If you’ve ever tried to “future-proof” a purchase by paying for everything you might eventually need up front, you know it can be a sucker’s game. The problem? We can’t actually see the future.
But today we got our hands on LPCAMM2 for the first time, and this looks like the future to us. LPCAMM2 is a totally modular, repairable, upgradeable memory standard for laptops, using the latest LPDDR chips for maximum speed and efficiency. So instead of overpaying (or under-speccing) based on guesswork about your future memory needs, you’ll hopefully be able to buy your next laptop and then install more RAM as needed. Imagine that!

We say “hopefully” because the laptop on our teardown table today is the first of its kind, thanks to a collaboration between Micron and Lenovo—and it remains to be seen how many other big laptop makers will adopt LPCAMM2 technology. But judging by the results of our initial hands-on, the writing is on the wall for laptops with soldered-down, non-serviceable memory.
What is LPDDR, and why do manufacturers solder it down?
Repairable, upgradeable RAM isn’t exactly a new idea. As anyone who has ever built a PC knows, we’ve had swappable DDR RAM sticks (also known as DIMMs, or Dual In-Line Memory Modules) since basically forever. From old Gateway towers to today’s gaming powerhouses to zillion-dollar industrial servers, upgradeable and replaceable RAM is still very much a thing. And for many years, the same was true of laptops, which used a slightly more compact (“Small-Outline” DIMM, or SO-DIMM) version of those same RAM sticks.
More recently though, we’ve seen increasing adoption of LPDDR—a low-power flavor of RAM (hence the “LP”) developed for mobile devices like phones and tablets. Whereas conventional DDR RAM excels at performance applications where power consumption isn’t a primary concern, like video editing or gaming, LPDDR wins the day when it comes to efficiency—a.k.a. battery life. And so for laptops in particular, the benefits of LPDDR are hard to beat.
The drawback of LPDDR, though, is that it has to be soldered to the main board in close proximity to the processor—making repairs and upgrades completely impractical. But why?
LPDDR operates at lower voltages compared to DDR, giving it the edge in power efficiency. But, the lower voltage makes signal integrity between the memory and processor challenging, requiring tighter tolerances and shorter trace distances—that is, the farther the signal has to travel, the more voltage you need for a reliable signal. This is why LPDDR is soldered down as close to the processor as possible.

In short, laptop makers and consumers alike have faced an unfortunate dilemma: conventional SO-DIMM RAM for serviceability and upgradeability, or soldered LPDDR chips for longer battery life.
Today, that changes.
Enter LPCAMM2
Standing for “Low-Power Compression-Attached Memory Module,” the new tech is as the name suggests: LPDDR chips on a compact board that screws in place very close to a laptop’s CPU. Combining the efficiency and speed of LPDDR with a thin, lightweight, upgradeable design and a trick interface that gets everything up close and personal with the CPU, LPCAMM2 seemingly does it all. And with dual-channel performance already baked in, a single LPCAMM2 module can do the job of a pair of the old socketed SO-DIMM sticks with a much smaller footprint and better thermals to boot. Finally: modular, performant, power-efficient laptop memory for the masses.



Even though LPCAMM2 is arriving initially from Micron, in a Lenovo product, the technology owes its existence to an alliance of tech companies working together over the course of several years. The first iteration, known as CAMM, was an in-house project at Dell, with the first DDR5-equipped CAMM modules installed in Dell Precision 7000 series laptops. And thankfully, after doing the initial R&D to make the tech a reality, Dell didn’t gatekeep. Their engineers believed that the project had such a good chance at becoming the next widespread memory standard that instead of keeping it proprietary, they went the other way and opened it up for standardization.
They were right. Only a few short years later, with the blessing of the JEDEC standards body, LPCAMM2 is here and ready to take the torch. You can now buy 32GB and 64GB modules.


Dell is one hero in this story, creating something the tech world sorely needed and then sharing instead of keeping it for themselves. Thankfully, this story is full of heroes: Micron and Lenovo are first to bring LPCAMM2 to market, with Samsung, ADATA, and others backing it as well. Instead of everyone going their own way, they’ve rallied around a new industry standard—meaning an off-the-shelf LPCAMM2 module should theoretically work in any laptop that adopts the technology, regardless of who manufactures it. With the industry as a whole on the same page, backing standards like this, the world becomes a more repairable place.
Designing for a repairable future
The advent of LPCAMM2 is especially gratifying for repair advocates, who for years have been told that repairability simply can’t coexist with cutting-edge tech in thin-and-light devices. We’re not ready to accept that, and we’ve long argued that OEMs who are willing to innovate with repairability in mind can do better. Maybe we can’t see the future, but we can envision one that’s more repairable than what we’ve been sold in recent years—and we’re grateful when companies like Micron and Lenovo take the leap to make that future a reality.

LPCAMM2 exemplifies our ability to advance technologically while designing with sustainability in mind. It represents a significant step forward in the fight against planned obsolescence. By fighting for a modular, upgradeable memory solution for chips previously stuck in a soldered hellscape, manufacturers are demonstrating their commitment to creating devices that stand the test of time.
There’s so much to gain here: from increasing device lifespan at schools and businesses, to reducing anxiety for consumers at the point of purchase, to enabling hassle-free repairs for devices that would otherwise be scrapped. As more companies rally behind this standard, we can look forward to a future where more laptops are built to last, and where repairs and upgrades aren’t only possible, but encouraged. There’s no question that the potential for this technology to make a tangible difference is real, and it’s right in front of us.
Full Disclosure: iFixit has prior business relationships with both Micron and Lenovo, and we are hopelessly biased in favor of repairable products.
댓글 22개
Awesome looking technology and I hope it will be widely adopted by the industry!!!!
zane - 답글
I'm kind of sick of tiny screws when a clipped in format should be an option. M.2 drives me nuts too. I've seen non-screw M.2 setups, and we should see a LOT MORE of them. Too bad this memory format takes 3 of those headache screws. As usually with tech as we move along, it gets both better AND worse at the same time. Not to mention probably a lot more expensive.
Scott Wilkins - 답글
Given the signal integrity issues discussed in the article, I wouldn't be surprised if the screws are required to get a solid connection on all the contacts. It looks like there's a whole array of pins here as opposed to a single row of contacts, so it doesn't look like the existing clipped-type solutions would be viable. Maybe something will come along later, but I'd imagine it would have to be more complicated and require tighter tolerances, both of which add cost.
Also, this is a replacement for soldered-in RAM, so it's not really fair to compare the expense/hassle to existing RAM sticks.
Gareth Chen -
It seems similar to the compression connections in a CPU LGA socket, this RAM has a vast number of connections not achievable in a compact format any other way. It's still 1000% better than soldered LPDDR, I'm all for it.
Dave C -
Go take an antenna module off any MacBook and come let me know how bad you think these screws are in comparison.
GARRETT KIPPS -
well, the thing you suggest is called sodimm socket :)
RipperDoc -
I agree that tiny screws are annoying, but compared to a heat gun and tweezers, this is far better.
Abc Bcd -
If you think laptop makers are going to give up the cushy margins they've been getting on soldered memory upgrades just because this exists now, you're freaking delusional. Pro-consumer companies like Framework will be all over LPCAMM2, but all the major laptop OEM players are going to avoid this like the plague.
Chris Fetters - 답글
I disagree. You've already got two of those big players on board.
What I suspect will happen is that the likes of Dell & Lenovo (+ HP & Samsung, when they pick it up) will incorporate this into their pro lines that are aimed at the business market, but stick to soldered memory in the cheaper domestic machines. Business buyers want longevity of platform configuration, for ease of software support - this gives that while allowing for incremental size upgrades.
This tech actually makes it cheaper for manufacturers to offer a range of configured memory sizes on the one base unit, across both price and time, because the memory is added later in the production sequence.
The interesting manufacturer will be Apple, who seem to be more locked-in to the concept of fixed, soldered-in configurations.
Brian -
My custom heat sink for these is going to be the outline of a recognier from Tron.
Chris W - 답글
I don't think you can get the kind of tolerances needed with clips. There is a reason everyone still uses screws for planar connectors that need to be level.
Antti-Ville Tuunainen - 답글
But why did they get rid of the Trackpoint buttons again? :-(
BoFiS - 답글
Some have complained about the tiny screws that hold the module in place. Never fear - they're captive screws, so they should line up with the studs automatically. All you have to do is put light pressure on them with the screwdriver, turn them counterclockwise until you feel the end of the screw thread drop into the stud thread, then turn clockwise to tighten. The backward rotation step greatly reduces the chance of cross-threading, especially with screws going into plastic or sheet metal.
PAB1130 - 답글
More pins for users to bend
Luis Giordano - 답글
To my understanding two issues remain (pls correct me if I'm wrong): (1) lpcamm2 does not support ECC ram, so if this becomes the new standard ecc won't ever become mainstream in laptop (2) one lpcamm2 module only support 128bit bandwidth (better than 64bit for sodimm, quote from article "with dual-channel performance already baked in") when apple silicon already uses 256bit and according to rumours even AMD/intel might in the future to improve igpu performance which is limited by bandwidth in most scenario, I hope the next steamdeck can have 256bit (8x32bit LPDDR5 channels) memory.
Agno - 답글
Hi, here to correct you. DDR5 has ECC (generally, required for the speeds it gets), and the JEDEC spec does also include ECC fully fledged response pins, so you could potentially even add fully-fledged ECC to this without the spec getting in the way:
"This specification defines channels and subchannels as follows:
• A channel is 64-bit data interface with varying number of subchannels.
o With ECC support, the channel becomes 72-bits wide
o A DDR5 channel consists of two 32-bit subchannels
o A LPDDR5/5X channel consists of four 16-bit subchannels"
You are right that you 'only' get 128 channels for dual memory, but Apple has four in an M-series chip and 'only' gets 200GB/s (vs 128GB/s at 8000MT/s DDR5), so a 256bit bus isn't everything. Also DDR4 ended up at 3600MT/s vs the 3200MT/s max in the spec, so one imagines DDR5 will get there. Plus LPCAMM2 gives us a way forward to wider buses without huge pin footprint problems! I can foresee this being a desktop tech eventually!
Ashley -
The problem is specifically LPDDR5 CAMM2.
DD5 CAMM2 does support full ECC, but LPDDR5 CAMM2 does not. LPDDR5 CAMM2 and DDR5 CAMM2 are not compatible, and no consumer device seems to be targeting DDR5 CAMM2. I sure hope they include proper ECC support in LPDDR6 CAMM2 when they standardize it.
You can download the doc free at https://www.jedec.org/document_search?se... , registration and login required.
Antti-Ville Tuunainen -
Another solution, used for example in the MNT Reform, is to put the CPU and RAM on a separate daughterboard. That way, you can have soldered RAM and still have an upgrade path. One downside is that you can't upgrade the RAM alone, you have to upgrade the processor too. But in most situations, if you really need more RAM, you'll likely need a gruntier processor to service it. Another downside is that it makes for a thicker construction. Even so, it can be a viable solution. In general, though, LPCAMM2 is very welcome news.
Les Kitchen - 답글
Looking forward to seeing this in Lenovo and Dell laptops … might be a long wait to see in a MacBook. 🤔
John Corr - 답글
Older laptops did have upgradable memory hell they even had a magic slot you could add other functionality to the computer. This is not revolutionary what was wrong with sodimm ram?
E Pluribus Unum - 답글
SO-DIMM connections don't have the bandwidth for today's processor speeds and low power requirements. It's a physical limitation in an environment where adding several millimeters to the connection length is significantly detrimental to performance.
PAB1130 -
The next step could be a contact adapter that mates directly with the memory bus pins on the CPU package, and permits addressing multiple memory sizes. That will remove most of the PCB trace lengths between the CPU and memory module. Hello, Apple? Talk to your customers - how many will gladly take an added several millimeters thickness to have replaceable memory and SSD storage?
PAB1130 - 답글