As a outside observer, NAND and DRAM prices have skyrocket ed with the AI infrastructure boom just as the China-based fabs are coming online.
It is wise for these Chinese fabs to eventually use a very aggressive dumping strategy to price well below cost push out other players forever, especially in DRAM.
But right now it seems they can max out their supply capacity without selling below cost.
Appears to me like China's endless state led (often unproductive) investment in semiconductor manufacturing subsidies (for decades) is about to pay off with some industry dominance soon.
That's the most obvious example of failure of Chinese central planning. That and one child policy were abysmal failures that resulted from shoddy science coupled to effective central authority.
Look at the 12/13/14th 5 year plan (the most recently passed). Do you think they achieved their goals?
If your headcanon is that the CCP is inept because they caused crop failures 60 years ago... you could stand to take a look at what they're doing today.
I guess the summary is as simple as: Good five year plans are great, bad five year plans are terrible.
There are sooo many variables in how one could go about making and executing five year plans. They must have figured out a couple of things that tend to work.
The big difference recently from the past is instead of philosophers its scientists who are making the plans and decisions in China so are willing to course correct instead staying the course despite bad out comes.
I don't know if that tracks, senior leadership was heavily influenced towards implementing the one child policy by the works of Song Jian, who came from a rocketry background and presented a model whereby the population would grow to an unsustainable level unless corrective control was applied.
I think it is unlikely philosophers would have suggested to treat population growth like tuning a PID controller.
If your headcanon is "5 year plans are great because some chinese supplier has cheap DDR-4", I would submit a gentle introduction to history is helpful (i.e. we took a couple irrational great leaps forward from cheap DDR-4 => China owns the RAM market => 5 year plans are the way to go)
I think it’s at least valid to conclude that a nation-level commitment to taking over a specific industry can work. You can gut your competitors and then dominate. It’s the Uber strategy applied at the geopolitical level.
If you could figure out how to get your country to dominate the world economy without also allowing your leaders to commit campaigns of mass-internment and extermination, then maybe you’d have a decent political system.
OP's point exactly: the Great Leap Forward is the classic example of society murdering people to make the line go up every quarter, no matter the cost or the truth.
Great leaders use human resources as resources, that's historically why they're great, acquire territory, build state capacity, both at expenditure of regenerating resources - lives. A few 5 year plans that traded a few million lives to save more millions later. And by million we mean low single digit percent, i.e. historic rounding error that isn't remarkable nor worth the fixation except by muh liberal value types.
There's a reason there was persistent Chinese famines before GLF, and none after, because early industrial policies sorted out land resource management via massive rural mobilization/infra/industrial efforts, i.e. why PRC industrialization % and lifespan was vastly higher than developing peers in 70s... that's all because GLF broadly worked, adding about cumulative 200 milliion lives in terms of extended lifespan and likely ~100m+ in terms of averted famine deaths. Most historically competent Chinese leadership is return to farseeing utilitarianism, willing to trade lives for progress, which always sucks for the people during time of upheaval, but ultimately net good.
Deaths in the Great Leap Forward were heavily concentrated as compared to the Industrial Revolution but the death tolls from IR-related famines weren't really all that far off. Industrialization was messy everywhere.
The Irish Potato Famine alone killed 15% of Ireland vs the GLF killing 5% of China.
That's not a reason not to plan 5 years in advance... is it? Any more than the Potato Famine is a reason we should't have capitalism.
I can't say that I've ever heard the argument that a plan led to a famine therefore we should never make plans, when we have great counterexamples that not planning also led to famine. Feels like learning the wrong lesson here.
[edit] I also think it's worth pointing out that America's response to the Dust Bowl was the Farm Bill, which it could be argued is one of the largest-scale examples of central planning in history. It continues to this day, and is part of the reason Americans pay less as a share of their spending on food than any other country on earth.
People say everyone remembers the hits, nobody remembers the misses - but that is the opposite of true for government. Everyone remembers the misses, nobody thinks twice about the hits.
The Irish Potato Famine was a constructed starvation by England to Ireland. All of their real foodstuffs were being stolen by England to run world-level wars everywhere, and the Irish grew what they were permitted to. Potatoes.
The CEO reads the WSJ and their McKinsey report and says “We must leverage AI to keep up and accelerate innovation!” Each division’s executive vice president sets a current year goal to use AI to increase productivity and simultaneously decrease cost. The VPs below them are given a goal to increase their department productivity by 30% by the end of Q3. The middle managers require line managers to train all individual contributors on AI (specifically LLMs) by the end of Q2 to get their bonus. Line managers email their direct reports links to LinkedIn courses and set a deadline by the end of Q1. When the end of the year comes AI has not raised productivity by 30% and AI spending has increased costs. To make Wall Street happy 10% of employees are laid off.
For decades the economic planners of the west have used the most effective tool at their disposal: the business class inflight magazine. Only recently surpassed by the VC telegram group chat.
This has nothing to do with 5 year plans, but with having a functional and competent government enable to enact a coherent long term policy.
In western countries every couple of years we elect a new clown show, which then proceeds to destroy whatever the last clown show tried to accomplish. That has happened again and again for decades, truly awesome "our democracy".
Dumping is when you sell things for below cost. It is not dumping when you charge a 500% markup instead of a 1000% markup, even if the market is currently selling at that markup.
All the laws listed there define dumping as something being sold below the "normal price" and there being some quantifiable harm being done to local industry of the country being exported to.
So it has nothing to do necessarily with the cost of production, and based on this it could be considered price dumping.
That's merely the marginalists at the WTO struggling to fit their beliefs that value is subjective and unkowable into the reality of commodity production.
Okay, so what official sources do you base your definition of price dumping on? Nobody is asking for your opinion on what you think price dumping is. This is a legally recognised thing, not just some random talking point.
The normal price is not the same as the market price. The normal price is what we had last year. The market price is higher than normal, due to a sudden demand spike and production shifting to HBM.
More manufacturing capacity coming online to return the price to normal is not dumping, it's how markets are supposed to operate.
Dumping would be e.g. if China used subsidies to sell DRAM at a price below what unsubsidised manufacturers can sell at, in an effort to push them out of the market.
> It is wise for these Chinese fabs to eventually use a very aggressive dumping strategy to price well below cost push out other players forever, especially in DRAM.
Crucial's departure from the consumer market left such a gaping hole, that CXMT doesn't even need to push other players out to gain a footing.
No, they raise price because they can and demand isn’t showing signs of stopping despite increased prices. This won’t affect whether there’s a shortage or not, besides we’re not talking direct to consumer float product, they inked commitments.
If they didn’t have a documented history of running cartel price fixing schemes for LCD/OLED display tech, NAND, and DRAM, I’d maybe agree with you but we have the history. They cry every time about China ‘dumping’ for not going along with the racket.
Can chose to increase production or embrace the scarcity. The later might look delicious on paper. More profit, less effort, less short term risk. All you have to do is ignore the whale in the room.
They are increasing production. Fabs take multiple years to come online. The modern semiconductor industry moves only very slowly and at very great cost.
It's funny that you call this an "very aggressive dumping strategy" while AI vendors are doing the same but with even greater losses and on a much larger scale.
It's all simply a fight for market share.
The original sin is the existing DRAM vendors selling their entire (spare) capacity to the likes of OpenAI.
Can we please stop with this irritatingly persistent myth? AI companies, at least the big ones, do not sell inference at a loss - far from it. This has been debunked and explained many times and yet it keeps being repeated.
The numbers aren't public but most guesses I've heard are that Anthropic's markup is around 50% on average, and that if considered in isolation, most models are profitable overall. The constant losses are instead due to training the next models, which will also eventually recoup but later, and forward capex investment.
This idea that big AI companies are normally and systematically selling inference at a loss as some kind of market share strategy is just not supported by the facts.
Yes, because it can't be sourced. These are private companies. That doesn't mean that anyone's guess is as good as anyone else's.
The CEO of Anthropic, for example, has publicly stated multiple times that if their individual models were companies in and of themselves, they'd be profitable. I have no reason to think he's lying.
I find people tend to miss the productive aspects of Chinese state led investments because they don't consider their value at scale. Take the HSR system, it has been derided time and again as being wasteful, and too expensive, and so on. Yet, now it's become a key artery for trade and commerce across China. It allows goods to move at an incredible speed, boosts tourism, and helps overall development of many regions which otherwise wouldn't see much economic activity.
The downside in general is that other countries lose production capacity in steel, heavy industry, semiconductors, machine tools etc - industries that took decades to build and can't be easily replaced.
Also they gradually lose the ability to meaningfully innovate in those sectors because there's no grounding against production reality anymore.
This has geopolitical consequences further down the line.
As though moving production to China wasn’t something the West did intentionally.
And now continues to push manufacturing out of Western countries by, for example in the UK and Germany, and Australia too, making electricity and gas so expensive it becomes cost prohibitive to manufacture much at all.
You forgot to mention embargoes against it. The US is free to sanction firms for their exports to China, but then shouldn't be surprised when China builds out domestic competitors.
> The downside in general is that other countries lose production capacity in steel, heavy industry, semiconductors, machine tools etc - industries that took decades to build and can't be easily replaced.
That's not really what happens though. You don't actually "lose" capacity, you just move to higher-valued special niches within the overall industry because (1) you can afford to, while low-cost competitors can't and (2) you can no longer expect to be the lowest-cost supplier for the bulk of the market. That's a win-win development and something to be encouraged.
> You don't actually "lose" capacity, you just move to higher-valued special niches within the overall industry
That's not what people mean by "lose" capacity.
Suppose DRAM companies expand capacity because prices are high, then demand levels off, the price crashes, and they all go out of business except for the one in China which gets a government bailout. That's fine, right? We're not interested in making DRAM, that's a fungible commodity, we want to make iPhones or something. (They make those too anymore, but never mind that.)
What happens now if China restricts what you can buy to give an advantage to their own companies who are trying to displace you in the higher-valued special niches? Or just raises the price for you and not them? What if there's a trade war? Or a conventional war?
When you still have a domestic industry, you go to them and have a source for the commodity. If only one country becomes the sole global supplier and that country isn't even particularly friendly, that's bad.
The way I see it, China has leverage once you arreive to that dependency situation. That leverages goes away the moment they restrict exports and every country scrambles to create local production once again.
We are seeing that with some rare earths, even tho china is back into exporting them (except to japan, I think?) everyone is looking for alternatives already. They may have killed their industry 10 years down the line for playing with the export lever a bit too much.
Just like how markets punish the ram cartel creating a chance for cxmt and ymtc to enter. It would create a chance for western companies to do the same if china messes with the markets they have "cornered".
The domestic industry is still there, only instead of mass-market DRAM it has started making higher-valued varieties of the same stuff. If there's a trade war, they can easily reconvert to making the mass-market stuff, just at much higher cost. You can't expect more than that, since they never really were as big or as low-cost as the lowest cost suppliers can be in normal times. That's not "losing" capacity, it's just acknowledging that you can't create capacity out of thin air.
> steel, heavy industry, semiconductors, machine tools
the question is if single country can carry all these industries at loss for prolonged period of time.
Another approach is to rely on international supply chain and speed of innovation, we can't produce steel domestically profitably today, fine, we may buy it from diversified international supplier network, and rebuild it fast tomorrow if needed using new tech, and focus on many other high margin verticals, instead of putting many billions of resources into infra which could be obsolete tomorrow.
First, they're not selling at a loss; the huge price increases have allowed them to push aggrssively in the legacy markets. They're making "slightly smaller" profits than other manufacturers (of which there are now very few).
Second, they can drive out all competition and then have a captive audience for whatever prices they want, as the barriers to entry in these markets are very high. This is essentially what's happened with all higher-end manufacturing in the west over the past 30+ years.
I don't know if it's still a thing, but China was getting a lot of heat about a decade ago for purposefully devaluing their currency to make their exports more attractive.
They kind of had to do this, because their large amount of exports were pushing the value of it up compared to others.
They are? Most of what VC funded companies do has been done before at a smaller scale, often with less polish and at a higher price.
VC money is used to scale up, cut costs with scale, capture markets, and then usually prices go up later depending on the economics.
The Chinese state is basically just acting as a big VC fund for Chinese manufacturing industries. A VC fund with a sovereign currency and the ability to sustain burn-mode for decades.
It doesn’t always work. There are some absurd examples of Chinese waste produced this way like “ghost cities.” But when it works it works, and at tremendous scale, and they can just dominate entire industries.
It's questionable whether the ghost cities truly exist though. I was under the impression they were a product of China's bizarre savings and investment market, and that a lot of them have since filled up?
Amazon was wildly profitable on a unit cost basis from relatively early on. They didn’t show profits on paper because they reinvested everything into their capital buildout to reduce costs even more.
> It is wise for these Chinese fabs to eventually use a very aggressive dumping strategy to price well below cost push out other players forever, especially in DRAM.
It's not dumping, it's the opposite.
Sam Altman's stunt has created massive amounts of fictitious demand (OpenAI isn't using those wafers it's ordering) and triggered massive panic-buying from everyone else.
Prices are arteficially high, this has turbocharged China's fab and R&D budgets as you observe.
> is about to pay off with some industry dominance soon.
They're not looking to dump the semiconductor markets. They're looking to invade Taiwan.
All this buildout in their semiconductor industry is to detach themselves from the western semiconductor industry that will either sanction them if they invade Taiwan, or in the case of TSMC, suffer major damage in the ensuing conflict.
That the collapse/destruction of the Taiwanese semiconductor and electronics industries will utterly ruin the western tech industry is somewhere between a happy coincidence and acceptable collateral damage to them. No dumping required.
China economic numbers don't have a winning tune to the in the least. I don't think taiwan could ever be taken by force, but that doesn't matter as Xi Jinping seems to think it's doable and is taking steps towards it (developing landing ships, purging the military of oppsition and pacifists, building a fleet and bombers...)
It would be very surprising to me that taiwan people think a reunification is feasable while the CCP still exists, just see how things are going in HK to see what would be waiting taiwan if they reunite.
> Why would they bother to invade Taiwan when they’re winning economically and diplomatically?
They are hedging their bets. If Taiwan refuses to accept re-unification, China wants to have the option of a military annexation.
They have been planning this for quite a bit longer than the current US administration. They're not going to bank on Donald Trump forever, he's not getting any younger and healthier, and November 2028 is sooner with every passing day. A military conflict is not off the table, and so it is considered and prepared for.
I realize Intel has done some serious ball dropping over the past two decades but you do realize the US has on shore cutting edge fabs, right? It's only luxury consumer electronics and the highest end corporate gear that use cutting edge nodes to begin with.
Disruption of the cutting edge would certainly wreak havoc on the pricing and specs of high end luxury electronics but that would hardly be the end of the world. I still use a desktop with DDR3 on a daily basis (granted the GPU is much newer with GDDR6) and my laptop is from the early era of DDR4 ...
The Intel CMOS process 18A, which they have launched a few weeks ago, is the first after almost a decade that is somewhat competitive with TSMC and Samsung.
Good for Intel: their new manufacturing process has demonstrated a much better energy efficiency than the TSMC "3 nm" process that was used to make Intel Arrow Lake and Intel Lunar Lake.
Unknown: TSMC now has a "2 nm" process and the first products using variants of this process are being launched. It is unknown how TSMC "2 nm" compares with Intel 18A, but it is almost certain that the TSMC "2 nm" is better.
Bad for Intel: they had difficulties to achieve high clock frequencies in Intel 18A in comparison with TSMC "3 nm", so most Panther Lake models have lower clock frequencies than their Arrow Lake counterparts. Moreover, it is also pretty certain that for now Intel 18A has much lower fabrication yields than even the latest TSMC "2 nm" process.
> I realize Intel has done some serious ball dropping over the past two decades but you do realize the US has on shore cutting edge fabs, right?
We could squabble about the finer details of Intel's fab capabilities. They have advanced nodes, but it's irrelevant. They simply do not have the capacity to support the entire demand that is currently supplied by TSMC.
It is not just "high end luxury electronics" that have modern CPUs. It's every bloody server in the cloud. (Have a look at who makes and distributes the mainboards. Same story, substitute Intel for Supermicro.)
The economic impact on this field would be a disaster. Compute becomes much more expensive, SaaS prices will follow, and with that a massive drop in demand.
Not to mention you can kiss the entire AI industry goodbye if the price of GPUs spike.
I don't think that's an accurate prediction. Currently less than 10 year old hardware gets recycled for pennies on the dollar. That's effectively due to the combination of how cheap and how much better the cutting edge hardware is. If it suddenly became more expensive it would just see slower adoption.
Case in point, this very comment section. The major suppliers have discontinued DDR4 production because it's "obsolete" meanwhile capacity for that exact same technology is coming online in China. What makes sense just depends on context.
>> as if Samsung didn't receive government subsidies when it was developing
That's because Samsung didn't get gov't subsidies when they developed DRAM in the early 80's. Semiconductors weren't really among the targeted industries by the South Korean gov't and were largely ignored as the govt prioritized HCI (Heavy Chemical Industry), shipbuilding, steel, and automobiles. The gov't didn't understand the potential of semiconductors and viewed the growing industry as risky investment.
Unlike China or Taiwan, the South Korea's semiconductor industry was very much organic, started by a private company, Korea Semiconductor, in the mid 70's which was later acquired by Samsung and became Samsung Semiconductors Inc. And they bled millions of their own money until their own DRAM came out in the early 80's.
The gov't still believed the industry was too risky and costly even after Samsung's development of 64K and 256K DRAM in 1983 and 1984. Samsung burnt through their own cash stockpile cross-financed by other divisions, or borrowed from foreign banks and financial institutions. There was really no major support until 1986 when ETRI, a gov't sponsored research institution, stepped in to promote cooperation among domestic semiconductor players -- ie, so Samsung could teach and help bring up other chaebols LG and Hyundai up to speed.
So PLEASE no more insane whataboutism to defend China's neo-mercantile practices or illegal state subsidies.
I've heard that Samsung's business practices can be quite predatory. Basically if you have cool tech and you try to sell it to Samsung, you'll often get a few meetings and then they will go silent and then what you were trying to sell them will be offered by them as a new product about a year later. At least this was the situation like a decade ago.
I think this is because they are a huge conglomerate and there are divisions and groups that specialize in everything and their (Samsung) culture is to do everything as much as possible in house.
> you'll often get a few meetings and then they will go silent and then what you were trying to sell them will be offered by them as a new product about a year later
This is how business works anywhere. there are no charities. whatever you say to investors or suppliers they can use so you better be careful have lawyers and set up correctly.
(The caveat is of course when Chinese companies do this your lawyers can do nothing while in a developed country you can have some recourse)
But even if Samsung was super predatory business wise it is beside the point. if they both get subsidies and de facto are close to their governments then you have to look at what their governments do. If you like what CCP is doing, it's your choice
So try to sell your high tech product to literal charity next and see what they offer. As you can imagine after this thought experiment, we're not talking about literal charities.
If you want to make money don't be surprised they also want to make money. If it is financially better to build it themselves they will, or they will tell their friends. Pro tip: don't tell them enough.
Just like in China if they can take your product and slap their own label and sell it at 50% discount they will.
How many countries has China bombed in, say, the last year, and how does that compare to the other superpower (and the washed out has-been superpower) in the room?
It's difficult to comprehend the gall and hypocrisy requires to kvetch about this when there are four carrier strike groups sitting in the Gulf right now.
That explains the cheap DDR4 DIMMs on AliExpress. Can get 2x 16GB DDR4-3200 DIMMs for A$252 delivered to Australia. A local PC store has same spec “name brand” RAM for around $380-$400.
Oh don’t worry I know. If I can pat myself on the back a bit, I was super vindicated at work last week when someone shared a WD drive shortage article - I had pushed us to grab 7x 14TB refurbished enterprise drives in December (6 bay raid, 7th is a spare). The 12TB seagate EXOS currently cost $110 more than we paid for those 14’s.
DDR4 going from $1.35 to $11.50 in a year shows this market was already distorted before CXMT showed up.
Legacy DRAM is still over half of Samsung and SK hynix's production capacity. That's where the volume pain actually lands while they're betting everything on HBM4.
A near 10 times price rise of legacy chips and surely a lot higher profit increase couldn't motivate SKamsunix to increase their capacity for that segment, even more, they proudly informed the market about their decision to stop making DDR4 at the end of this year... to focus on even higher margin products.
But they aren't going to stop whining about China, no matter how much pain the market experiences.
> CXMT is in the process of converting wafer capacity equivalent to about 20 percent of its total DRAM output — some 60,000 wafers per month — at its Shanghai plant to the fourth-generation HBM3 chip production
Apple has planned to explore cooperation with Chinese memory chip manufacturers Yangtze Storage (YMTC) and Changxin Storage (CXMT) to strive for more favorable supply contracts [from the big three]
This feels like a classic business blunder. Focus hard on a single business segment, leaving an opening in the market for your competitors. Not because it wasn't profitable, but because it wasn't profitable enough for you, right now. Only downside is that now you've created an opening for a new player in the market.
This feels like a short coming of western business/stock market thinking. Focusing on profit within the next few quarters, and not caring about the longer term consequences. For all it's flaws and shady business practises at least China can think beyond a single fiscal year.
Ok but this is how the market is supposed to work. If the incumbents aren't doing what their customers want, then competitors can rise and fill the gap and compete.
This isn't a shortcoming, it's a competitive market working as intended.
Who said I thought businesses would always make the right move?
Businesses blunder. "The market" is just a set of observations, including that competitors will tend to take advantage of those blunders. It is not a failure of the market that businesses have blundered, nor surprising that it will happen occasionally, and neither I nor mrweasel implied otherwise.
NO you see, we have to hate Chinese companies because they are unfair competitors since they get state funding from the Chinese government, unlike Intel, Micron, TSMC, ASML, Samsung who don't get state funding from the US, EU, Taiwan, ROK ... oh wait.
Scratch that, we have to hate Chinese companies because they do business with the Chinese military, unlike Intel, Nvidia, Samsung who don't do business with the US and ROK military ... oh wait.
I know you are being sarcastic but the reason why we have to hate Chinese is simply because the standard of living of Americans depends on China not succeeding, simple as that.
Does it, though? If anything it seems like the opposite: China's success had directly enabled my standard of living as an American to be as high as it is.
I suppose it'd be true that the standard of living of some Americans depends on China not succeeding — specifically, those Americans who own corporations competing with Chinese firms — but I think they'll survive just fine with only 10 yachts instead of 15.
Well then hey can just say THAT, instead of coming up with hypocritical BS that doesn't pass the smell test. People internationally have enough IQ to see through the double standards BS, especially since youtube is a thing.
And the standard of living of working class Americans has been on a steady decline since Reagan by the hand of US administrations, not by the hand of CHina.
This isn't defending anyone's standard of living, it's defending profits of domestic monopolies like Micron, who indulge in state subsidies from US taxpayers and then fuck then over on prices.
Lets not forget there is no competitors. There is one competitor- the chinese state, one huge company willing to subsidize any endeavor that will help it fmgain more marketshare with already captured markets.
CXMT sells the vast majority of their bits at the prevailing market rate, just like everyone else. They are adding capacity as quickly as they can, with a 5-10 year planning horizon, just like everyone else. It’s really not that deep!
Yes of course their messaging to customers and the investment community is that they will be rational and measured in their investments. In reality, they are adding capacity as quickly as possible as margins are too high. However, capacity addition leading edge semiconductor manufacturing has a multi-year lead time.
> In reality, they are adding capacity as quickly as possible [...]
Even if we ignore the fact that you can't build out factories in secret, this would be securities fraud by publicly listed companies.
> ...as margins are too high
This is not the first RAM boom-bust cycle, and memory makers are an actual cartel convicted of coordinating in the past - none if them are going to break rank when they can invest the minimum and reap outsized benefits. Also, no one wants to invest in additional capacity when the bottom can fall out at anytime, and shareholders demand your head- not even the AI companies want to bear that risk, which is saying something.
PRC asked them to curtail DDR4 production so they didn't bottom out the market a year or two ago, and to focus on latest gen development, like HBM. They were the world leader in cost efficient DDR4 production at the time.
It's not really a blunder though. Given that total capacity is tightly constrained, Samsung and SK Hynix are happy to focus on what they do at their best and with the highest margins. Why shouldn't they supply the HBM market?
Maybe not to starve whole segments of the society and economy of critical inputs ?
This is why we have strategy reserves - to avoid greedy companies to make a quick buck when they find out they can sell the equivalent of crack and dump all the other now low margin stuff like food or other essentials.
Sure, that company might make a killing for a short while, but people will die from hunger, missing medicine or freeze to death.
So let's buy a strategic reserve of DIMM sticks from CXMT. But that's not a great strategy either, because RAM tech evolves. DDR3 to DDR4 to DDR5 to DDR6. So you need to rotate your strategic reserves over time, and luckily enough for you, the used goods market is doing exactly that for free!
It's putting all eggs in one basket. If/when the higher-margin category collapses, they'll have no fallback. Imagine Chevrolet had discontinued Impalas and other low-margin cars, and switched to Corvettes during the pandemic
Even if there's a collapse in the high-margin market (I don't think anyone is expecting this right now), it will be slow and telegraphed in advance, giving them plenty of time to refocus on lower-margin products.
> Focusing on profit within the next few quarters, and not caring about the longer term consequences
Anything new? From my non-American view, American companies has done similar things for a very long time now. It happened in the consumer electronics, it might happen again in the IT industry.
It's not the fault of the companies, they simply just wanted more certainty and the consumer market is not (when compare to cooperate contracts).
But from the stand point of a nation, if no one creates low-end products, then no one will be providing low-end/entry-level jobs. That's when you got structural problems.
There is really nothing about the stock market that means only thinking about the mext few quarters. See all the losses on the profit and loss statements of AI tech giants, or, say, game console companies? Why are their stocks still valued so highly during these periods? The answer: investors are thinking long term.
It is really impossible to have quality long term thinking without capitalization accounting and similar instruments that come out of the "wester" system of business that chinese free enterprise gladly and speedily copied when it was made free.
I think the sentiment here is about management's tie of bonuses to near-term stock performance. Maybe not about the market itself, I agree with your view on investors want long term gains over short term fluctuations mostly.
Everyone here wants to be able to buy RAM at a reasonable price again.
After reading articles about CXMT and repeatedly reviewing the comments here - my take is there's nothing in play that will lead to reasonably priced RAM anytime soon.
If I'm wrong please illuminate us. We could use some hope.
Can Apple, Dell and HP lobbyists convince US regulators to lift restrictions on working with Chinese memory manufacturers, at least until Micron's new semiconductor fabs can ship US-made DRAM and SSDs to US OEMs in 2028? https://wccftech.com/cxmt-ymtc-removed-from-pentagon-list-op...
The Pentagon has withdrawn the document that suggested updates to Section 1260H
Each fab will be 600,000 square feet—the size of more than 10 football fields—making them some of the biggest “clean rooms” ever built in America. To prepare the site, engineers have already blasted through more than 7 million pounds of dynamite. An army of construction workers, building contractors and architects have set up a small city’s worth of trailers so they can work around the clock.
I was wondering when people would find out about CXMT. I wish them luck and hope the US doesn't sabotage them. We need diversity and competition right now.
Because that’s just the price for the dies, still need packaging, integration and then retail distribution. Also raw BOM is like 1/5th of the retail price usually.
You can buy oil on the market and take delivery, yes. Figuring out the logistics (tanker trucks, rail cars, storage tanks etc) will also be your problem though.
Commodities markets aren’t like eBay, nobody is just going to FedEx you a barrel of oil.
But since when? There are public announcements about new energy deals since summer 2024. But I'm missing any information about similar RAM/NAND/HDD deals back then, so that corresponding shortages could be only for short time until, say, summer 2026.
It ain't the price raise that's the mistake (even if that's what's currently painful for those of us looking to buy RAM). It's the willingness to only raise prices, and not meaningfully expand production, that's the mistake.
Back in the 1990’s everyone had to have a unix workstation for unclear reasons (why not run Linux for < 10% the cost?).
There were crazy bubble economics schemes that meant doomed startups got unix boxes for free.
When the bubble popped, the workstation vendors hit a triple whammy: Inferior $/perf, unlimited used inventory at low prices, and an economic downturn.
The same exact thing is happening now, except the hardware is being jammed into data center models.
Anyway, when the bubble pops, people making affordable consumer stuff will be fine (like this CXMT company).
People that went all-in on firing all non-hyperscaler customers (like micron/crucial) will find they’re building the wrong chips for end-user devices, there is no server market anymore (for a few years), and they have a total addressable market of maybe 1000 distressed companies, globally.
I predict the people making these decisions and destroying their companies to juice Q2 2026 financial outlook numbers will genuinely be surprised when the bankruptcies start.
great comment. actually reminds me of how intel screwed up with developing the first 64 bit processor, then amd came along with theirs and managed to force intel into compatibility.
All these big companies think they're too big to fail until they do.
at the end of the day the consumers hold the purchasing power. ignore them at your peril.
Has DDR5 caught up to DDR4 latency yet? I remember it was worse at least in the beginning. There's more bandwidth per channel but a hw design can always add more channels for the desired BW. Not so for latency.
and unfortunately increase latency even more with registered DIMMs. Comparing bandwidth increase (50 GB/s) to the stagnated latency (~80..120 ns total, less than ~0.1 GB/s) over last decades, I'm wondering, whether one still can call today's RAM random memory (though sure it can be accessed randomly). Similar to hard disk drives. Up to 300 MB/s sequentially but only up to less than 1 MB/s 4KB random (read).
It might be very effective marketing. The big non-Chinese OEMs trust and use Korean and Japanese DRAM, and they might have been unwilling to put DRAM from CXMT into their products. (CXMT is newish, does not have access to ASML gear, which ASML would like you to believe makes it harder to make high-quality DRAM, DRAM is historically not a very large fraction of the cost of most non-huge-memory machines, and a bad DIMM is an expensive mistake for a company like Dell or HPE that is on the hook for repairs.)
But now CXMT seems to have gotten at least Dell, HP (I wonder if the article meant HPE), Acer and Asus to buy and attempt to qualify samples. If CXMT lands some serious purchasing agreements while still selling well above cost, that’s a win for them.
Because market rate is a 400%+ markup right now and not everyone is a greedy American kleptocrat with a diagnosable addiction to extracting every possible cent of wealth on the planet within a single fiscal quarter.
I can see a few other benefits to the "Sell at a reasonable markup, rather than 'market price'" strategy.
* You get a reputation for not price-gouging.
* Your company gets used to operating with "enough" cash, rather than having "crazy" amounts of cash that it might be tempted to spend on misadventures.
* When the collapse of LLM Mania burns the LLM industry and its hangers-on to the ground, you get to keep your prices the same and keep your business practices the same... rather than scrambling to figure out how to make do with far less cash.
Does the last part of your comment explain it? They need revenue to expand capacity and the market has opened up a window to become a bigger supplier while still being profitable.
The pertinent question, then, is who's gonna eat China's lunch?
My guess would be south/southeast Asia (India and Vietnam seem especially promising), but if the US was smart it'd put its efforts fully toward as many infrastructure investments and trade agreements and immigration agreements as possible to create a pan-American economic union. We have the resources and technology to turn every country in North and South America into an industrial and technological powerhouse. We have the resources and technology to finally conquer the Darién Gap and connect North and South America with highways and high-speed rail. We have the resources and technology to go on the offense against drug cartels (while also eliminating the failed border controls and drug prohibitions that keep those cartels in business in the first place).
If we're gonna be imperialists, then by golly let's at least be productive about it.
China's big. It has a long way to go before all of it becomes developed enough that they have to worry about someone else eating their lunch.
And Latin American nations can't get started on economic development because their governance sucks. It's actually a decades-old problem, and not one that the U.S. can do much about. The one country that probably has the best shot right now is Argentina, let's see how they do.
Is there a reason GPU's don't use insane "blocks" of sdcard slots (for massively parallel io) so the model weights don't need to pass through a limited PCI bus?
Yes. Let's do the math. The fastest sd cards can read at around 300 MB/s (https://havecamerawilltravel.com/fastest-sd-cards/). Modern GPUs use 16 lanes of PCIe gen 5, which is 16x32Gb/s = 512Gb/s = 64 GB/s. Meaning you'd need over 200 of the fastest SD cards. So what you're asking is: is there a reason GPUs don't use 200 SD cards? And I can't think of any way that would work
SD is obviously the wrong interface for this but "High Bandwidth Flash" (stacked flash akin to HBM) is in development for exactly this kind of problem. AMD actually made a GPU with onboard flash maybe a decade ago but I think it was a bit early. Today I would love to have a pool of 50GB/s storage attached to the GPU.
Oh definitely. The AMD past product just stuck 4x m.2 slots onto the board. Today that approach would be 50-60 GB/s read speed which would be useful enough for something that any of the vendors could build with existing components.
One thing to note, those aren't the fastest SD cards, those are the fastest UHS-II SD cards. The future is SD Express and you can already get microSDs at 900 MB/s.
SD expires cards are still on my watch list after seeing this https://www.theverge.com/2021/9/9/22665216/sd-express-card-s.... But even then, that's only 3x faster, so you're still putting down 71 ish (I think my original number was 213.33) cards, which means 71 PCIe PHYs and NVMe stacks.
Some years ago I realized that if I had oodles of money to spend I would totally get someone to make a PCIe card with like several hundreds microSD cards on it.
You can buy vertical microSD connectors, so you can stack quite a lot of them on a PCIe card. Then a beefy FPGA to present it as a NVMe device to the host.
Goal total capacity, as you can put 1TB cards in there. And for teh lulz of course.
The next gen inference chips will use High Bandwidth Flash (HBF) to store model weights.
These are made similarly to HBM but are lower power and much higher capacity. They can also be used for caching to reduce costs when processing long chat sessions.
The random access memory models is not really representative of ML workloads (both training and inference), where multiplying large tensors result in predictable memory access patterns.
As far as I understand, the "entity list" you are referring to is part of the "Export Administration Regulations", so it restricts sales from the US to restricted entities, not the other way around.
Consumers generally safe to import for personal use, but for larger orgs, there's likely strict supply chain compliance risks that makes companies on 1260H list, i.e. military concern list no go.
OTOH, now I read small Taiwanese manufacturers who are left out of the Nvidia supply chain are reverting to DDR4 motherboards because of the DDR5 shortage. Strange times.
Geopolitics and industrial policty aside, I think it's important to check how stable and reliable these chips are. I wouldn't count on them being on par with "western" ones. Correct me if I am wrong here.
Even Apple is looking into them as providers. And imagine what the money Apple can provide in advance can do for their manufacturing needs if they are serious about securing memory supply for their product lines.
It is wise for these Chinese fabs to eventually use a very aggressive dumping strategy to price well below cost push out other players forever, especially in DRAM.
But right now it seems they can max out their supply capacity without selling below cost.
Appears to me like China's endless state led (often unproductive) investment in semiconductor manufacturing subsidies (for decades) is about to pay off with some industry dominance soon.
Like the electric vehicle sector.