i do care about performance but really for MOST USE CASES the benefit of going beyond sata SSD just doesn't matter significantly.
sadly the price has just not moved down in the past 6 months+ and instead it has gone up since December 2016. Personally I am waiting for 1TB m.2 drive for around $150USD which should have been possible with the number of enhancements - TLC, VNAND, ... yet, it seems there has been no downstream impact. worse yet it seems like the enthusiasm has just died down to even release new products. we are rarely seeing new models getting launched like previous years - dozen every quarter across all producers...
Actually the NAND scarcity is artificially produced by chip makers so they get better margins. It is not an issue of manufacturing capacity, they are deliberately holding out to maximize profits. And regulators do not mind.
No, it really is demand growing faster than fabs can be brought online. Fabs are a fixed cost to set up, chip makers gain NOTHING by artificially reducing output. All that accomplishes is reducing sales, which reduces profits.
Don't be naive. Couple of months back China announced the construction of a NAND mega-fab and all "analysts" were whining how "disruptive" that would be to the flash market.
If the shortage is genuine, how is more production disruptive in any way? That would be good news everyone should cheer at.
It is only disruptive if it will end the artificial scarcity and reduce prices.
There is the cost to build a fab. There is the cost to operate a fab. Chip makers are making more money on doing less work. The fab is not going anywhere, it is already built. It doesn't reduce sales, they are not throwing away demand, they just keep supply really tight in order to keep prices inflated. They still sell as much as people are willing to buy, just at considerably higher prices.
It is not unprecedented, misuse of artificial scarcity goes as further back as human history.
Or explain how come SSD availability is actually great, if there is indeed nand SHORTAGE then there should also be SSD shortage. Yet there is none, 95% of the ~200 SSD models my local supplier offers are available in stock and in good quantities.
Or explain how come day after day we have more and more obscure brands of SSDs popping out, offering no tangible benefits, reeking of "why even bother"? Why would NAND makers, who are all vertically integrated and aside from NAND also make SSDs, will be willing to sell their limited supply chips to third parties to generate profits?
There are numerous hints that the NAND shortage is but a play for the public, and numerous indications that supply is actually ample. There is no shortage, just careful planning to keep supply tight, and not between OEMs and consumers, but between chipmakers and OEMs. That's a classic example of price gouging.
Companies are seeing that 2D NAND generation is not cost effective in the long run. The manufacturers are converting 2D NAND production capacity to 3D NAND production capacity. That production capcity will not be on-line until later in 2017.
"For 4Q16, its NAND flash business registered a sequential growth of 11~15% for bit shipments and an increase of more than 5% in ASP. These results were attributed to strong demand for high-capacity eMMC, UFS and SSD products. Thus, Samsung's NAND flash revenue for the fourth quarter advanced by nearly 20% sequentially.
Looking at Samsung's production plans, the 2D- to 3D-NAND conversion at the Line 16 facility continues. Line 17 and the plant in Pyeongtaek, South Korea are designated for 3D-NAND production, but they are not expected to make significant contribution to Samsung's 3D-NAND capacity until after the second quarter of this year. Therefore, bit shipments for this first quarter will fall by around 4~9% compared with the previous quarter. If Samsung's plans proceed according to schedule, then its NAND flash bit shipments will again expand after the second quarter."
SK Hynix: In 4Q16, SK Hynix scaled back SSD shipments while increasing eMCP shipments in response to the demand from Chinese smartphone brands. Consequently, the supplier's bit shipments for the quarter post a slight sequential drop of 3% but its ASP of NAND flash chips saw a large sequential hike of 14%. On the whole, NAND flash business grew its revenue by 9% sequentially in the fourth quarter, totaling around $1.16 billion.
Going into 1Q17, SK Hynix's NAND flash bit shipments will also be affected by the transition to 3D NAND as well as declining smartphone shipments during the period. The supplier's bit shipments could drop by around 5% sequentially. However, ASP of NAND flash chips is going to be on a steady rise due to the ongoing market undersupply.
With regard to SK Hynix's transition to 3D NAND, the advanced architecture is estimated to account for 10% of the supplier's bit shipments for the first quarter of 2017. The firm is currently releasing 48-layer products and will also be launching 72-layer products in the second half of this year.
Toshiba, WD, similar information...
Micron: NAND flash revenue for Western Digital's 2FQ17 rose by about 20% sequentially due to increase in both bit shipments and ASP. The firm also continue to perform strongly in the SSD market, indicating that the company's acquisition of SanDisk has led to an effective integration of HDD and SSD lines.
In terms of product planning, Western Digital has already added its own branded 64-layer 3D NAND chips into its retail portfolio. Sampling of PC-OEM SSDs based on the same memory technology is also taking place in this first quarter. The share of 3D NAND products in the supplier's total NAND flash output is expected to exceed 50% by the end of 2017.
Intel: Intel Intel posted a large sequential increase of more than 25% for its NAND flash bit shipments in the fourth quarter of 2016 because of the high demand for enterprise-grade SSDs. The supplier's NAND flash revenue for the quarter also grew 25% sequentially to $816 million.
With regard to product planning, it will gradually scale down the production of its 20nm and 25nm products while increasing the weight of enterprise-grade SSD products based on both 3D NAND and MLC technologies. These 3D NAND MLC SSDs can be priced competitively and help Intel improve the overall cost structure and profitability. At the same time, the firm is now starting the mass production of 3D NAND TLC products.
It's within the realm of possibility that price fixing is indeed going on, but unless someone has proof, it is pointless to speculate on it. The prices are either worth buying to you or they are not. Vote with your wallet.
ddrive, you're mixing up two different definitions of "shortage". Right there, there is a NAND shortage in a colloquial, non-economic sense, simply meaning that NAND is scarce, and that there is not as much NAND as we'd like there to be. In this sense, there is a shortage of almost all goods, because we'd always like there to be more of everything. So when we say that there is a NAND shortage, we simply mean that the supply of NAND is less than what we are accustomed to its being. (The most likely reason is that NAND fabs are retooling to transition from planar to 3D NAND, so in the meantime, there has been a negative supply shock.)
On the other hand, when you discuss an SSD shortage, you silently switch you a different definition of shortage, viz. the economic definition. In economics, "shortage" refers to when the quantity demanded exceeds the quantity supplied ***at a given price***. You are assuming the price is given, and you are saying that the shelves are well-stocked. Well, this just means that the price is right.
For example, suppose a new 1 TB SSD cost only $100. We'd find a ton of willing buyers, but almost no willing sellers. That would be a shortage. But when the price rises to say $1,000, we find an equal number of buyers and sellers, and the market clears without incident. You're saying that SSDs are in stock, and so there is no shortage. Well, this just means that the price is apparently being set efficiently, so that quantity demanded equals quantity supplied (over some reasonable span of time).
Now, suppose the price of NAND skyrocketed, and so the price of SSDs skyrocketed as well. But suppose that the shelves remained well-stocked. This would mean there was a shortage in the colloquial sense of "not as much as we'd like", but NOT a shortage in the economic sense of "the price is equating supply and demand".
I'd like you to substantiate YOUR claims with fact. Where is your evidence of artificial scarcity and a global conspiracy to fix prices? Extraordinary claims require extraordinary evidence, so the burden of proof is on you as the party with the more outlandish claim.
I am not saying it is impossible. My memory IS long enough to remember when the SRAM manufacturers got nailed for conspiring to keep prices artificially high. But... your certainty speaks of either foolishness or ironclad evidence, and we would ALL like to see your evidence.
Reducing supply will increase profits only for a monopolist. If you think that a NAND producer is behaving this way, then you are implicitly asserting that there is a single monopolistic NAND producer, or else that all of the NAND producers have successfully formed a monopolistic cartel amongst themselves, so that they no longer compete with each other.
I think that both possibilities are highly dubious. Therefore, it is more likely that the NAND industry is competitive, and that the the scarcity is natural, not artificial.
Incidentally, this time-lag in retooling factories to produce 3D instead of planar, and the disruption to the industry caused by this time-lag, is a great illustration of Ludwig Lachmann's claim, in Capital and Its Structure, that capital is not a generic lump of clay that can be costlessly and instantly reallocated, the way some macroeconomists imagine. Instead, Lachmann says, capital is heterogeneous and multi-specific - meaning, that capital equipment can be used for multiple purposes, but not for infinite purposes, and that any given piece of capital equipment is better-suited for some purposes than others.
"Couple of months back China announced the construction of a NAND mega-fab and all "analysts" were whining how "disruptive" that would be to the flash market"
I didn't know that a NAND mega-fab just takes a couple of months to come online and start making MEGA NANDs.
Really depends on what your "most use cases" is. Full shutdown, restart, office and browser stuff? Sure. high-res video editing with multiple uncompressed streams? Not so much. For myself, I really wish my laptop with 16 gigs or ram would hybernate faster... and I wouldn't have to use RAM disk to speed up large scale database stuff.
Anandtech has consistently been reporting that there is a global NAND shortage, that's driving prices higher. The market is shifting to 3D NAND. That takes time. New broducts based on that 3D NAND. They're not out yet, but there's been a number of 3D NAND product announcements in the Anandtech pipeline. When these 3D products start hitting the market, I would expect to see some prices drop.
Are you complaining about all the tempered glass too? Some people actually enjoy their hardware and don't want a crusty box tucked away without thought.
Motherboards with full OLED coverage that can double as a 100% sRGB color space monitor. Make sure you other components have the same coverage or they'll obscure the display. Also, purchase this nifty new tool for calibrating components to each other ... also with full coverage OLEDs.
Looking forward to this. Hope the performance lives up to specs and that the price is reasonable. We need a good Samsung alternative in the same performance segment.
I almost picked up a 1TB M8PeG (M.2 w/heatspreader) when they went on sale right before going out of stock *everywhere*. Had I known that was going to be it I'd have definitely pulled the trigger. I'm still somewhat tempted to pick up the HH.HL AIC model, but I can see that being far less practical (compared to M.2) if I want to transfer it to a new build.
So we are looking at a sequential performance somewhere between Samsung 960 Pro and 960 EVO, presumably with longer sustained sequential writing speed than EVO. Not bad for a TLC drive, though the reported writing speed is probably until the SLC cache fills up. However we need to see IOPS numbers. Where are the IOPS numbers? I doubt they will manage to break the read & write 300K barrier (for their 1TB drive). If they do, and it is priced aggressively against 960 EVO Samsung could finally have some competition.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
41 Comments
Back to Article
peevee - Friday, June 2, 2017 - link
The PCIe card is x4, right?DanNeely - Friday, June 2, 2017 - link
Yes. It's x4 physically, and to hit the rated speeds (>2GBps) needs a 3.0 x4 link.XZerg - Friday, June 2, 2017 - link
i do care about performance but really for MOST USE CASES the benefit of going beyond sata SSD just doesn't matter significantly.sadly the price has just not moved down in the past 6 months+ and instead it has gone up since December 2016. Personally I am waiting for 1TB m.2 drive for around $150USD which should have been possible with the number of enhancements - TLC, VNAND, ... yet, it seems there has been no downstream impact. worse yet it seems like the enthusiasm has just died down to even release new products. we are rarely seeing new models getting launched like previous years - dozen every quarter across all producers...
BillyONeal - Friday, June 2, 2017 - link
Neither TLC nor VNAND reduce costs enough to do that, and demand for NAND has gone through the roof. Higher demand == higher prices.ddriver - Friday, June 2, 2017 - link
Actually the NAND scarcity is artificially produced by chip makers so they get better margins. It is not an issue of manufacturing capacity, they are deliberately holding out to maximize profits. And regulators do not mind.edzieba - Friday, June 2, 2017 - link
No, it really is demand growing faster than fabs can be brought online. Fabs are a fixed cost to set up, chip makers gain NOTHING by artificially reducing output. All that accomplishes is reducing sales, which reduces profits.ddriver - Friday, June 2, 2017 - link
Don't be naive. Couple of months back China announced the construction of a NAND mega-fab and all "analysts" were whining how "disruptive" that would be to the flash market.If the shortage is genuine, how is more production disruptive in any way? That would be good news everyone should cheer at.
It is only disruptive if it will end the artificial scarcity and reduce prices.
There is the cost to build a fab. There is the cost to operate a fab. Chip makers are making more money on doing less work. The fab is not going anywhere, it is already built. It doesn't reduce sales, they are not throwing away demand, they just keep supply really tight in order to keep prices inflated. They still sell as much as people are willing to buy, just at considerably higher prices.
It is not unprecedented, misuse of artificial scarcity goes as further back as human history.
melgross - Friday, June 2, 2017 - link
No, it's very high demand. Don't make up conspiracies that don't exist.ddriver - Friday, June 2, 2017 - link
Care to substantiate those claims with facts?Or at the very least provide an adequate explanation why increased NAND supply would be DISRUPTIVE?
Humor me.
ddriver - Friday, June 2, 2017 - link
Or explain how come SSD availability is actually great, if there is indeed nand SHORTAGE then there should also be SSD shortage. Yet there is none, 95% of the ~200 SSD models my local supplier offers are available in stock and in good quantities.ddriver - Friday, June 2, 2017 - link
Or explain how come day after day we have more and more obscure brands of SSDs popping out, offering no tangible benefits, reeking of "why even bother"? Why would NAND makers, who are all vertically integrated and aside from NAND also make SSDs, will be willing to sell their limited supply chips to third parties to generate profits?There are numerous hints that the NAND shortage is but a play for the public, and numerous indications that supply is actually ample. There is no shortage, just careful planning to keep supply tight, and not between OEMs and consumers, but between chipmakers and OEMs. That's a classic example of price gouging.
mischlep - Friday, June 2, 2017 - link
Companies are seeing that 2D NAND generation is not cost effective in the long run. The manufacturers are converting 2D NAND production capacity to 3D NAND production capacity. That production capcity will not be on-line until later in 2017.http://www.storagenewsletter.com/2017/03/10/nand-f...
SAMSUNG:
"For 4Q16, its NAND flash business registered a sequential growth of 11~15% for bit shipments and an increase of more than 5% in ASP. These results were attributed to strong demand for high-capacity eMMC, UFS and SSD products. Thus, Samsung's NAND flash revenue for the fourth quarter advanced by nearly 20% sequentially.
Looking at Samsung's production plans, the 2D- to 3D-NAND conversion at the Line 16 facility continues. Line 17 and the plant in Pyeongtaek, South Korea are designated for 3D-NAND production, but they are not expected to make significant contribution to Samsung's 3D-NAND capacity until after the second quarter of this year. Therefore, bit shipments for this first quarter will fall by around 4~9% compared with the previous quarter. If Samsung's plans proceed according to schedule, then its NAND flash bit shipments will again expand after the second quarter."
SK Hynix:
In 4Q16, SK Hynix scaled back SSD shipments while increasing eMCP shipments in response to the demand from Chinese smartphone brands. Consequently, the supplier's bit shipments for the quarter post a slight sequential drop of 3% but its ASP of NAND flash chips saw a large sequential hike of 14%. On the whole, NAND flash business grew its revenue by 9% sequentially in the fourth quarter, totaling around $1.16 billion.
Going into 1Q17, SK Hynix's NAND flash bit shipments will also be affected by the transition to 3D NAND as well as declining smartphone shipments during the period. The supplier's bit shipments could drop by around 5% sequentially. However, ASP of NAND flash chips is going to be on a steady rise due to the ongoing market undersupply.
With regard to SK Hynix's transition to 3D NAND, the advanced architecture is estimated to account for 10% of the supplier's bit shipments for the first quarter of 2017. The firm is currently releasing 48-layer products and will also be launching 72-layer products in the second half of this year.
Toshiba, WD, similar information...
Micron:
NAND flash revenue for Western Digital's 2FQ17 rose by about 20% sequentially due to increase in both bit shipments and ASP. The firm also continue to perform strongly in the SSD market, indicating that the company's acquisition of SanDisk has led to an effective integration of HDD and SSD lines.
In terms of product planning, Western Digital has already added its own branded 64-layer 3D NAND chips into its retail portfolio. Sampling of PC-OEM SSDs based on the same memory technology is also taking place in this first quarter. The share of 3D NAND products in the supplier's total NAND flash output is expected to exceed 50% by the end of 2017.
Intel:
Intel
Intel posted a large sequential increase of more than 25% for its NAND flash bit shipments in the fourth quarter of 2016 because of the high demand for enterprise-grade SSDs. The supplier's NAND flash revenue for the quarter also grew 25% sequentially to $816 million.
With regard to product planning, it will gradually scale down the production of its 20nm and 25nm products while increasing the weight of enterprise-grade SSD products based on both 3D NAND and MLC technologies. These 3D NAND MLC SSDs can be priced competitively and help Intel improve the overall cost structure and profitability. At the same time, the firm is now starting the mass production of 3D NAND TLC products.
sleepeeg3 - Friday, June 2, 2017 - link
I think the whole hypothetical argument is stupid, but for the record, there has been collusion with the memory companies in the past to fix prices. Multiple times.https://www.law360.com/competition/articles/5518/h...
https://www.law360.com/articles/336763/sandisk-cut...
It's within the realm of possibility that price fixing is indeed going on, but unless someone has proof, it is pointless to speculate on it. The prices are either worth buying to you or they are not. Vote with your wallet.
woggs - Friday, June 2, 2017 - link
OMG. The demand is from ENTERPRISE customers for ENTERPRISE drives.lmcd - Friday, June 2, 2017 - link
Do you not get how econ works? Higher prices = lower demand at the price point.Mikewind Dale - Saturday, June 3, 2017 - link
ddrive, you're mixing up two different definitions of "shortage". Right there, there is a NAND shortage in a colloquial, non-economic sense, simply meaning that NAND is scarce, and that there is not as much NAND as we'd like there to be. In this sense, there is a shortage of almost all goods, because we'd always like there to be more of everything. So when we say that there is a NAND shortage, we simply mean that the supply of NAND is less than what we are accustomed to its being. (The most likely reason is that NAND fabs are retooling to transition from planar to 3D NAND, so in the meantime, there has been a negative supply shock.)On the other hand, when you discuss an SSD shortage, you silently switch you a different definition of shortage, viz. the economic definition. In economics, "shortage" refers to when the quantity demanded exceeds the quantity supplied ***at a given price***. You are assuming the price is given, and you are saying that the shelves are well-stocked. Well, this just means that the price is right.
For example, suppose a new 1 TB SSD cost only $100. We'd find a ton of willing buyers, but almost no willing sellers. That would be a shortage. But when the price rises to say $1,000, we find an equal number of buyers and sellers, and the market clears without incident. You're saying that SSDs are in stock, and so there is no shortage. Well, this just means that the price is apparently being set efficiently, so that quantity demanded equals quantity supplied (over some reasonable span of time).
Now, suppose the price of NAND skyrocketed, and so the price of SSDs skyrocketed as well. But suppose that the shelves remained well-stocked. This would mean there was a shortage in the colloquial sense of "not as much as we'd like", but NOT a shortage in the economic sense of "the price is equating supply and demand".
DanNeely - Friday, June 2, 2017 - link
An increase of supply well in excess of the expected future demand will cause a price crash. Textbook disruption right there.Lord of the Bored - Thursday, June 8, 2017 - link
I'd like you to substantiate YOUR claims with fact. Where is your evidence of artificial scarcity and a global conspiracy to fix prices? Extraordinary claims require extraordinary evidence, so the burden of proof is on you as the party with the more outlandish claim.I am not saying it is impossible. My memory IS long enough to remember when the SRAM manufacturers got nailed for conspiring to keep prices artificially high. But... your certainty speaks of either foolishness or ironclad evidence, and we would ALL like to see your evidence.
Mikewind Dale - Saturday, June 3, 2017 - link
Reducing supply will increase profits only for a monopolist. If you think that a NAND producer is behaving this way, then you are implicitly asserting that there is a single monopolistic NAND producer, or else that all of the NAND producers have successfully formed a monopolistic cartel amongst themselves, so that they no longer compete with each other.I think that both possibilities are highly dubious. Therefore, it is more likely that the NAND industry is competitive, and that the the scarcity is natural, not artificial.
Mikewind Dale - Saturday, June 3, 2017 - link
Incidentally, this time-lag in retooling factories to produce 3D instead of planar, and the disruption to the industry caused by this time-lag, is a great illustration of Ludwig Lachmann's claim, in Capital and Its Structure, that capital is not a generic lump of clay that can be costlessly and instantly reallocated, the way some macroeconomists imagine. Instead, Lachmann says, capital is heterogeneous and multi-specific - meaning, that capital equipment can be used for multiple purposes, but not for infinite purposes, and that any given piece of capital equipment is better-suited for some purposes than others.Gasaraki88 - Thursday, June 8, 2017 - link
"Couple of months back China announced the construction of a NAND mega-fab and all "analysts" were whining how "disruptive" that would be to the flash market"I didn't know that a NAND mega-fab just takes a couple of months to come online and start making MEGA NANDs.
woggs - Friday, June 2, 2017 - link
OMG you're an idiot. Factories MUST run at full capacity to make money.Reflex - Friday, June 2, 2017 - link
Yay for unsupported assertions!nagi603 - Friday, June 2, 2017 - link
Really depends on what your "most use cases" is. Full shutdown, restart, office and browser stuff? Sure. high-res video editing with multiple uncompressed streams? Not so much. For myself, I really wish my laptop with 16 gigs or ram would hybernate faster... and I wouldn't have to use RAM disk to speed up large scale database stuff.mischlep - Friday, June 2, 2017 - link
Anandtech has consistently been reporting that there is a global NAND shortage, that's driving prices higher. The market is shifting to 3D NAND. That takes time. New broducts based on that 3D NAND. They're not out yet, but there's been a number of 3D NAND product announcements in the Anandtech pipeline. When these 3D products start hitting the market, I would expect to see some prices drop.CheapSushi - Tuesday, June 6, 2017 - link
Wait for QLC.Gasaraki88 - Thursday, June 8, 2017 - link
NAND shortage has cause prices to go up this past year and the shortage will last till the end of this year.mooninite - Friday, June 2, 2017 - link
Why does everything need RGB LEDs in 2017? Who decided this?close - Friday, June 2, 2017 - link
Maybe they ran out of idea to make many otherwise relatively boring product launches "interesting". "Now with RGB LEDs".melgross - Friday, June 2, 2017 - link
Gamer influence. They need LEDs on every internal component, or they can't play.CheapSushi - Tuesday, June 6, 2017 - link
Are you complaining about all the tempered glass too? Some people actually enjoy their hardware and don't want a crusty box tucked away without thought.close - Friday, June 2, 2017 - link
Wow, RGB LEDs on a PCIe SSD. What will they think of next.ddriver - Friday, June 2, 2017 - link
RGB LEDs with RBG LEDs on them. Produce better color coverage and accuracy.hyno111 - Friday, June 2, 2017 - link
SATA SSD with RGB LED. (But it already exist..)StrangerGuy - Friday, June 2, 2017 - link
Watercooled PSUs (what a great idea!) has been done, so SATA and power cables with RGB leds.BedfordTim - Friday, June 2, 2017 - link
An OLED information panel?Water cooling?
BurntMyBacon - Friday, June 2, 2017 - link
Motherboards with full OLED coverage that can double as a 100% sRGB color space monitor. Make sure you other components have the same coverage or they'll obscure the display. Also, purchase this nifty new tool for calibrating components to each other ... also with full coverage OLEDs.CheapSushi - Tuesday, June 6, 2017 - link
Tempered glass on an SSD.jjj - Friday, June 2, 2017 - link
Given the previous gen, looking forward to this one.Any M8Se review in the pipeline?
MrCommunistGen - Friday, June 2, 2017 - link
Looking forward to this. Hope the performance lives up to specs and that the price is reasonable.We need a good Samsung alternative in the same performance segment.
I almost picked up a 1TB M8PeG (M.2 w/heatspreader) when they went on sale right before going out of stock *everywhere*. Had I known that was going to be it I'd have definitely pulled the trigger. I'm still somewhat tempted to pick up the HH.HL AIC model, but I can see that being far less practical (compared to M.2) if I want to transfer it to a new build.
Santoval - Friday, June 2, 2017 - link
So we are looking at a sequential performance somewhere between Samsung 960 Pro and 960 EVO, presumably with longer sustained sequential writing speed than EVO. Not bad for a TLC drive, though the reported writing speed is probably until the SLC cache fills up. However we need to see IOPS numbers. Where are the IOPS numbers? I doubt they will manage to break the read & write 300K barrier (for their 1TB drive). If they do, and it is priced aggressively against 960 EVO Samsung could finally have some competition.