Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
I don't need a lecture from you. This isn't "their" forum, but a forum for all members.
Apple products in general, and its top-tier items in particular, are pretty much luxury items if only because of the margins Apple makes on them. I certainly won't shed a tear because Jimmy's annual Ultra Pro impulse, which he buys on credit, will cost him $2,000 instead of $1,700 the next time around, and he might decide to pass as a result.
Just want to point out that more than half of all people old enough to have a phone in the US have an iPhone.
 
I don't need a lecture from you. This isn't "their" forum, but a forum for all members.
Apple products in general, and its top-tier items in particular, are pretty much luxury items if only because of the margins Apple makes on them. I certainly won't shed a tear because Jimmy's annual Ultra Pro impulse, which he buys on credit, will cost him $2,000 instead of $1,700 the next time around, and he might decide to pass as a result.
I can’t say I’ve ever heard that definition for luxury items before, but that would put prescription medicines in the upper class of luxury items. I guess that tracks with the direction of the world these days, though.
 
  • Like
Reactions: vantelimus
The cartel's pricing is creating one hell of a price umbrella. It'll be very attractive for competition to enter the market. I doubt the entire computer industry is going to just sit there and not do anything.
It's not possible to have another entrant into the market. The patent base/walls are high. Apart from CXMT, there's nobody with a patent base that permits them to enter the market without violating intellectual property. So, the total supplier base is the big4 now: Samsung, SK Hynix, Micron and CXMT. Imagine if a new startup tried to develop DRAM technology from the ground up without stepping on the toes of the big4. It would be a decade-long exercise that might not prove fruitful and would require very deep pockets willing to sustain a decade of heavy losses. No chance Apple will try that. Also, note how long it took Apple to come up with its own modem that got around Qualcomm's patents.
 
It's not possible to have another entrant into the market. The patent base/walls are high. Apart from CXMT, there's nobody with a patent base that permits them to enter the market without violating intellectual property. So, the total supplier base is the big4 now: Samsung, SK Hynix, Micron and CXMT. Imagine if a new startup tried to develop DRAM technology from the ground up without stepping on the toes of the big4. It would be a decade-long exercise that might not prove fruitful and would require very deep pockets willing to sustain a decade of heavy losses. No chance Apple will try that. Also, note how long it took Apple to come up with its own modem that got around Qualcomm's patents.
The difference with the modem was that this wasn't an existential threat aka "must have" like memory, it was at best a "should have", or even a "could have". Not having this wasn't threatening the entire industry or Apple's position in the market.

The car also comes to mind. It ended up as "won't have" 🙂.

I'm not the CEO of a major corporation, but I'm pretty sure Apple and other companies with significant hardware businesses are going to attack the current situation from all angles, since it's their job. Some of these ways can't be discussed here due to the rules. I doubt patents will be the ultimate obstacle. FRAND exists. I'm not a lawyer.
 
Also, note how long it took Apple to come up with its own modem that got around Qualcomm's patents.
Good point with a nuance... Apple's modem is fabricated by some other company like TSMC.

Also, there is the tier with many oligopolies for many different things memory fabs buy such as equipment (ASML in one category, Lam in another, etc), and materials like wafers and gases, photomasks, and so-on.

And each category of that first tier has additional oligopoly suppliers, particularly ASML. (There is a youtube channel "Asianometry" that has produced episodes that focus ASML components including perfect mirrors, tin‑droplet generator that fires microscopic droplets, and the laser system that vaporizes those droplets to produce the EUV light required for chip manufacturing.)

Good chance the suppliers of the materials for those mirrors, tin, and lasers each add a third tier of very limited material suppliers. Like Zeiss does not just head to the beach to get what it needs to make perfect mirrors.

The USA plays no part in many of these areas.
 
Last edited:
It’s my intention never to buy another Micron or Crucial product ever again. Last year they closed all consumer-facing operations to focus on data centers. Now, they’re raking in record profits and then blaming others for the prices. Keep in mind, Micron has made a profit in the billions in most years since 2010, so the idea that Apple is at fault for Micron not investing their massive war chest in expansion rings almost criminally hollow.
 
It’s my intention never to buy another Micron or Crucial product ever again. Last year they closed all consumer-facing operations to focus on data centers. Now, they’re raking in record profits and then blaming others for the prices. Keep in mind, Micron has made a profit in the billions in most years since 2010, so the idea that Apple is at fault for Micron not investing their massive war chest in expansion rings almost criminally hollow.
I mean, they killed the crucial brand and all consumer production completely so that wont be too hard. Unless you’re an AI company buying HBM you pretty much cant buy *anything* from Micron right now
 
I mean, they killed the crucial brand and all consumer production completely so that wont be too hard. Unless you’re an AI company buying HBM you pretty much cant buy *anything* from Micron right now
And who's to blame? Apple of course!!!!!!!!!!!!!!!!!!

In unrelated news, a few weeks ago I bought soda on sale at Costco, and now they have closed all branches and only sell it for $800 per bottle, and only to businesses. Shame on me!!!!!!!!!!!!!!!!!!

😉
 
It's not possible to have another entrant into the market. The patent base/walls are high. Apart from CXMT, there's nobody with a patent base that permits them to enter the market without violating intellectual property. So, the total supplier base is the big4 now: Samsung, SK Hynix, Micron and CXMT. Imagine if a new startup tried to develop DRAM technology from the ground up without stepping on the toes of the big4. It would be a decade-long exercise that might not prove fruitful and would require very deep pockets willing to sustain a decade of heavy losses. No chance Apple will try that. Also, note how long it took Apple to come up with its own modem that got around Qualcomm's patents.
And CXMT is only somewhat viable because they steal from the other 3. A non-Chinese company would not be able to compete as they wouldn’t have the advantage of corporate espionage.
 
And CXMT is only somewhat viable because they steal from the other 3. A non-Chinese company would not be able to compete as they wouldn’t have the advantage of corporate espionage.

CXMT is also facing production constraints as they serve China’s massive domestic AI market, along with dealing with yield issues in high-bandwidth memory. While they could help ease demand for cheaper DDR4 and basic DDR5 memory, they’re already running at full capacity (300,000 wafers per month) and won’t have additional capacity until their new factories open in late 2028.
 
And CXMT is only somewhat viable because they steal from the other 3. A non-Chinese company would not be able to compete as they wouldn’t have the advantage of corporate espionage.
FYI, CXMT bought their way in.
"In 2019, CXMT signed an agreement to access a massive cache of Qimonda's patents (roughly 7,000 patents and applications) and 2.8 terabytes of technical documentation. This gave CXMT a legitimate, non-infringing foundation outside the dominant Samsung-SK Hynix-Micron triangle."
Later on, process optimizations from Samsung ended up at CXMT. CXMT recruited people who were unhappy and no longer wanted to work at Samsung or had retired. According to Korean prosecutors, this saved CXMT a couple of years of trial-and-error.
 
Last edited:
CXMT is also facing production constraints as they serve China’s massive domestic AI market, along with dealing with yield issues in high-bandwidth memory.
CXMT doesn't sell HBM. Only prototype HBM3. Maybe in two years' time. But we have moved to HBM3E and HMB4 already. And HBM5 and HMB6 are coming. This is a much harder race for CXMT than regular DDR4 and DDR5.
 
CXMT doesn't sell HBM. Only prototype HBM3. Maybe in two years' time. But we have moved to HBM3E and HMB4 already. And HBM5 and HMB6 are coming. This is a much harder race for CXMT than regular DDR4 and DDR5.

They’re producing HBM3 and have started early mass production, but the yields are terrible, cutting into their wafer capacity. The point is, they’re essentially a non-starter for easing the RAM crunch in the US (sanctions notwithstanding), even in the consumer market, since they’re already at full capacity and the situation is only getting worse.
 
They’re producing HBM3 and have started early mass production,
Not really. "CXMT’s filings indicate that roughly 99% of revenue consists of DDR and LPDDR products in 2025."
Reports indicate they are considering punting, they can't start HBM3 production until they can master how to do stacking at 12-hi. And since they've failed, the report says they'll skip making HBM3 altogether.
 
Not really. "CXMT’s filings indicate that roughly 99% of revenue consists of DDR and LPDDR products in 2025."
Reports indicate they are considering punting, they can't start HBM3 production until they can master how to do stacking at 12-hi. And since they've failed, the report says they'll skip making HBM3 altogether.
The report about them failing 12-hi stacking is true, but them "skipping" HBM3 altogether makes no sense from an engineering standpoint. You can't just skip the TSV learning curve. They are falling back to 8-hi HBM3 stacks to try and stabilize yields. Plus, what exactly are they skipping to? They physically can't make HBM4 because it requires an advanced logic base die, and US sanctions on EUV machines prevent them from fabricating it. They are stuck trying to perfect HBM3 whether they like it or not and pray they can perfect 12-hi stacks while HBM3E still has a market. Regardless of what their commercial success or revenue is, they are burning, as of reports earlier this year, 60K wafers (20% of capacity) on it monthly.
 
The report about them failing 12-hi stacking is true, but them "skipping" HBM3 altogether makes no sense from an engineering standpoint.
HBM3E 8-hi/12-hi instead. But CXMT doing HBM is a waste of capacity/profit according to some reports. They can use that capacity for regular DRAM, which they're doing fine on and have no problem selling everything they make.

Since the Hebei provincial government has deep pockets and willingly took 10 years of losses for CXMT to make it in the DRAM market, it's likely that HBM will be funded no matter what. One way to look at it is that CXMT has to succeed in HBM eventually for strategic reasons. But by the time they succeed, they have to make the on-ramp at maybe HBM4 or HBM5.
 
Bring it on.

Just for fun, looked into it a bit and seems frivolous. The plaintiffs are trivial in terms of damages. The lead attorney
Brian J. Dunne firm bio lists 8 cases as his relevant experience. Of the 8, one is still pending, one is a win and 6 were dismissed.

Also a quote from Tom's Hardware: "However, this case revisits a precedent that was already established in the same courthouse. A 2018 class action brought by law firm Hagens Berman made similar claims about parallel production cuts, and the district court dismissed it in 2020. The Ninth Circuit upheld that dismissal in 2022, ruling the trio's conduct was “more likely explained by lawful, unchoreographed free-market behavior” than by an illegal agreement."

Interestingly, a company named "My Florida PC" - a mom & pop PC repair shop in Plantation Florida - is listed as a plantiff in both that 2018 and this new class action.
 
... Of the 8, one is still pending, one is a win and 6 were dismissed. ...
To most laymen, this would probably feel like justification enough for ignoring Dunne and his ilk entirely. The problem is, the legal system is one of the few employment opportunities where a losing streak punctuated by a single win can keep a practitioner not only gainfully employed, but embolden them to go forward and continue doing more of the same.

Ergo, frivolous or not, that guy's not going away anytime soon.
 
  • Like
Reactions: Fallforward
To most laymen, this would probably feel like justification enough for ignoring Dunne and his ilk entirely. The problem is, the legal system is one of the few employment opportunities where a losing streak punctuated by a single win can keep a practitioner not only gainfully employed, but embolden them to go forward and continue doing more of the same.

Ergo, frivolous or not, that guy's not going away anytime soon.
Ya, even when they win often everybody else loses, like when a large amount of money is spread over a large number of class members. The attorneys get millions while each class member gets pennies.
 
  • Like
Reactions: zarmanto
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.