RAM prices expected to rise another 40-50% in Q3 2026, and then 30% more in Q4 as AI demand outpaces supply

Because they will be the only ones capable of making the economics make sense.

This is something you don't seem to understand, this has the potential to kill off huge portions of the DIY market. ASUS, Gigabyte, MSI, and ASRock are reportedly facing a projected 28% contraction, with ASUS alone expected to sell millions fewer boards. And then there's all the accessory companies like Corsair, Thermaltake, Thermal Grizzly, etc. They depend on people actually building, upgrading, tweaking, cooling, and maintaining PCs. If the “normal enthusiast build” disappears, those companies either shrink, pivot, or die leaving nothing but pre-builts that don't necessarily need externally built accessories, they can be made in-house.

Your field of vision is very narrow, I'm looking out across the horizon.
If you were looking across the horizon, you would see the corpses of every other hardware rush out there where people decried the end of DIY and see that maybe you shouldn't listen to chicken little.
 
I haven't studied English, and I don't think that's enough.
On that we can absolutely agree.
I think I'd better quietly leave this forum.
Or just follow through on your comments. You make an argument then totally switch your line of reasoning in a response. This isnt a language issue.
But you shouldn't address someone in such a familiar manner.
If you leave a comment, others can respond to it. Welcome to the big boy internet.
 
It's not what I said at all. I believe demand has not risen beyond supply, it's just fake demand sustained by the bubble
Again: learn economics. It doesn't matter whether you believe these buyers are "in a bubble", or if their purchases are ill-advised: they are still buying RAM. There is no "fake" demand. Stop deluding yourself.

I believe it's plain market manipulation...
Please explain how Micron and Samsung are forcing literally every business on the planet to invest in AI. Do you imagine armies of little invisible RAM-bots invading the minds of world-wide CEOs?
 
If you were looking across the horizon, you would see the corpses of every other hardware rush out there where people decried the end of DIY and see that maybe you shouldn't listen to chicken little.
Then tell me how the heck any of this will continue when projections state that RAM prices won’t come down until at least 2030? That’s four years of plummeting sales for just about anything PC hardware related!

At this point, only the very big companies will survive leaving a graveyard of smaller boutique hardware suppliers. Companies like Thermal Grizzly and other smaller brands probably won’t survive this or end up being bought up by bigger companies.

The likes of Dell, HP, Lenovo, Apple and other such big OEMs are most definitely going to survive. Sure, Asus and Gigabyte will as well but as this RAM shortage continues on, watch them pull back on enthusiast boards and concentrate on supplying the big fish.
 
AI cannot continue to greedily consume everything without making the golden sht promised to investors.

Not buying tech in the next 18 months, like many people, will be a great move.
 
Again: learn economics. It doesn't matter whether you believe these buyers are "in a bubble", or if their purchases are ill-advised: they are still buying RAM. There is no "fake" demand. Stop deluding yourself.


Please explain how Micron and Samsung are forcing literally every business on the planet to invest in AI. Do you imagine armies of little invisible RAM-bots invading the minds of world-wide CEOs?
Hey look, you are defending price gauging, price fixing and fraud again. Who would have guessed that the guy who thinks rich people shouldn't have to pay their fair share of taxes is once again defending the rich in the detriment of normal people.

Do you really think that these "CEOs" didn't talk it out before raising prices by 5-10x? That's just pathetic dude.
 
Again: learn economics. It doesn't matter whether you believe these buyers are "in a bubble", or if their purchases are ill-advised: they are still buying RAM. There is no "fake" demand. Stop deluding yourself.


Please explain how Micron and Samsung are forcing literally every business on the planet to invest in AI. Do you imagine armies of little invisible RAM-bots invading the minds of world-wide CEOs?
Again, I wasn't saying that, there's no fake demand if they're buying, duh, I'm saying they're not, the whole bloated industry is based on intent not actual sales and so they create scarcity and price fixing and all that. And I'm not saying there's no demand or little or some, it's big, just not as it is presented. But hey, we'll see in a couple of years who's right. The only real competition can come from the Chinese, but they re not there yet. And then you'll see all this bullsh1t unfold and probably the whole bubble will collapse. Here's some strategies: slowing new fab investments, allocating supply strategically, or using inventory builds/draws in coordinated ways, trade associations, public forecasts, or analyst calls can align expectations without direct deals. One firm announces increases; others follow without explicit talk. Public statements about "market conditions", "costs" can serve as signals, each limits production so total supply stays below demand, pushing prices up. With sufficient capacity, this is artificial scarcity. Ofc there's just theory. But why is it so inconceivable that they are actively doing it when they have total control atm? You think market is fair when oligopolies rule?
 
Last edited:
Again, I wasn't saying that, there's no fake demand if they're buying, duh, I'm saying they're not, the whole bloated industry is based on intent not actual sales and so they create scarcity and price fixing and all that. And I'm not saying there's no demand or little or some, it's big, just not as it is presented. But hey, we'll see in a couple of years who's right. The only real competition can come from the Chinese, but they re not there yet. And then you'll see all this bullsh1t unfold and probably the whole bubble will collapse. Here's some strategies: slowing new fab investments, allocating supply strategically, or using inventory builds/draws in coordinated ways, trade associations, public forecasts, or analyst calls can align expectations without direct deals. One firm announces increases; others follow without explicit talk. Public statements about "market conditions", "costs" can serve as signals, each limits production so total supply stays below demand, pushing prices up. With sufficient capacity, this is artificial scarcity. Ofc there's just theory. But why is it so inconceivable that they are actively doing it when they have total control atm? You think market is fair when oligopolies rule?
What you said reminded me of this news article title from last year:

"Memory makers have no plans to increase RAM production despite crushing memory shortages"

And only in june of this year did SK hynix "announce" that they will increase production "by 2030".
 
What exactly does Microsoft's forced obsolescence of perfectly good hardware have to do with this?

There's no grand conspiracy here. Microsoft wasnt trying to "save" consumers from the AI price jacking. They wanted you to pay them more money to replace perfectly good hardware. This was criticized because it was a blatant money grab.
How does buying new hardware help Microsoft directly? There are a lot more suppliers of hardware than Microsoft and most of Microsoft's computers are premium priced.
 
Historically, memory prices spiked because PC or smartphone demand outpaced supply, manufacturers expanded production, and eventually oversupply drove prices back down. AI has changed that equation. Hyperscalers are deploying AI infrastructure that consumes enormous amounts of DRAM and HBM, and that's not a one time buying spree...it's becoming an ongoing source of demand.

I do expect prices to come down from whatever peak they reach. Markets rarely move in one direction forever. What I'm not convinced of is the idea that they'll return to 2024-2025 pricing anytime soon. Looking at GPUs, CPUs, and other hardware over the last decade, prices often settle at a new baseline after a major market shift rather than reverting to where they were.

Unless AI investment slows dramatically or memory manufacturers create enough excess capacity to flood the market again, I think "slightly lower" is more realistic than "back to normal."
 
Last edited:
Oh there is plenty of ram. The memory cartels funded by GPU cartel, Nvidia, decided to lock in prices till 2028 and most likely till 2030. Government which is supposed to stop it is actually profiting from this deal and the cycle of corruption goes full circle.
 
Again, I wasn't saying that, there's no fake demand if they're buying, duh, I'm saying they're not, the whole bloated industry is based on intent not actual sales
They're actual sales of actual RAM chips.
...And I'm not saying there's no demand or little or some, it's big, just not as it is presented.
This is what psychologists term a "security belief": based not on logic and evidence, but your own wants and desires.

you reminded me of this news article title from last year:

"Memory makers have no plans to increase RAM production despite crushing memory shortages"
Yet again you fell for a gaslighting news article, and turned off your critical thinking skills. What the truth is (and what that THG article itself said, had you read more than the headline) is that memory makes are planning modest increases to traditional RAM production, and far larger increases (600% above current levels) for HBM memory.

Given that 100% of the price pressure on traditional RAM is coming from the AI & DC segments buying this because HBM memory is unavailable, this will certainly return prices to normal.
 
They're actual sales of actual RAM chips.

This is what psychologists term a "security belief": based not on logic and evidence, but your own wants and desires.


Yet again you fell for a gaslighting news article, and turned off your critical thinking skills. What the truth is (and what that THG article itself said, had you read more than the headline) is that memory makes are planning modest increases to traditional RAM production, and far larger increases (600% above current levels) for HBM memory.

Given that 100% of the price pressure on traditional RAM is coming from the AI & DC segments buying this because HBM memory is unavailable, this will certainly return prices to normal.
This is what happens when you are tech and economically illiterate but still want to give your opinion on things.

All manufacturers did was pivot production from one product to another. They didn't build any new facilities.

Only recently SK Hynix and Samsung announced plans to build new manufacturing facilities over the course of the next decade (2030-2035), and they specifically mention production of HMB modules. And the 2027 facility that SK Hynix mentioned earlier this year is just for advanced packaging, not chips.

"this will certainly return prices to normal" - Tell us when and what "normal" means to these bastards who priced fixed their products. Will the current prices drop by 80%? rhetorical question, we already know the answer.

FYI
"According to industry projections cited by SK Hynix, the HBM market is expected to expand at a compound annual growth rate of 33% between 2025 and 2030." - even with increased production of HBM, prices won't go down. We are talking about a demand growth of 4-5x by 2030.

Learn how to use context and facts dude... it's just embarrassing to have to correct you every single time you write something.
 
Last edited:
And they won't do that. It happened before causing prices to free fall. They hated that and from this point on, they'll make sure that that never happens again.
I agree somewhat.

The big memory manufacturers learned some expensive lessons from previous crashes and are much more disciplined about expanding production today.

However, no company can completely prevent a future oversupply if demand falls or new competition ramps up.

I agree with the broader point....I think it's far more likely prices settle at a higher baseline than return to the lows we saw before the AI boom.
 
I agree with the broader point....I think it's far more likely prices settle at a higher baseline than return to the lows we saw before the AI boom.
This is exactly what will happen. RAM will never go back to the dirt-cheap, rock-bottom prices we saw before this whole mess. It just won't.
 
This is what happens when you are tech and economically illiterate but still want to give your opinion on things.

All manufacturers did was pivot production from one product to another. They didn't build any new facilities.
Why do you pollute every thread with this nonsense? Micron broke ground on two new fabs in Idaho in 2024 (one scheduled for production in 2027, the other 2028). In 2025, Samsung opened one new DRAM/NAND fab (P4 Pyeongtaek), and SK Hynix's new Fab M15X also began production in Dec 2025 -- four months earler than planned.

But it isn't simply the new fabs that count, but the production expansion occuring at existing fabs. In 2024, SK Hynix was averaging 350K DRAM wafer starts/month, currently it's in the 550K range. In 2024, Samsung was averaging 410K-430K DRAM wafer starts/month, but capacity expansion has now increased it to over 660K starts/month.

Total DRAM production in 2023 was roughly 120B gigabytes. In 2024 it was 160B and in 2025 180B. It would have risen faster, but cutting-edge DRAM fabs are not Romanian cabbage rolls. They take years to build.
 
Last edited:
Why do you pollute every thread with this nonsense? Micron broke ground on two new fabs in Idaho in 2024 (one scheduled for production in 2027, the other 2028). In 2025, Samsung opened one new DRAM/NAND fab (P4 Pyeongtaek), and SK Hynix's new Fab M15X also began production in Dec 2025 -- four months earler than planned.

But it isn't simply the new fabs that count, but the production expansion occuring at existing fabs. In 2024, SK Hynix was averaging 350K DRAM wafer starts/month, currently it's in the 550K range. In 2024, Samsung was averaging 410K-430K DRAM wafer starts/month, but capacity expansion has now increased it to over 660K starts/month.

Total DRAM production in 2023 was roughly 120B gigabytes. In 2024 it was 160B and in 2025 180B. It would have risen faster, but cutting-edge DRAM fabs are not Romanian cabbage rolls. They take years to build.

RISK PRODUCTION is at BEST next year we are 2 MORE years away from TODAY for ''full production'' assuming the plants even complete full construction. I 1000% guarantee those plants will CLOSE DOWN and lay off all workers in the next 6 years when *****s in the current administration realize the increase in materials cost will kill the demand and thus idle the plant and then shut it down as manufacturing will swing back overseas for cheap labor at the behest of billionaire investors in the US.

Book mark this this post .. come back in 6-7 years and tell me how right I am.

things were cheap for a reason, now they will pay for ::
UNIONS scam artists
US WAGES at 4-5 times the cost
tariff increase material imports
US costs of grid power fighting AI data center prices
US costs of water fighting AI data center price
loosing a trillion DOLLARS in the from of ''tax assistance, or tax incentives'' to have someone build a FAB

The ONLY people who will make money on the new fabs will be anyone who owns construction contracts, owns land rights and who sells the stock early before the FAB crash of 2032/2033.
 
Last edited:
RISK PRODUCTION is at BEST next year we are 2 MORE years away from TODAY for ''full production'' assuming the plants even complete full construction. I 1000% guarantee those plants will CLOSE DOWN and lay off all workers in the next 6 years when *****s in the current administration realize the increase in materials cost will kill the demand and thus idle the plant and then shut it down as manufacturing will swing back overseas for cheap labor at the behest of billionaire investors in the US.

Book mark this this post .. come back in 6-7 years and tell me how right I am.

things were cheap for a reason, now they will pay for ::
UNIONS scam artists
US WAGES at 4-5 times the cost
tariff increase material imports
US costs of grid power fighting AI data center prices
US costs of water fighting AI data center price
loosing a trillion DOLLARS in the from of ''tax assistance, or tax incentives'' to have someone build a FAB

The ONLY people who will make money on the new fabs will be anyone who owns construction contracts, owns land rights and who sells the stock early before the FAB crash of 2032/2033.
"1000% guarantee."
That's usually the part where I stop expecting evidence and start expecting fan fiction.

You're arguing against fabs that aren't even in full production yet to explain ram prices that are rising today because AI demand is consuming available memory supply. Those are two completely different timelines.

Then you toss in unions, wages, electricity, water, tariffs, tax incentives, billionaires, overseas labor, and a stock market crash...all in one post...as if saying more things somehow makes the prediction stronger.

Modern semiconductor fabs aren't 1975 assembly lines. They're some of the most automated manufacturing facilities on earth. Labor isn't even close to being the dominant cost driver. If high wages alone determined where fabs could exist, companies wouldn't be building them in the US, Europe, Japan, or South Korea.

Could some projects fail? Sure. Could subsidies end up being a bad investment? Absolutely. That's a reasonable debate.

But "every fab will shut down by 2032 because I said so" isn't analysis. It's a screenplay.

And the best part? You've already built yourself an escape hatch. If one fab succeeds, you'll just claim it only survived because of subsidies. If prices fall, you'll say it wasn't because of new capacity. If prices rise, you'll say you were right all along. That's not a prediction....it's called an unfalsifiable narrative.

By all means, bookmark the thread. Just remember that confidence isn't evidence, and typing "1000% guarantee" doesn't magically turn speculation into economics.

See what I did there ...
 
Back