if you're waiting to buy a computer, do it soon. prices are going to skyrocket (again) in 2027 and 2028. especially ram. i've linked to some of my sources in the post.

qImPMgxRpOZiUbxz4hWq@lemmy.world · 2 days

i’m seeing some estimates stating ram prices are going to jump 400% by the end of june 2026. major price increases are inbound.

here is some of the information i’m seeing:

prices may never come down again.

“post consumer computing”

spacegoat@lemmy.world · 16 hours

I’ll just keep stealing them from my local data center

jlow@slrpnk.net · 2 days

Nice try, RAM manufacturer.

django@discuss.tchncs.de · 2 days

No thanks, I’ll buy old stuff, until we are done with this bullshit.

teawrecks@sopuli.xyz · 1 day

If the new stuff gets more expensive, the old stuff will too

LammyToGo32@lemmy.zip · 2 days

DDR3 ftw 💪 no shit, my newest piece of equipment in my homelab (besides the switch) is a ThinkPad T540p (i5 4200M)

Janx@piefed.social · 2 days

No source? Some experts are saying an AI crash will happen sooner rather than later…

jlow@slrpnk.net · 2 days

For multiple years it’s been “any minute now!”

Rivalarrival@lemmy.today · 2 days

Every real estate agent, every mortgage broker, could see the fraud in the housing market long by 2003, but either didn’t recognize it as fraud at the time, or didn’t care. They were all making money hand over fist to prop it up.

Zero down payment, 110% LTV mortgages, seller-paid closing costs, buyer is receiving a $3000 check at closing? ARM set to triple their rate in a year? Completely absurd, but the agent is only getting paid if the deal closes. The mortgage broker is only getting paid if the deal closes. The lender is only getting paid if the deal closes. The seller is only getting paid if the deal closes. Everyone needed this deal to close, so they did what it took to close the deal.

Lenders were offering “Continuing Education” credits that RE agents needed for their licenses. That “continuing education” consisted of seminars coaching agents to encourage these deals.

By 2003, buyers were underwater with their mortgages as soon as they signed them. Yet it took 5 more years for the bubble to actually burst.

The circular financing ponzi scheme behind AI will survive until someone can’t service debt. What happens then is either the collapse of the whole scheme, or the government declaring them “too big to fail” and allowing them to reorganize into a single corporate entity.

Ooops@feddit.org · 2 days

So exactly like an AGI emerging after just one more update.

hyperion626@lemmy.world · 1 day

I mean, AI is getting genuinely terrifying.

Ultragramps@lemmy.blahaj.zone · 2 days

I enjoyed how someone put it:
“Just one more round of funding, bro.”

Mordikan@kbin.earth · 2 days

There is so much AI hype and so much anti-AI hype, that I doubt it really will.

For the last 4 years, we have perpetually been 90 days from both the AI market crashing and AI stealing all our jobs at the same time.

My thought is the AI market won’t crash because especially in data engineering and devops there is an inherent worth. I think there will be a potential downturn in the volume of cash flowing into perhaps after the big players like OpenAI and Anthropic go to IPO, but the market itself will remain stable. Similar to how the housing bubble crash didn’t actually wipe out the housing market (with or without the US govt saving the big players - it still would have survived on its own even if those players did not).

LilB0kChoy@reddthat.com · 2 days

My thought is the AI market won’t crash because especially in data engineering and devops there is an inherent worth.

Did you see Anthropic’s IPO filing? The company is upside down and there’s so much commingling between 3 or 4 companies that when one goes the rest will topple like dominoes.

What won’t crash are the billionaires at the top. See, they’re going to cash out with piles of money and leave the index fund investors (e.g. the every day person) holding the bag.

Mordikan@kbin.earth · 1 day

AI isn’t anywhere close to being a pump and dump scheme as you speculate.

The reality is it would be closer to the dot com bubble (if anything at all). Most of the damage would be in physical infrastructure like datacenters which would be an asset write-off. However, there is tangible profitability in AI which we didn’t see in dot com. That would imply if the demand does not exist (which we know it does looking at DevOps, BI, etc) that should things go south, it would be a market correction and not a market collapse.

Again, the weak point and therefore the most hurt would be overvalued segments which is basically the datacenter infrastructure, but that doesn’t imply a collapse/pop only a downturn and correction.

zebidiah@lemmy.ca · 13 hours

The difference with the dot Com bubble is that we built the modern internet on it’s ruins, all that fibre that went into the ground and stayed dark is now lit up and we are all benefiting from it. The gpus and ram going into datacenters is a consumable resource that will be burned out and useless for hobbyists, these aren’t going to be the next gen of xeon chips hitting the market as soon as someone brings a cheap motherboard to market that sockets them…

I think nothing under the sun is new and we’ve all seen all this before. Specifically where I see a lot of parallels is with GPS hardware manufacturers and also Dropbox.

Back in the day Tom Tom and Magellan and Garmin ruled the world, then iPhone came out with maps and they evaporated. Dropbox was hugely successful and a revolutionary offering, until email providers offered the same thing for free and with more convenience.

The tech is super cool, and it’s only going to get better but don’t pay attention to the fools who run AI companies, they will go down and they will crash the economy. I would rather trust Chinese AI than anything clammy Sam Altman and Wario try to sell me on…

hyperion626@lemmy.world · 1 day

There is no way it doesn’t get propped up by the government at this rate. AI is the next big weapon in, well, everything digital. And they won’t want other countries having the biggest stick.

LilB0kChoy@reddthat.com · 1 day

There is no way it doesn’t get propped up by the government at this rate.

There absolutely is a way. It starts with people hold their governments and representatives accountable.

The AI companies can and absolutely should fail. AI itself isn’t going anywhere. That Genie is out of the box.

hyperion626@lemmy.world · 11 hours

Right, but it’s an arms RACE. Can’t have a race by crippling your best runner.

Mordikan@kbin.earth · 1 day

“It starts with people hold their governments and representatives accountable”.

That is basically a non-answer and doesn’t mean anything. The reality is AI is classified a national security priority and therefore has the US govt to prevent it from running out of money.

You could argue that the US has already lost the AI race to China, though. The race isn’t won by building the gigantic LLMs. It isn’t about 1 million token context windows or 1 trillion parameter models. It’s about sentiment. Deepseek and the other distilled models take a novel approach with MoE and strict use case training, but ultimately its the willingness of people/companies to use the AI that results in the win.

US adoption is still the highest world-wide, but China is starting to take the lead on that and have they have stronger willingness to use it.

LilB0kChoy@reddthat.com · 23 hours

That is basically a non-answer and doesn’t mean anything.

You struggling to understand a statement doesn’t make it a non-answer.

The reality is AI is classified a national security priority and therefore has the US govt to prevent it from running out of money.

Your conflating the current technology AI and AI companies.

but ultimately its the willingness of people/companies to use the AI that results in the win.

Ah, you’re one of those people.

PushButton@lemmy.world · 2 days

Well, when the bubble will -pop- explode, the whole economy is going down - yours too.

You might still want to buy… Sooner rather than later…

BCsven@lemmy.ca · 2 days

It will at some point. The amount the need to charge to make a profit means most people won’t use it

django@discuss.tchncs.de · 2 days

Thanks for updating the post and supplying links. My original statements stands though: I’ll keep buying and using old hardware. Permacomputing will rise!

B0rax@feddit.org · 2 days

Source? Trust me bro?

silver_wings_of_morning@feddit.dk · 2 days

From the infornation he has seen. Did you not read his post?

B0rax@feddit.org · 2 days

Oh sorry, must have missed that part.

FiniteBanjo@feddit.online · 2 days

If enough people trusted him then it would become true, lmao

Supply and Demand

hendrik@palaver.p3x.de · 2 days

Yeah, they probably bought 2 RAM sticks and think they can now manipulate markets with a Lemmy post and turn it into a great investment.

Tolstoy@lemmy.world · 2 days

This kind of posts from random user, especially with those names are a clear sign of cracks in the bubble…

solrize@lemmy.ml · 2 days

By the end of June 2026? Old news.

yardratianSoma@lemmy.ca · 2 days

I mean, I just bought two used systems for under $500 CAD combined, both capable of being used as servers, if I wanted to. I’m more worried about HDD prices, nevermind flash storage. I have no idea when I will be able to afford a 10+ TB HDD in the next couple years, at least.

HelloRoot@lemy.lol · 2 days

I got myself a whole new ARM mini pc for 600$ with soldered 64GB of ddr5 RAM. I think of it like a very beefy rpi5 and use it as my homelab server. They were not very popular and selling leftover stock for the same pre-bubble price.

Ooops@feddit.org · 2 days

You mean they are reaching geo-synchronous orbit after already skyrocketing this year?