This has become like opec's best years. Only there's no cartel at least not openly.
But there's another key bottleneck. Even with all the money in the world, getting those machines that etch the RAM could be a multi year ration shop queue. And they're not making those companies every day!
Yes, except no DRAM maker is taking this as a signal to go deep into debt to double or triple capacity. It takes a minimum of 3+ years to dramatically increase capacity and no one expects this inflated demand bubble to last that long. Because they've seen this cycle before, they'll bump next gen capacity up a bit more than planned, maybe 15-25% because the current windfall profits are enough to build a buffer to absorb the hit if that excess capacity comes online in the next DRAM demand crash.
In the meantime, they'll just apologize to everyone for the "market conditions beyond our control" while banking as much of these crazy profits as they possibly can while it lasts. But deep down they remember they were starving just yesterday and know they'll be starving again tomorrow.
I understand why everyone's pissed at the "evil DRAM makers" but I also remember boasting about how I scored some crazy cheap RAM sticks not so long ago. None of us were shedding tears when they were selling at a loss just to survive.
> Yes, except no DRAM maker is taking this as a signal to go deep...
Incorrect. CXMT is taking this as a signal to open new fabs, create new products, develop new silicon manufacturing techniques, and enter the consumer market worldwide (except USA).
CXMT is a company with heavy state backing and control that started its current trajectory years and years ago. It has infinity money and isn’t allowed to go bankrupt. CXMT is reading political signals (Beijing wants indigenous AI up and down the vertical) as much as market.
> CXMT is a company with heavy state backing and control that started its current trajectory years and years ago. It has infinity money and isn’t allowed to go bankrupt.
This also describes the U.S.A. steel industry. "Heavy state backing" and "not allowed to go bankrupt" often result in particularly terribly-performing industries. If anything, it's notable that China seems to buck that particular trend.
Not sure "only the Chinese model can solve the normies pain" is a great look.
Lately consumers are told capitalism is cheap TVs, phones, and computers while housing, healthcare, education, energy, and food climbed further out of reach. The "bread" part of bread and circuses has shrunk dramatically, "but at least you're getting cheaper circuses" is what they've been saying.
Now you're saying only a heavily state-directed Chinese model is still willing to aggressively finance new fabs and meet demand to provide that? "Capitalism demands we don't expand capacity and don't meet demand' is kinda a tough sell after capitalism has been sold as expanding access. "Consolidation, maximal extraction, shareholder preservation, and AI" isn't going to be a winner. I get it makes business sense and wouldn't be a big deal in the past, but the rapidly changing dynamics hitting nearly every part of daily life across the entire economy feels destabilizing.
We're down to wearing dystopian sci-fi level cheap clothes, can't afford chips/sodas, candy/sweet treats have been enshitified to no longer be fulfilling, we can't afford to call an ambulance. Media is polarizing instead of mass media calming. Heck thrift stores are becoming unaffordable to the low end as the middle now resorts to them.
"You don't have the foods you are used to, decent clothing, shelter, health, or social cohesion but you have a sweet 6 year old phone and flock cameras keeping you safe". Jesus it feels... not good. Even cyberpunk dystopia understood you needed to at least jack people into something.
> Lately consumers are told capitalism is cheap TVs, phones, and computers while housing, healthcare, education, energy, and food climbed further out of reach.
Who says this? The latter are heavily regulated and not exposed to market forces, which is why they didn’t get cheaper, unlike TVs etc.
The result of equivalent deregulation would be slums and shantytowns. But hey, cheap rent! I don't think this is s desirable outcome either.
Also, from the article:
> In other words: you can buy a computer thousands of times more powerful than the best consumer device from 40 years ago, for something like 0.3 percent of the price. No other good in history has experienced a decline in cost on that scale: poor people can now carry around in their pockets computers many orders of magnitude more powerful than what the richest slice of the world’s population could afford a few decades ago.
That extreme level of efficiency increase is something very specific to computers and not something translatable to other areas of the economy.
Okay, you got me. I was reflecting the prudent way long-term players generally interpret the market. But from time to time external macro-economic and/or geopolitical factors may shift. Those changes may persist over time or eventually swing back like tides. Other times someone thinks they've got a new angle or a new technology (the RDRAM format was a notable attempt). Occasionally, the people who always say "But this time it's different" are lucky enough to actually say it one of the rare times it IS different.
Time will tell if CXMT's gambit will pay off. Even if it doesn't maybe their state sponsors will absorb the losses for other strategic reasons.
Everyone knows the commodity market (outside HBM) is going to be margin crushed within a few years… by CXMT, demand swings, and so on.
So nobody cares to fully meet the demand today. They are perfectly happy telling customers (those not willing to sign a 10 year contract paying up front at least) to go pound sand.
They don’t even need an OPEC like arrangement. They are effectively perfectly coordinated already in dismissing customers.
Yes, there are a bunch of bizarre comments in here. "Why aren't they rushing to make more fabs? Don't they know China is using this moment to the enter the market in a big way?" Yes, they do! And building more fabs would make that worse. The last thing you want to do before a state-backed player enters the market in a boom cycle is load up on debt to produce more capacity just before the bust.
It's a bit surprising many here are referencing the memory supply cycles without mentioning this revolutionary new application for memory people call AI. It'd be like talking about weather cycles and climate without mentioning global warming. Just like the planet is not going to get colder on average than it was for the foreseeable future, we're not going to need less memory for the foreseeable future.
China is trying to be vertically integrated, completely independent of outside influences and own the future supply/means of production. Memory and chips are piece of their larger plan. In any event, whatever China brings online will be absorbed - imagine a future where people's home computer has 100 to 1000GB of RAM in one variety or another. Folks are going to want better chips and more memory for years to come, supply will be absorbed.
> this revolutionary new application for memory people call AI.
Most analysts think LLMs will elevate the long-term base RAM demand level. I mentioned 15-25% above prior projections without AI, which I think everyone agrees is highly likely. That's actually a lot because it's an overall market number and RAM goes lots of places other than PCs, servers and high-end mobile (depending on how you segment, 25% overall could be in the neighborhood of doubling PC, server, high-end mobile demand).
Above that range analyst estimates diverge. Some are more bullish, and a few are much more bullish. But everyone's error bars get much wider when the numbers go over 30% overall. It's hard to tease out exactly how much of the current demand bubble will persist in the long-run. Clearly, the current market is distorted by short-term dynamics but which part is base demand and which distortion?
How much consumer AI compute will be on-device vs aggregated in load-balanced clouds? How much RAM will that kind of compute require? Will the market find it's more efficient to consolidate around two or three mega-datacenters or will each frontier lab (and geopolitical block) continue drag racing each other to tie-up future RAM (as much to keep it away from competitors as for their own needs). I don't know. I've been watching this game as an interested bystander for several decades and I wouldn't bet too much of my own money on the most bullish estimates.
I think there will continue to be latent demand from PC, server and laptop users who have put off purchases as hyperscalers skip to the front of the wafer line with large outlays of spending for chips and RAM. This would likely extend out beyond the datacenter demand which analyst are pretty confident will be in place for the next 2 to 3 years, looking at Nvidia estimates alone.
I'm concerned that the regular Joe/Jane won't be getting a GPU or a system upgrade for some time. And then you have people in Africa and elsewhere who can't afford new phones.
There probably is some distortion in that the top spenders aren't terribly against cornering the RAM market and forcing everyone to pay more for local AI due to higher RAM, GPU and CPU costs while squeezing people through their APIs to access AI services, but that's an expensive gambit that depends on local models not getting good enough and frontier models to continue to get better while requiring large swaths of memory and compute, keeping consumers demanding the latest models from the APIs. But if the supply increases and people have more access to RAM, GPUs and CPUs, we'll likely see more local model usage and people doing all sorts of creative work, even training their own models.
I'm bullish, not so much of an overzealousness as having witnessed so many of these computing explosions to not doubt it much anymore. We'll collectively find all sorts of creative uses for these wafers and tech. If I were to bet on it, I'd simply buy a basket of stocks and hold, riding out the noise - nothing too surprising there.
The demand for AI simply doesn't exist at the real prices. It barely exists at the current subsidized rates - Microsoft, Google, Meta, Anthropic, OpenAI are spending hundreds of billions to make mere billions.
And then these data centers will be worthless, future ones won't get built, memory demand will evaporate on the spot.
Very interesting and thoughtful comment. I agree. When demand from hyperscalars drops rapidly after the tide goes out and everyone realizes that they can't continue building out the data centers, then fabs will be left with overcapacity that will flood the market. Wondering what this will mean for local LLMs when good RAM is available for cheap?
I'm yet to see any convincing argument that inference is subsidised in any substantial way. Training and speculative expansion are where the spend is from what I can see.
A few days ago Gemini redid their rate limits, making images/audio/video generation much more expensive, shrunk limits across the board (including a new weekly limit) and added more expensive tiers.
At the moment you can pay $20/month to do thousands of expensive queries a month (involving file uploads, the Pro model, extended thinking), and evidence suggests that heavy users are not profitable.
I agree that heavy users are probably not profitable but that's the way the economics of subscription services tends to work across the board.
I'm arguing that even if inference isn't profitable right now it's not orders of magnitude off. Whatever pricing emerges for models equivalent to current frontier models won't be significantly higher than the current API pricing.
There are already enough small companies without tons of VC money to burn that are serving up nearly-frontier llms at prices lower than the big players are charging. They can't all be subsidising? These are companies without any moat or any IP.
If inference was profitable - they'd tell us. Msft, goog, public companies. They'd break out the numbers and show us, if they were good.
But instead, all we get is known liars going on podcasts and repeating "stylized facts" that aren't literally true about their supposed profitability on inference, from companies losing billions per year in a situation where they don't have to tell the truth.
That is VERY far from a convincing argument that they are profitable. So I can & will safely conclude that the opposite is true.
The Chinese memory makers can scale up but the amount of capacity they currently have is negligible. They're in the same boat as everyone else, it's just that nobody expects them to curtail production on purpose to raise prices even higher, whereas that's expected behavior (according to the article) from the established memory makers.
>Guess his boyish looks and his videos educating outsiders and students about AI contributed..
IDK but if I judge a head of ML, I'd care more about whether he has delivered technically, not if he's good at PR.
The ZIRP era lead to SV ML scene to be full of false prophets who were just good at PR and advertising for their employer because they were popular with the industry hype, than at actually getting shit that works delivered out the door.
It's good that the money and hype bubbles bursts every now and then as only then are the scammers, false gods, and naked emperors revealed and can the house get cleaned of them.
Very good article. I am not referring to the RSS part.
Interesting thing is, much of what AI is now regurgitating is human output, accumulated over the years. Model training dataset. Stuff like Reddit posts, even posts here?
If, say, AI output becomes THE 99% over the next few years, we will enter the era of incestuous inbreeding within AI -when it simply regurgitates its own output.
This is good news. I use a Motorola device and feel it was the best (or at least the least troublesome) among the PRC based brands. Clean UI that's near pure Android..
If they can offer it as choice then hopefully banking apps etc wont get knocked off. And we can have best of both.
Any comments on how the copyright issues are handled in corporate settings? I mean both in terms of staying clear of lawsuit+ ensuring what we produce remains safe from copying
But there's another key bottleneck. Even with all the money in the world, getting those machines that etch the RAM could be a multi year ration shop queue. And they're not making those companies every day!
reply