machine does, even if we only assume 1 PBW (which I think is fairly conservative, since it's not only this guy whose SSDs have lasted this long).

However, your point about the manufacturer using failure rates to set the warrantied TBW does make sense. Despite that, I still think it's just where, say, 0.1% of SSDs begin to fail so Tim Apple (or now John Apple) can wipe his ass off any customer RMAs. It isn't right, but I don't think the hardware's the problem.
Posted on Reply
#44
Wirko
R0H1TEven playing endless media/YT videos and/or opening multiple tabs can do that especially if the system is low on memory!
That would point to an extremely stupid virtual memory manager, one that constantly moves hot data to the swap file while it keeps cold(er) data in RAM.
Posted on Reply
#45
mrnagant
It isn't just about all out failure, it is about performance. It is there and you might just "deal" with it.

I have an nvme that is on its way out. But because of prices I have not yet replaced it. It is an Intel 600p 2TB. When I download a game, it is also installing at the same time right. These completely slow down my computer. It can take 2-3x times longer than usual. If I have another download going, one download slows to an absolute crawl. Everything just takes a little more time. The computer is useable, but to me noticeably slower. But when I am not installing big things, things are fine. But in fact, the health of my nvme is on its way out. Only has been getting worse.

Not always will it just one day stop working. It will slowly stop working as good, day after day, until it dies or your replace it.

So to say "oh my users are fine. everything is fine here. I haven't seen any issues" doesn't point to what in fact could and is actually happening.

8GB might be ok amount. But also remember it is shared with the GPU. On like a desktop or something the likelihood of having a terrible discrete card at least comes with its own RAM. GPUs memory pool is basically shared these days with all the applications that use GPU memory. It is incredibly easy to use 2, 3, 4GB of RAM without playing a single game. So, this adds to stuff getting kicked out of RAM into swap file. I also don't know how the NPU impacts RAM utilization, how much of what it can do can end up in swap and hit it constantly.

If the alorgithms suck on the nvme, the swap file will hit the same part of the ssd as well which will cause an increased degridation on the size of that swap file. So hopefully Apple is rotating the swap file location in real-time. I think this is why my 600p is slowing down a lot. Its SLC cache using the same space and just constantly getting hammered. With my SSD largely being filled up this whole time, the SLC cache has been at its smallest allotment, further speeding up degradation of that area.
Posted on Reply
#46
Dukenukemx
dlgh7I mean the bread would be in the pudding right? Because it means we would see this exact issue on the iPhone. Not only that but it would date back numerous generations on all of them with 8gb of Ram or less. And they have all failed right?
Except iPhones don't run MacOS and don't get to see gigs worth of file transfers. iOS is a media consumption OS while MacOS is a productivity OS. This is why running a cell phone chip in a laptop is a bad idea.
Posted on Reply
#47
misha1350
MrDweezilThe good news is this can be solved with more RAM or a bigger SSD, both of which are super cheap and trivially replaceable by the customer.
Super cheap and trivially replaceable by the customer is something that cannot be done on any Macbook.

The correct answer here would be to ditch the Neo and grab oneself a second-hand ThinkPad T14 Gen 2, or a T14 Gen 4 AMD with 32GB RAM.
sudothelinuxwizardThese seems overblown to me, that French guy who makes Mac mini and Studio upgrades (don't remember his name) tested his two 256GB Mac minis and they kept working well after 4.5 PBW. The "lifespan" in the S.M.A.R.T data is just what Apple will replace for you under warranty as far I I'm concerned, random failures aside there's no way they die at under 500GBW (my Chinesium Silicon Power 256GB SSD with a shitty JMicron controller and B-grade repackaged SK Hynix NAND is at around 600 TBW and still works).

At 200 GBW per day (which seems like an awful lot given UFD Tech's video says that his heavy usage would cause around 100GBW) that's well in excess of 35 years even if we assume they don't make it anywhere near that long.

Why? Because Apple SSDs have 25% overprovisioning (A "256GB" one actually has 320GB of free capacity, 64GB of which being spare blocks). A quarter of the NAND could die and the SSD would keep going.

I also assume given they reuse the same NAND modules across iPhone, iPad, and Mac, they can be stricter with the binning.
They also have their S5E/S6E PHYs on every NAND and a controller on the SoC instead of one weak controller which might make wear leveling work better (just speculation on my part for that bit).
Your Silicon Power SSD only works because it isn't a repackaged chinesium memory die. Silicon Power has been around for longer than you think, and while it may be cheap and use low-end and old controllers, it's nowhere near the levels of Netac or KingSpec, which are to be avoided like wildfire.
Posted on Reply
#48
Tess
Good article, thank you!
Cpt.JankThe good news is that the 512 GB SSD in the MacBook Neo appears to have at 2-3× the durability of the 256 GB model, resulting in around 5,000 hours of constant use before burning through its rated durability.
I thought that's how it just always works? A 512GB SSD is twice the "durability" as a 256, and a 1TB twice that of a 512 (all other things equal)... No?
Cpt.JankEven Apple seems to realize that RAM is the MacBook Neo's weakest link, as rumors have been swirling about the hardware maker switching to the A19 Pro before long, upping the memory from 8 GB to 12 GB.
Apple apps love more RAM! 32GB ~ 48GB is likely the sweet spot for most of the creative apps - given small to medium project files. At least in 2007/8 it was 24GB (tested extensively) so I'm assuming on a projection. :D
Posted on Reply
#49
sudothelinuxwizard
misha1350Your Silicon Power SSD only works because it isn't a repackaged chinesium memory die.
It is, I opened it and googled (DuckDuckGoed?) the part number.

I know Silicon Power has good models too, in fact even my 512GB version of A55 has a much nicer Maxiotech controller and branded YMTC 128-layer TLC (although that is probably more component lottery than the higher capacities being better in general given the prices).
Posted on Reply
#50
user556
Very much overblown. The fact that app degraded performance isn't the topic of conversation says it all.
Posted on Reply
Add your own comment
Nov 20th, 2026 12:39 PST change timezone

New Forum Posts

Popular Reviews

TPU on YouTube

Controversial News Posts