LPCAMM2 32+ to 16GB downgrades: will your 16GB module be reused, resold, or end up in a drawer?

Framework Laptop 13
,
1

I’m interested in one downstream effect of the recent 32GB+ to 16GB LPCAMM2 preorder downgrades that I haven’t seen discussed much, if at all.

I’m not really trying to relitigate whether Framework could absorb the supplier price increase, or whether the downgrade was the least-bad short-term fulfilment option. I can see why they did what they did, even if I don’t agree with it. I’m more interested in what happens afterwards.

My read of the situation is this:

People who originally ordered 32GB+ presumably did so because they expected 32GB+ to be the appropriate capacity for them. Supplying 16GB first does not necessarily remove that demand, it probably just defers it (unless people are en masse cancelling orders).

Because LPCAMM2 systems are generally single-module, a displaced 16GB module has little natural upgrade demand (I think?). Somebody wanting more RAM has to replace their existing module, while somebody genuinely satisfied with 16GB generally already has their only slot occupied. There may well be demand from RAM-less systems, failed modules and other niche cases, but outside that fairly narrow set of buyers the natural purchaser starts to look a little mythical.

The difficult bit is that Framework cannot really know yet how much latent module churn it has created. Someone accepting the 16GB downgrade today might genuinely keep it for the life of the machine, or they might simply tolerate it until 32GB+ becomes more affordable or available and then replace it. Both currently look like an accepted 16GB order, but the latter is really deferred higher-capacity demand plus a future displaced 16GB module.

That makes me wonder how many of those modules will ultimately remain in service, how many will successfully find another user, and how many will simply end up sitting in drawers.

If that happens at meaningful scale, I think it creates a fairly perverse market effect: two modules are manufactured and sold to satisfy one customer’s original requirement, while one ultimately provides no ongoing useful service.

In that sense, Framework’s chosen solution looks potentially net inflationary to the wider RAMpocalypse. Obviously I am not claiming that Framework itself causes global RAM inflation, but in a supply-constrained market, scarce product being manufactured and purchased only to become effectively unused is avoidable demand. Within the already immature LPCAMM2 ecosystem, enough of that behaviour seems directionally capable of worsening scarcity and adding inflationary pressure to the same market conditions that caused the downgrade in the first place.

That is my current read of the economics, though, and I may be missing something - hence asking here rather than presenting it as settled fact.

I suggested to Framework Support that they consider a trade-in, buy-back or reuse program for the forced 16GB modules. They confirmed there currently isn’t one, but said they would pass the suggestion on. They also told me there is no confirmed timeframe for further third-party 32GB+ validation or an officially supported alternative. I don’t think community interposer workarounds/modules that Framework itself currently lists as incompatible, are realistically usable alternatives for most buyers.

So I’m curious what affected buyers actually expect the eventual outcome to be:

If your 32GB+ preorder was downgraded to 16GB, what do you realistically expect to do?

  • Cancel the FW13 Pro order
  • Keep the 16GB module long-term
  • Reconfigure back to 32GB+ before shipment
  • Remove Framework RAM / source third-party RAM
  • Upgrade later and sell the 16GB module
  • Upgrade later and reuse the 16GB module elsewhere
  • Upgrade later and keep it deliberately as a spare
  • Upgrade later and it’ll probably become drawerware
  • Not sure yet
0 voters

If selling - to whom?
If acquiring third party - what, from where with what availability? Will you be waiting with a non functional FW13 Pro if you haven’t already acquired some?

Personally, I’m probably in the reconfigure-before-shipment group. I could get by with 16GB for perhaps a year, but not without making some compromises to the experience, so my unfortunate likely outcome is paying to go back to the 32GB configuration I originally ordered - because anything third party, without some jank (as at 8 Sept 2026) is unobtanium.

What I’m particularly interested in is the split between people who genuinely settle on 16GB and those for whom accepting 16GB today is simply deferred module churn. Framework can see the former and latter as identical preorder outcomes now, but the actual memory demand and fate of those modules may not become apparent for months to years - and Framework may never be able to observe how much of the displaced RAM was successfully reused versus simply stranded.

My concern is therefore that the downgrade may solve an immediate fulfilment problem while creating a longer-term combination of deferred higher-capacity demand and stranded lower-capacity hardware that simply isn’t visible in Framework’s short-term numbers.

Interested to know whether that matches what affected buyers actually plan to do, or whether I’m sitting in my own little thought bubble here.

1 Like
3

I read somewhere that some big-brand LPCAMM2 laptop lines have been delayed, and I assume that’s an externality that FW could not have accounted for or controlled. If folks want to sell their 16GB module they’ll need people to sell it to, and if those products aren’t getting released, the customer pool shrinks.

I wonder if that will change in the future? If someone is running a single stick of 16 or 32, an additional 16 would be desirable.

4

I definitely agree that FW couldn’t have controlled the situation, but I don’t necessarily agree that the risk was not something that could have been accounted for. The 3rd party market isn’t mature no matter which way you look at it with the lack of variety, the needed workarounds and FW specific validations for compatibility. That said, if everyone isn’t trying to get LPCAMM2 off the ground then it does fall into a catch 22.

If folks want to sell their 16GB module they’ll need people to sell it to,

This is exactly the problem. I tried to change the poll to prompt that option to make people say who they thought they would be able to sell to, given it is my current expectation that:

There may well be demand from RAM-less systems, failed modules and other niche cases, but outside that fairly narrow set of buyers the natural purchaser starts to look a little mythical.

I wonder if that will change in the future? If someone is running a single stick of 16 or 32, an additional 16 would be desirable.

Here’s hoping, but I suspect it’s wishful thinking. Mainboard real estate is as much the problem here

5

The alternative may have been going bust. We’ve seen some punters (more on Reddit than here) that felt that such an eventuality was moral or justified, but I think each of us need FW to survive regardless of our feelings on whether economic markets and/or capitalism are presently working well.

The need for FW to keep going of course is partly from self-interest: in-date warranties can only be claimed upon if there is a company to receive the claim. But in a wider sense I think FW is doing far more good than harm, even to the degree that other manufacturers are starting their own repairability initiatives.

1 Like
6

Let’s just hope what they learned from this was not that they can get away with it but to do better next time.

1 Like
7

I don’t disagree with your broad sentiment and I’m in the same camp as you, both from net market externality and personal perspectives but:

my point/gripe here was not so much about the decision they made to deal with the shortage (I don’t think what they did is unfair given the position they were in, even if I don’t agree), it was about the choice to go LPCAMM2 over SODIMM in the first place. Similarly, I can see why they did it (also not what I would have done, but I’m also not a tech producer), but one hopes that the situation was that they actively considered the risk of entering into an immature market and got caught with their pants down, as opposed to being ignorant of it and being hopelessly optimistic. Either way, the decision to go LPCAMM2 has shot much of the financial viability of going Panther Lake with them.



I like to think that there is general good faith engagement on the whole



All of that said, while this is all interesting more generally, I mainly made this topic to discuss the eventual market (or lack of it) for the 16GB modules that are being pushed out and/or what the approach that FW took means for it in general, in the absence of a buy-back scheme

8

Many of us on here urged them to do so, myself included although that was pre everything. I maintain that it was the right decision. Prices for RAM can vary and will (:crossed_fingers:) eventually go down. Designing it to accept SODIMM is permanent and nerfs the hell out of iGPU performance. I have not purchased a FW 13 Pro because of what I consider to be a terrible track record regarding firmware support. When situation changes, then I’ll buy. Had I purchased a Pro, I would’ve been the person to sell the 16GB kit, failing that, I would’ve repurposed it. So potentially drawer-ware for some period of time but not forever.

1 Like
9

I think as a decision before everything knowing what was known then, fair, on the basis of performance and power - but for a product with an identity built around the replaceability of parts, the lack of available validated LPCAMM makes it all a hard pill to swallow, especially given that first party is the only way to get that right now (RIP Crucial). Hindsight 20/20 as far as I’m concerned, but appreciate we don’t agree on that.

The core issue with LPCAMM2 16GB (I think) is that it has no meaningful/useful second hand market because almost no one who wants/cares about the performance wants 16GB outside of very niche conditions.
So then, how long before most people are likely to have a system that has an empty LPCAMM slot that needs filling, with a module that is already undercooked - 5? 7 years? Ever? That’s a lot of drawer time.

10

If lpcamm2 catches on and the market doesn’t go on with socamm or something else (given they deliberately want to split the server and consumer market that isn’t too likely). In time there will be relatively cheap minipc barebones that take lpcamm which would be pretty decent hosts for cheap 16gb modules. At the current market prices I’d expect the people that got downgraded to 16 and didn’t cancel to just learn to live with it for a while.

Still much better than systems with 8/16gb soldered, in that case the whole system goes in the drawer/bin.

2 Likes
11

16GB of RAM is arguably excessive if one looks for the correct application. If the board gets converted to homelab use, you can do an awful amount of stuff with “just” 16GB. Additionally, I’d also argue that while what I just offered is niche; FW’s target market has serious overlap with that niche. Take one of them 3D printed mounts to mount a FW motherboard in an ATX case, do what needs to be done in software to prevent sleeping of TB ports, cheap eGPU dock to mount a SATA expansion card and fill the case with drives and boom, you now have a NAS. Add in a disc drive on one of the those SATA ports and it’s a media server. Scrap the SATA and just use a bunch of cheap 10Gb network cards and suddenly it’s a ripping fast router fully capable of routing 10Gb.

Just takes imagination and none of those applications would require more than 16GB of RAM. None of them are economical to do today but they might in 5-10 years. Old enterprise 10Gb equipment is relatively inexpensive today and will continue to plunge in value. HDD prices will fall along with RAM prices.

I’m sure plenty of others can think of other ideas. But no, I don’t think 16GB is worthless/useless today.

3 Likes
12

But this is arguing exactly my point.

I completely agree that 16GB is useful for plenty of things. I’m basically the person you’re describing: I run a NAS, HTPC and HAOS box in addition to a desktop and laptop. In most of those non-daily-driver applications I might care about memory efficiency far more than bandwidth, but neither at the expense of the base capacity needed.

Even within that niche, though, how many of us are building an extra system because we genuinely need one rather than because we can/for fun? I’m often the latter. And how many NAS/router/HTPC-type workloads actually care meaningfully about LPCAMM2 over SODIMM?

More importantly, even the uses you’re describing probably only become economical (or perhaps useful at all) in 5–10 years, and that assumes enough barebones LPCAMM2 systems exist by then to absorb these modules.

That’s the bit I’m getting at. Framework’s chosen approach is potentially directionally worse for the RAM shortage in the short/medium term (with long as an unknown): some customers whose actual requirement was 32GB+ will first consume a 16GB module, then later consume a 32/64GB module as well. For some period, both have been manufactured and purchased to satisfy one original requirement, and the now displaced 16GB may be sitting idle either for an extended period, or indefinitely/e-wasted.

If it eventually finds a NAS/router/mini-PC home, great - that is obviously much better than permanent e-waste. But future reuse does not undo the extra demand that occurred during the constrained period. The policy is effectively accepting a near-term inefficiency in the hope that a sufficiently liquid reuse market develops later.

I don’t think 16GB is useless. I’m questioning whether its time-to-reuse and secondary-market demand are remotely matched to the timing and volume of modules being displaced. If not, the downgrade adds demand pressure now and hopes to recover that efficiency later through fairly niche reuse cases.

13

I believe this to to be the case. The things that FW theoretically need to improve upon are business decisions (which technology to support) or economic predictions (expecting a 15% price rise and getting 100%), and here in the threads, we’re commenting with the benefit of hindsight.

Modern business is all about iterating on improvements, moving quickly enough to experiment, and learning from mistakes. I think FW is doing fine on those measures.

14

None to my knowledge, but I do care about iGPU performance in the here and now. Switching to SODIMMs would nerf performance into the ground. If I’m doing light to moderate gaming, I’m going to want the extra performance. The improved power efficiency is just icing on the cake.

You can’t run a business with no cash flow. Framework made the best decision they could with the knowledge they have. They need to sell laptops, they didn’t predict costs would rise like this. What would you have them do?

I see the point you’re making and I don’t intrinsically disagree (although I will say that you know exactly as much about future use of the modules as FW does).

I promise you this however, if we presume that the next generation of boards also use LPCAMM, then anyone who wishes to upgrade their board and RAM capacity will simply sell both at the same time. We can further assume that there will be a number of people who upgraded RAM first and will carry the old RAM to the new board, leaving old boards with no RAM and a ready market for 16GB modules to be snapped up on the cheap. The market will sort itself out. The price merely needs to drop to a point where people buy, whatever price level that may be.

Those of modest needs might even buy new mainboards sans any memory, confident they can save a buck buying used 16GB modules.

I’m just saying I can easily envision these modules getting used in a variety of ways.

15

They put like terabytes of lpddr5x on on ai servers and then put them in ai datacenters by the rack, frameworks whole ram consumption probably doesn’t even register in the overall supply. (the main constraint is supply of the lpddr5x chips, putting them on modules is a lot easier so it is much easier for extra parties to show up that can fulfill demand if there is more than the traditional suppliers can handle even if it is a bit more complicated than making sodimms)

For the secondary market having a surplus of available lpcamm2 modules (even small one that are less desirable) would at least make the platform slightly more desirable as you are more likely to get access to some memory more easily.

16

As I’ve said, I see why they have done what they’ve done. And yes, sure, the market will sort itself out - but on what time horizon, and to what short/medium term detriment +/- exacerbation of length of the current problems?

Whilst this is true in the context of the broader market, outright non-availability of SODIMMs is not what we are seeing relative to LPCAMM2, and so it is still relevant to this corner of the market, given the fork was about the architectural choice to begin with. This is also why I’ve said this to be directional with caveats about scale.

I agree that iGPU performance is probably the strongest argument for LPCAMM2. The question then becomes how large the overlap is between people who specifically value that bandwidth and people for whom 16GB is an attractive long-term capacity. That cohort obviously exists, but I suspect it is relatively small (…admittedly I am probably biased in my numbers/reference points here though because I’m running 4K on my desktop and can routinely get into the mid-to-upper 20s of system RAM, esp if I’ve left things running in the background). I’m not convinced it is large enough to create much natural demand for loose second-hand 16GB modules.

Exactly, and I think that uncertainty cuts both ways. I’m not claiming these modules will definitely become permanent drawerware, but the plausible time horizon is still a decent way away given the general lack of LPCAMM systems out there in the wild. Whilst the eventual market may prevent physical waste, it doesn’t undo the extra demand that occurred during the constrained period. And then there is also the LPDDR6 elephant too potentially complicating matters given the probable time horizon, which then would lead into:

They may well end up that way, but that still depends on enough people upgrading boards on the relevant timescale, selling the old boards RAM-less, and wanting to carry their existing higher-capacity (if they could get it) LPCAMM2 forward. I can see that creating some future demand for cheap 16GB modules, I’m just not convinced we can assume it will be large or timely enough to absorb the downgraded cohort (not to mention if selling together, will people have upgraded earlier because of RAM constraints? But that’s a whole other kettle of fish).



We’re really all just probably different sides of the same coin :slight_smile: I’m just less optimistic about the easiness of the eventual outcome, let alone anything nearer-term

1 Like
17

For those who have voted for either

  • Upgrade later and sell the 16GB module
  • Remove Framework RAM / source third-party RAM

can you please state to whom/which buyer/when, or where you are intending to get the third-party RAM/from what stock?

18

Well, it’s upgrade “later”.
Once third party ram hits the market at a reasonable price.

1 Like
19

Framework could have blocked the pre-order. For example, making batch 5 only available for pre-order after shipping batch 1, instead of dishonoring the “They are actually going to get the configuration they ordered at the price they ordered it at” on the nextgen event. If Framework actually did that, batch 1 to 4 would still be the same and customers ordering batch 5 and later were still going to consider the elevated RAM prices in a transparent manner, and Framework still wouldn’t lose money by absorbing the cost. If batch 5 and later were made unavailable for pre-order, customers would’ve bought and got the RAM or other (non-pre-ordered) computers before the price hike instead of putting all eggs into the same busket.

2 Likes
20

I don’t think they would have gone back on this without a lot of deliberation and perception that this was a last resort/close to it. Preorders for later batches were already in by the time the announcement was made (I’m batch 16 and I ordered prior to the announcement), so the hypothetical requires them to have known the magnitude of price increase before, unless you have evidence of them knowing earlier and were dishonest about it.



Given you’d be tolerating the lower capacity in the meantime, to what degree will it affect you?

21

A bit more swap space use so things might lag a little more. But I have survived on 16 for 3 year with my last framework.

2 Likes