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

Server DRAM prices surge 50% as AI-induced memory shortage hits hyperscalers

tomshardware.com

Post reply on HN
Print view View original

111–120 of 129 posts

Re: Server DRAM prices surge 50% as AI-induced memory shortage hits hyperscalers

#111

Post by gruez »

binarycrusader wrote: post #109
gruez wrote: post #82

Earlier quoted context omitted.

>Pet peeve: Contrary to a persistent popular belief, inflation != currency debasement. Not to mention that "debasement" doesn't make sense anymore given that there basically aren't any currencies on the gold standard anymore. At best you could call a pegged currency that was devalued as being debased (with the base being the pegged currency), but that doesn't apply to USD. "debasement" therefore is just a pejorative…

It might be subjective, but doesn't this count at least partially as a currency on the gold standard? https://en.wikipedia.org/wiki/Zimbabwean_ZiG

From wikipedia:

>A gold standard is a monetary system in which the standard economic unit of account is based on a fixed quantity of gold.

and

>The Zimbabwe Gold (ZiG; code: ZWG)[3] is the official currency of Zimbabwe since 8 April 2024,[2] backed by US$900 million worth of hard assets: foreign currencies, gold, and other precious metals.

>...

>Although the rate of devaluation of the ZiG may vary,[13] the ZiG has consistently lost value since its introduction, and its long-term prospects are dim so long as large grain imports continue and the government continues to overspend.

sounds like it's not "fixed" at all, and "backed by ... hard assets" just means it has central bank reserves, which most fiat currencies have.

Re: Server DRAM prices surge 50% as AI-induced memory shortage hits hyperscalers

#112

Post by nubinetwork »

654wak654 wrote: post #28
Ekaros wrote: post #25

I hope this AI craze will crash soon enough. Maybe then various things normalize in price again. And consumers get cheaper products with less limitations.

Best we're getting is probably a stop to the price raises, but no price cuts. Kids will continue to grow up not knowing a $600 flagship GPU or a $1000 gaming PC.

The Radeon 8500 from 2001 was even cheaper than that, roughly 300 USD... the voodoo 3 3500 from 1999 was roughly 200 USD... if you ask me, we don't need graphics chips as intense as we have now, but the way they crank them out every year and discontinue the older models ruins most of the value of buying a GPU these days.

Re: Server DRAM prices surge 50% as AI-induced memory shortage hits hyperscalers

#114

Post by muldvarp »

bigbadfeline wrote: post #90
muldvarp wrote: post #81

Earlier quoted context omitted.

> We need more of this kind of optimization. Who is the "we" in this sentence? The ultra-rich that don't want to pay white collar workers to build software? The advantages of LLMs are tiny for software engineers (you might be more productive, you don't get paid more) and the downsides are bad to life-altering (you get to review AI slop all day, you lose your job).

> The ultra-rich that don't want to pay white collar workers to build software? This is already a fact and it's set in stone - making AI cheaper won't change anything in that regard. However, a cheaper AI will allow the laid-off software engineers to use models independently of those firing them, and even compete on an equal footing.

Compete for what exactly? Under the assumption that AI agents will make human software engineering obsolete there won't be a market for you to compete in. Everyone that wants a piece of software will ask their AI agent to create it.

My ability to create software is only really useful to me because other people pay me for it. If AI agents take that from me, it won't matter that I can now create awesome software in minutes instead of weeks. The software was never really the thing that was useful for me.

Re: Server DRAM prices surge 50% as AI-induced memory shortage hits hyperscalers

#116

Post by bigbadfeline »

muldvarp wrote: post #114
bigbadfeline wrote: post #90

Earlier quoted context omitted.

> The ultra-rich that don't want to pay white collar workers to build software? This is already a fact and it's set in stone - making AI cheaper won't change anything in that regard. However, a cheaper AI will allow the laid-off software engineers to use models independently of those firing them, and even compete on an equal footing.

Compete for what exactly? Under the assumption that AI agents will make human software engineering obsolete there won't be a market for you to compete in. Everyone that wants a piece of software will ask their AI agent to create it. My ability to create software is only really useful to me because other people pay me for it. If AI agents take that from me, it won't matter that I can now create awesome software in min…

> Under the assumption that AI agents will make human software engineering obsolete

That assumption is wrong, humans will always have to guide the bots to achieve human-worthy goals, be it as supervising engineers, small business owners or a combination of the two - that helps competition and avoids lock-in.

Re: Server DRAM prices surge 50% as AI-induced memory shortage hits hyperscalers

#117

Post by wiredpancake »

jonas21 wrote: post #77
654wak654 wrote: post #28

Earlier quoted context omitted.

Best we're getting is probably a stop to the price raises, but no price cuts. Kids will continue to grow up not knowing a $600 flagship GPU or a $1000 gaming PC.

For $50, kids these days can buy a Raspberry Pi that would have run circles around the best PC money could buy when I was a kid. Or, for $300, you can buy an RTX 5060 that is better than the best GPU from just 6 years ago. It's even faster than the top supercomputer in the world in 2003, one that cost $500 million to build. I find it hard to pity kids who can't afford the absolute latest and greatest when stuff that…

[dead]

Re: Server DRAM prices surge 50% as AI-induced memory shortage hits hyperscalers

#118

Post by autoexec »

some-guy wrote: post #57
zonkerdonker wrote: post #39

Earlier quoted context omitted.

It really is a damn shame, but before AI, it was cryptomining. Desktop GPU prices have been inflated to nonsense levels for gamers, to the point where console vs. PC isnt even really question anymore.

In some ways though, the increase in visual fidelity has been _marginally_ improved on a per-year basis since the PS4/Xbone era. My GPUs have had much, much longer useful lives than the 90s/early-2000s.

AMD just tried to get away with stopping support for cards that were still being sold new in stores. Nvidia cards are just getting worse and more expensive over time (https://www.xda-developers.com/shrinkflation-is-making-nvidi...).

Part of what made PC gaming in the late 90s/early 2000s so exciting was that the improvements were real and substantial instead of today where we're stuck with minor improvements including bullshit like inserting fake frames generated by AI, and the cards back then were usually pretty easy to get your hands on at a normal price. You might have had to occasionally beat your neighbors to a best buy, but you didn't have to compete with armies of bot scalpers.

Re: Server DRAM prices surge 50% as AI-induced memory shortage hits hyperscalers

#119

Post by KurSix »

This shortage is the best thing that could have happened for R&D in model efficiency. The "who has more parameters" race is about to hit the physical wall of hardware availability. Now the real race for efficiency begins: quantization, distillation, Mixture-of-Experts, new architectures

Hardware constraints are the single biggest driver of software innovation

Re: Server DRAM prices surge 50% as AI-induced memory shortage hits hyperscalers

#120

Post by KurSix »

Ekaros wrote: post #25

I hope this AI craze will crash soon enough. Maybe then various things normalize in price again. And consumers get cheaper products with less limitations.

Even if the hype around LLMs dies down, the demand for AI compute won't disappear. It will just shift from giant language models to more specialized areas: computer vision, scientific computing (like AlphaFold), drug discovery. All of that requires massive amounts of hardware
Post reply on HN