I want to believe

6 will actually be “RAM suppliers buy the stock in bulk and sell as refurbs, keeping prices up”

They aren’t PC components. They’re server GPUs.

They don’t even have ROPs anymore. There was a time when the Nvidia P100/A100 could technically game, but that’s generations ago.

FYI, during the crypto busts, Nvidia bought back GPUs and basically incinerated them to keep demand for new stuff up.

…I’m pretty sure that’s what they will do again.

It’s frustrating to see the tech industry take this path, because there were alternative paths that didn’t lead us here. We are in a selfish and shortsighted time.

So you’re not 100% incorrect, but does it happen in 2027? 2028?

Logistically it’ll probably happen eventually, but when?

OpenAI is likely to be the first domino to fall that causes the whole bubble to pop. They’re burning over $10-15 billion a year in losses, and have shown that any attempts to scale only increase that burn rate. They’re running out of investor money to toss into the fire.

And they just announced a last ditch free unlimited for everyone plan to…I dunno drum up interest or hope it converts to paying customers or something. So their costs are about to skyrocket even more, accelerating their death spiral

Just like how GPUs became cheap and plentiful after mining crypto became passé, right?

…They actually did.

I got an AMD 7950 is a crypto bust. Then a Nvidia 980 TI in the next one, dirt cheap.

Nvidia figured it out in the most recent bust though, and controlled supply fairly well.

After the last crypto bust, prices stayed up though. They’re still way higher than before the last crypto boom.

And nVidia, AMD, Intel all know better now than to overproduce consumer GPUs.

PC components flood the market

*Server-grade purpose built AI components flood the market

SXM to PCIe adapters already exist hehe

Not to mention my dumbass would chuck money at a whole server just to throw it in my home lab.

Its not trivial.

Check the level1 forums, where some electronics gurus are trying to get a single adapter MI250X to even boot. The problem isn’t the adapters, but firmware that makes it difficult.

I see no problem if I can snag some generic Instinct accelerators, and a server to go. I’ve already got half a rack full.

Eventually, lenders call their margins

Lending to companies doesn’t work like that.

Softbank’s $40bn loan will be called on March 25, 2027. If they can’t roll this over it’s the banking equivalent of a margin call.

HBM and Vera Rubin are not useful for PC Gaming, at least not for current architectures. Main issue is that DDR5 production simply got cut to make more HBM. But there will be a huge amount of LPDDR, NVME / SSD drives and refurbished Hard Drives available, and hardware for running inference at home for pretty powerful models will become cheap as fuck, liberating us all from the hyperscalers; the open weights models from china are available for everyone.

At least i pray it will be this way.

But aren’t the wafers uncut? It could be used for anything if they’re just in storage.

If the memory wafers are largely HBM, so they can’t be used for gaming hardware.

If you’re thinking of HBM gaming GPUs, the chip design pipeline is too long. If Nvidia/AMD/Intel started making one right now, and rushed it, it would still be many years away.

…However, a lot of LPDDR5 is being made too.

Hence the rumors of AMD hedging their bets and using LPDDR for some future GPUs.

Could gpus use HBM? I thought AMD used it 10 years ago in theirs.

I’m not sure I worded this right before, but it takes ~half a decade to plan and produce a GPU.

They can’t just switch out the bus or memory on an existing design.

So they could make a HBM gaming GPU. But they would have had to have started ~2 years ago.

…That’s extremely unlikely.

It’s unlikely they’ll even plan one now.

AMD did indeed try making HBM gaming GPUs, and it turns out it’s supremely uneconomical; they immediately pivoted to the opposite extreme (narrow bus + larger cache).

However, LPDDR GPUs appear to already be in the pipe, so that will provide some GPU market relief.

It’s good they did find an alternative

According to Tom’s Hardware, it includes “DDR5 and LPDDR4/LPDDR5 as well as premium HBM.”

I’ve never heard of uncut wavers hitting storage or the market. These things are made to be used for whatever is needed at the moment. If you hear this news regarding “Ram cap sold out”, it means all available wavers they expect to be produced for that year have been allocated to whoever paid for them. If 50% of the wavers go into HBM, 30% into DDR including LPDDR and 20% go into NAND storage (i do not have the exact numbers, but i know that HBM needs around 3x the amount of silicon compared to DDR), then you are done selling your capacity - whatever happens afterwards does not make DDR5 appear magically on the market.

Only thing that could change that would be that the AI bubble crashes before 2027, then the allocation numbers will be changed.

Not sure what you mean with the wafers hitting the market. I was only referring to the specific deal that OpenAI made with Samsung and SK hynix to purchase a metric boatload of 900,000 undiced DRAM silicon wafers per month.

According to Tom’s Hardware it includes “DDR5 and LPDDR4/LPDDR5 as well as premium HBM.”

Honestly I’m not sufficiently familiar with how that works to understand if any of that is consumer grade stuff.

That means they have bought out the wavers - with a specific allocation for HBM/DDR/LPDDR/NAND as they need it. they know what percentages of what they need to reach the balance they use in their racks, and that will be what was ordered in detail.

The DDR5 can be used in common PCs, but it’s a different story with LPDDR. I have heard of an adapter to make LPDDR useable in common PCs, while i haven’t seen anything like that for HBM - and i wouldn’t bet on such an teck being available for the consumer market amytime soon, the architecture is quite different.