Analysts Say We Are Headed For a Flash Memory Price Crash (techspot.com) 99
With the industry currently facing a very large surplus of NAND flash memory, analysts suggest we could see very significant price drops in SSD and even DRAM in 2019. They say to expect a price correction over the next several quarters. Techspot reports: Jim Handy, a market analyst with Objective Analysis, predicts that the flash memory industry is headed for a "downward pricing correction" in 2019, if not a full-on collapse. If prices crash, we could be looking at NAND prices as low as eight cents per gigabyte. At last week's Flash Memory Summit, Handy said that even without a full collapse, the downturn will be the biggest "price correction in the history of semiconductor products."
The Register reports that currently, NAND flash prices are hovering around $0.30/GB. A 66-percent dip would bring SSDs into a more competitive range to HDDs causing cannibalization leading to a downturn for some manufacturers like Seagate and Western Digital. Manufacturers could allocate more NAND to producing DRAM, but this, in turn, would result in an oversupply in that sector. If Handy's predictions pan out, the industry could be in for a 25-percent price reduction in NAND and a 75-percent drop for nearline/high-cap SSD's. This could result in significant stock valuation shifts for some manufacturers.
The Register reports that currently, NAND flash prices are hovering around $0.30/GB. A 66-percent dip would bring SSDs into a more competitive range to HDDs causing cannibalization leading to a downturn for some manufacturers like Seagate and Western Digital. Manufacturers could allocate more NAND to producing DRAM, but this, in turn, would result in an oversupply in that sector. If Handy's predictions pan out, the industry could be in for a 25-percent price reduction in NAND and a 75-percent drop for nearline/high-cap SSD's. This could result in significant stock valuation shifts for some manufacturers.
Wait a minute. (Score:5, Insightful)
Re: (Score:3, Interesting)
It wasn't collusion. Moving production from planar to 3D NAND caused supplies to be limited as bit-demand was increasing. The transition to 3D NAND is set to be largely complete, but bit-demand isn't accelerating at current price points.
Supply will increase, prices will drop, bit-demand will go way up. Prices will stabilize. NAND storage for all at reasonable prices.
Margins at producers will fall, but not far enough for long enough for them to intentionally idle production.
Re: (Score:2)
Because of high bit-demand from data centers and smartphones.
Re: (Score:3)
Because of high bit-demand from data centers and smartphones.
Fortunately, byte-demand is not as high, but they cost more so it all works out. Gigabyte demand even more so.
Re: (Score:2)
Between the 3 DRAM producers? Could be, but there are only 3 of them and they each make very good margins on massive, growing amounts of DRAM production. Why would one of them decide to raise production, cut margins, and flood the market with cheap DRAM? All they have to do is act rationally and be disciplined and the profits will continue to be huge for the foreseeable future.
When 3 guys are acting rationally, maybe they conspired. But more likely, they're just rational.
In the USA, "rational" == antisocial asshole? (Score:1)
I know this is backed by sla shitton of scientific evidence nowadays, but it always surprises me, how casually Americans consider antisocial asshole behavior, or even psycopathic behavior, as "just normal" or "just business"...
Yes, if you need money, and can murder somebody, to take all their money, without ever getting caught, ... then doint it is "just" rational.
Doesn't mean it is sane behavior for heathy actual human beings!
Seriousy, *this* right here is the actual core of all the reasons the USA and cap
Re: (Score:2)
Did you ever stop to think about what makes you entitled to benefit from other people’s labor? What did you do for them? What do you do for anyone?
Why should companies make excess DRAM as a favor to you?
Re: (Score:2)
IBM was not stupid and sold it's hard disk drive division at the right time, not only will it kill hard disk drives but also optical drives. In terms of form factor and reduced sales cost, even for typical media content, the tiny form factor of flash ram versus optical storage media, plus digital delivery on top, will kill optical storage media, numbers will drop to low. It was inevitable, just greed and patents kept the price artificially high. This of course is sick psychopathic behaviour as flash ram is
Re: (Score:2)
No, Moore's Law doesn't apply to NAND storage.
Re: (Score:2)
Yes, rly.
Re: (Score:2)
Read the Wikipedia article you cited yourself:
“Due to its relatively simple structure and high demand for higher capacity, NAND flash memory is the most aggressively scaled technology among electronic devices. The heavy competition among the top few manufacturers only adds to the aggressiveness in shrinking the design rule or process technology node.[34] While the expected shrink timeline is a factor of two every three years per original version of Moore's law, this has recently been accelerated in th
Re: (Score:1)
uhm, I did.
Moores law is about the number of transistors in a dense integrated circuit. NAND is a dense integrated circuit full of transistors.
Any questions?
It's phase shift, not collusion (Score:2)
Those without the math background to understand differential equations and harmonic oscillation don't understand why a phase shift should exist, so incorrectly attribute it to collusion all the time. Yes sometimes it's collusion. But even in the complete absence of collusion, it will still happen.
Re: (Score:2)
Re: (Score:3)
Re: (Score:2)
Yes alongside the crimes regarding 10gb nics... I'm sorry is 10+ yrs of this technology commonplace in thousands of datacenter systems and massive transistor density increases still not enough for 10gb nics to be the slowest you see... period. Can someone explain to me why the $300 ultra high end cpu's of a few years ago are now 4x+ that price for the same tier? Must be all these improvements in density and reduction in manufacturing cost.
The problem is that 10GBASE-T transceivers draw a lot of power; the requirements for the physical interface are incompatible with processes smaller than 40nm. This is not so much of a problem for desktop NICs but it is a massive problem for switches.
2.5GBASE-T and 5GBASE-T have been developed because this problem has proven to be so intractable.
Re: (Score:2)
For $500 the consumer could easily have a private cloud in a box at this point. The major moves seem to be to ramp us prices until only the wealthy and enterprise can afford technology i
Finally! (Score:2, Insightful)
Looking forward to spending under $100 for a 1tb 2.5inch laptop drive. I'd also like to get a nice fat 512gb m.2 chip as well.
I can't justify their current prices. It was only a matter of time for the prices to drop. A collapse is a good thing!
Re: (Score:2)
Looking forward to spending under $100 for a 1tb 2.5inch laptop drive.
Sata 2.5" is dying fast, nearly all new designs are M.2. Faster, better. Desktop PC and enterprise will be the last island.
Re: (Score:2)
That and replacing the HDD shipped in a laptop whose motherboard has a SATA port but no M.2 slot.
Re: (Score:2)
Sata 2.5" is dying fast, nearly all new designs are M.2. Faster, better. Desktop PC and enterprise will be the last island.
SATA is much more universal than M.2. I like drives I can freely move between my desktops/servers and laptops. This was a welcome change from IDE/PATA, where smaller drives had smaller connectors, so you'd need adapters to put a 2.5'' drive into a desktop, for instance. I'm used to building small and silent Mini-ITX machines, and I've seen too many unnecessary distinctions between "large" desktop/server and "small" laptop/mini hardware.
I understand the benefits of newer tech, for example my latest laptop
Re: (Score:2)
Re: (Score:2)
Won't happen. At the "Crashed" price of 8 cents a GB, 1000GB is $80. Given markups and profits, that would be at best, around $150-$200.
And generally speaking, it's not likely to crash. If they can predict a crash a year ahead of time, then manufacturers will scale back their production. Prices might drop a bit, but not to the extremes of a quarter the price.
This is especially so as memory makers have a company that they basically hire to
Temporary (Score:2)
Demand for flash storage will be very high if flash storage becomes very cheap. Equilibrium won't be at a very low price.
Re: (Score:2)
Tell it to the market analysts. They aren't factoring in a huge demand increase at low prices.
Re: (Score:2)
DRAM needs it (Score:2)
While Flash has been steadily dropping in price for the past decade, DRAM has been jumping in price, nearly quadrupling over the past two years. If a Flash price crash will cause fabs to switch over to DRAM and bring those prices down, I say it can't happen soon enough.
Re:DRAM needs it (Score:4, Interesting)
Fabs mostly can't switch over to DRAM production. They aren’t technologically similar. And there are only 3 DRAM producers versus 6+ flash producers. Barriers to entry for competitive DRAM production for new producers are astronomically high.
Re: (Score:3)
... that's not what TFA says (I know, reading the articles? On Slashdot?). Planar flash and DRAM are similar enough to be switched easily, about as easily as CPU fabs switching from one design to another on the same node. 3D flash is another story, but there's still several planar flash fabs running, which (being the least cost-effective flash fabs) would almost certainly switch to DRAM if the price for flash crashes.
Re:DRAM needs it (Score:4, Interesting)
I disagree with TFA on that. The 3 DRAM makers have shown a very rational approach to increasing DRAM production — they try to prevent oversupply and keep margins up. They don’t have an incentive to change, and even if they did, the new generations of DRAMs are technologically very challenging to make, and DRAM bit density isn't growing very fast.
SK Hynix just announced they are spending $3B on a new fab:
https://www.anandtech.com/show... [anandtech.com]
If NAND is in vast oversupply and it's reasonable to simply convert NAND production to DRAM, then why build new fabs? Answer: because the combination of those two things isn't true enough to make that decision economical.
Re: (Score:3)
Which is good for nobody... including them. Producing and selling twice the product for the same profit is neutral for the manufacturer and twice as good for absolutely everyone else (except their competitor).
Re: (Score:2)
It's cool that you have more insight into what's best for those companies than all the people involved in making decisions there. It must be frustrating to be so smart and have all the answers regarding a subject where you have absolutely no direct involvement or experience.
Re: (Score:2)
In other words you have absolutely no logical rebuttal to my argument and therefore you attack me. The messenger has no logical impact on the validity on argument, the validity of the message is independent of any qualification or lack
Re: (Score:2)
You should write the CEOs of these 3 companies a letter explaining what they are doing wrong and how things will be better if they do it your way. Maybe they will put you on the board of directors.
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Does that mean we can expect lower prices for smartphones and laptops that contain flash storage
Of course it does, and this will further exacerbate the value gap between flagship phones and second tier phones. Historically, the higher value tier has always won this competition. It also means even more Chromebooks, with 128GB becoming the new minimum and 512GB in the midrange. Good news for schools and Linux road warriors.
Meanwhile (Score:2, Interesting)
Spinning platter drives have been the same price per Gb for a decade.
Re:Meanwhile (Score:5, Informative)
Do you always post nonsense to the internet without 10 seconds of research? Hard Drive Cost per GB Over Time [backblaze.com]
Re: (Score:2)
Nearly flat for half a decade, my mistake.
Re: (Score:2)
So in the world you live in, dropping from $.06/GB to $.025 is "flat". Or maybe you have issues admitting an error. Combined with overestimation of your own knowledge and a trigger posting finger, it's a kooky cocktail of intellectual sludge.
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Everybody knows that solid state will eventually push hdd fully into backup, mass storage and nearline roles. Nobody yet knows whether solid state will eventually take over those roles too, if they say they know then they are lying. And in fact, most of the world's data is on mass storage right now. Even if hdd does eventually fall into a purely legacy role, how long will it take? It could be as long as twenty years or as quick as five.
One trend that is pretty easy to predict: enterprise 5400 rpm will start
2011, 2TB for $49 (Score:2)
Re: (Score:2)
Re: Meanwhile (Score:2)
Hard drives have reached a density plateau. 12TB is pretty common these days but reading/writing 12TB at 10MB/s is ludicrously slow. I am currently deploying 6TB because itâ(TM)s the closest rebuild times come to acceptable (24-48h). Having a RAID6 rebuild in 3-5 days is too risky.
Re: (Score:2)
They've also not been found in any of my computers.
Re: Meanwhile (Score:2)
I couldnâ(TM)t disagree more. 13 years ago a commercial spinning-platter 158T array was nearly a million beans, and a project I was in offsited (but never expected to use) an essential 30T of that array into a small beige box cluster for about a kilo buck per terabyte.
Now I can buy 6T for $110. 30T is $500-600.
The descent is less steep than a generation ago, but the physics are harder. And youâ(TM)re nuckin futz to say pricing has been stagnant over 10 years.
Good! (Score:2)
The fewer moving components that are put in computers the better! I still remember reading posts on /. which declared the price of FLASH storage would never drop below the price of HDD storage. I'm looking forward to the world of solid state storage everywhere.
HDD death has been greatly exaggerated (Score:2)
Re: (Score:2, Informative)
The problem with Flash storage is not only price. It's also data retention time. The times where unpowered flash was holding data for 10 or more years are long gone.
How long can a current SSD hold the data without power? I can take an old HD from my stack, power it up after 2 or more years and still read the data from it. Can you do the same with a current SSD?
Or could it mean that if you don't power up your computer for a month or two will already have you lose data?
Re: (Score:2)
I'm planning on buying a few 2.5" SSDs and enclosures to back up my music and movies for long term archival/storage, since a mechanical HDD isn't the best to stick on a shelf for years at a time in between uses. SSDs aren't perfect, but they have no moving parts and are harder to damage (BluRays and DVDs can get scratched and degrade over time). SSDs should last a long while for a back up medium that's only occasionally written to.
SSDs are not a good choice for archival use because of the low retention time of dense unpowered NAND Flash memory which is only months to years. This might be avoided if they are left powered assuming they do idle time scrubbing but that presents other problems.
Free hard drive magnets may come to an end... (Score:2)
...if spinning rust disappears from the market!
Harvest 'em while you can. Get a crate of dead drives from the local recycler, and strip all the magnets. You'll be telling your mystified grandkids about the glory days when incredibly-powerful magnets were just free for the taking, for anyone willing to wield a screwdriver.
About 2 years ago, when I started putting 480GB SSDs in things, I commented that I'm probably never gonna buy another spinning drive in my life. SSDs are at the point where they can meet my
Awesome! (Score:1)
Ferret