Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

2027 memory capacity is reportedly sold out

ign.com

Post reply on HN
Print view View original

451–460 of 479 posts

Re: 2027 memory capacity is reportedly sold out

#451

Post by i_never_listen »

vannevar wrote: post #175
bob1029 wrote: post #141

One unit of HBM capacity consumes roughly the wafer capacity that could have produced three units of DDR5 capacity. HBM dies need to be larger than ordinary DRAM dies because of how the final packaging works. > As discussed previously, the ramp of HBM production will constrain industry supply growth in non-HBM products. Industrywide, HBM3E consumes approximately three times the wafer supply as D5 to produce a given n…

>As the memory industry is still recovering from the challenging environment in 2023, this tight supply environment will help drive the considerable improvements in profitability and ROI (return on investment) that are needed to enable the investments required to support future growth. The problem is that they need actual chips tomorrow. I fear we are reaching the point in the semiconductor industry where, in order t…

Do they really need them tomorrow? Or everyone is locking in supply to ensure that they have the chips if the data center actually gets built in the next few years?

Re: 2027 memory capacity is reportedly sold out

#452

Post by Arodex »

linsomniac wrote: post #417
Arodex wrote: post #404

Earlier quoted context omitted.

You are basically describing predatory pricing or dumping, where the goal is not to be the best but to starve your competitors to build a monopoly and gouge customers later. Uber tried that and failed. Now they cost as much as normal taxis if not more. Airbnb tried that and now cost as much as hotels if not more. And tesla is a trillion dollar company on paper only. It can vanish into thin air in a day.

>You are basically describing predatory pricing I'm describing business . Period. It's incredibly rare to encounter products above a fairly small revenue level (niche products), that haven't come from a company that spent significant time reinvesting profits and therefore being unprofitable. Same for using outside investment. Yes, there are companies who have burned outside investment on things that didn't pan out. L…

Betting massively on onion futures is also "business" period air quotes, and yet it is widely accepted as a bad idea and banned.

https://en.wikipedia.org/wiki/Onion_Futures_Act

And it is just one of several "business" methods that are widely regarded as abusing the rules/manipulation and regulated or outright banned.

Your AI companies are not just reinvesting profit, they are burning a fuckton of predatory investors'money to try to drive adoption, but also competitors to bankruptcy.

Re: 2027 memory capacity is reportedly sold out

#453

Post by i_never_listen »

baby_souffle wrote: post #423
vannevar wrote: post #175

Earlier quoted context omitted.

>As the memory industry is still recovering from the challenging environment in 2023, this tight supply environment will help drive the considerable improvements in profitability and ROI (return on investment) that are needed to enable the investments required to support future growth. The problem is that they need actual chips tomorrow. I fear we are reaching the point in the semiconductor industry where, in order t…

> they're going to have to start selling future chips that cannot possibly be physically produced. This is already happening with gpus, no? Tsmc and McDaniel capacity are sold out into the future. Data center that might not be needed in 1 to 3 years is where they’re headed.

Yes, everything so far points to about half of 2026 capacity being delayed or cancelled. How much of that is due to construction delays or the power requirements being brought online I don't know, but those are the main culprits (plus possibly hubris on the AI companies part regarding demand)

Re: 2027 memory capacity is reportedly sold out

#454

Post by milch »

inigyou wrote: post #343
milch wrote: post #289

Earlier quoted context omitted.

I've been in the same boat since it seems something about the memory controller on my gaming PC's MB is broken, but price wise it feels like it just doesn't make sense. I could get something from eBay relatively cheap but then I'd probably spend an hour disassembling and reassembling just to maybe learn that the eBay unit has some equivalent issue or is slightly broken in some other way. I could old new stock for out…

Last time I replaced an old MB it was only $60. What prices are you seeing? It wasn't the same model or vendor MB but it was a compatible one. Same chipset.

The cheapest compatible one I can see new (and one of the only ones...) is currently at 120: https://www.newegg.com/p/pl?N=100007625%20601413455%20601409...

That's actually not bad, last time I checked the only ones that were in stock cost 300+ (IIRC all x570). My plan then was to maybe get a new case and rebuild it on a micro ATX board which are generally more available and seem to have a few decent options in the ~80 price range.

Anyway, again it's a question of is it worth it spending several hundred and the time it takes to rebuild. It currently "works" fine except it takes 5 minutes to boot and is unable to save any BIOS settings like XMP, fan curves, and so on (and yes I changed the CMOS battery - there's a long thread on the ASRock forums of others having the same issue even after the battery is swapped). The non-optimal settings probably eat a few % of performance but other than that it's mostly annoyance, so it still feels like if I'm going to invest any more in it, it better be more worth it

Re: 2027 memory capacity is reportedly sold out

#455

Post by burnte »

cortesoft wrote: post #255
docjay wrote: post #236

Earlier quoted context omitted.

Why must you increase the price?

The alternative is to have shortages. If there is demand for N chips at $X price, but you only have N/2 chips, then half the people aren't going to be able to buy them. The people who really need the chips, and would be willing to pay a lot more than $X for them, will be competing with people who are only willing to pay $X for them. You will end up getting scalping and shortages and hoarding. Since people know that t…

> The alternative is to have shortages.

This is a myth. The volume of inventory remains the same regardless of the price. When demand is higher than supply you have shortages regardless of price simply because you have unmet demand. If you raise prices, that demand is still unmet, you're now filtering that demand by who is willing to pay more so you maximize your own value and profit. Not even rationing prevents shortages because, again, the demand exists. Rationing simply tries to spread the supply around so more people get some rather than none. Price hikes instead say "who wants it badly enough?"

Re: 2027 memory capacity is reportedly sold out

#456

Post by Forgeties79 »

taneq wrote: post #427
Forgeties79 wrote: post #386

Earlier quoted context omitted.

To some degree sure but how many people/businesses are tearing ram out of their existing hardware to resell?

Depends on the market, innit?

My ram is worth 5x what I paid but if I pull it out I don’t have a functioning computer lol plus so many people have laptops which are increasingly soldered/otherwise permanently locked together, so stripping for parts isn’t always an option. For Macs it’s outright impossible.

Re: 2027 memory capacity is reportedly sold out

#457

Post by niek_pas »

socketnaut wrote: post #426
noduerme wrote: post #328

So LLM profiteers have bought up all the consumer RAM for the next several years in order to deliver more of a service we're already oversaturated with that doesn't turn a profit..? I guess by then there'll be no consumers left with jobs to afford a PlayStation 6, so no problem. Except how will the AI trainers find revenue? If maximum potential is reached by laying off 90% of workers at companies that sell stuff... h…

Someone less cynical might recognize prioritization of technological advancement over video games as a good thing.

Prioritizing art over machines that generate mediocre code is not cynical

Re: 2027 memory capacity is reportedly sold out

#458

Post by undersuit »

embedding-shape wrote: post #354
esskay wrote: post #333

Earlier quoted context omitted.

IIRC this wont be possible as a lot of them aren't being packaged on traditional DDR memory boards. They're being put on things like the MI350X. I'm sure there will be a ton of standard DDR5 sticks, but not so much from AI-only datacenters.

> this wont be possible as a lot of them aren't being packaged on traditional DDR memory boards. They're being put on things like the MI350X Even better, looking forward to having hardware like that at home for "cheap" in the future, compared to what it costs right now :)

The MI350X has a 1000W TDP. Maybe I'd use it in the winter.

Re: 2027 memory capacity is reportedly sold out

#459

Post by zenoprax »

jonhohle wrote: post #294
crote wrote: post #281

Earlier quoted context omitted.

> If demand outpaces supply then there’s a shortage no matter what you do. Demand is elastic. If the price of potatoes rises to $1000 / kg most people will switch to using pasta or rice for their dinner instead. There might be a shortage of potatoes at $1 / kg, but not at $1000 / kg. If my 7-year-old computer were to die tomorrow I would've normally replace the entire thing. It has had a good life, the newer generati…

I just did exactly that. I had an old i7 with 64GB of DDR4 that I bought when it was cheap. DDR5 was so expensive I just bought an AM4 and moved the RAM over. My MacBook is showing its age, though. Hearing that 2027 wont be any better makes me hopeful all the vibe coders can ask for memory efficient implementations to stretch existing hardware a little further.

> Hearing that 2027 wont be any better makes me hopeful all the vibe coders can ask for memory efficient implementations to stretch existing hardware a little further.

I've been thinking about this too. After being forced to work off a laptop with 6GB of RAM for a few weeks it made me appreciate my 32GB at home where I run Electron apps with abandon. It still seems crazy to me that we've normalized running an entire browser for a single program just because the UI frameworks are too hard/unfamiliar to the majority of programmers.

Re: 2027 memory capacity is reportedly sold out

#460

Post by syncsynchalt »

justin66 wrote: post #434
vannevar wrote: post #175

Earlier quoted context omitted.

>As the memory industry is still recovering from the challenging environment in 2023, this tight supply environment will help drive the considerable improvements in profitability and ROI (return on investment) that are needed to enable the investments required to support future growth. The problem is that they need actual chips tomorrow. I fear we are reaching the point in the semiconductor industry where, in order t…

> they're going to have to start selling future chips that cannot possibly be physically produced. …but that would be unethical. cue the laugh track

Contracting for more chips than you can make isn't a fun situation to find yourself in, nobody is doing it for profit (it's actually quite expensive in time, money, and reputation to litigate the fallout).

It happens due to failed estimates (we thought we could start production in January but we couldn't figure out the process until March), upstream supply failures (our surface-mount supplier can't get us enough capacitors), or global market conditions (concrete and steel unexpectedly doubled in price and we can't afford to build the whole building this year).

It happens all the time and isn't your fault (but it is your responsibility to handle).

Post reply on HN