Intel and Micron are set to announce 25nm chips

PC World is reporting Intel and Micron plan to announce 25 nanometer flash memory chips on Monday. The news comes from a research note by Objective Analysis sent out ahead of the official unveiling. According to PC World, an Intel rep confirmed the new chips, which are set to be used in solid state drives, portable media players, and smartphones. The representative states production is expected in the second quarter. Samples are given to customers to allow adequate time to integrate the chips in product designs.

The 25nm NAND technology puts Intel and Micron ahead of competitors, such as Samsung, who is set to deploy 30nm technology in product lines by the end of 2010. Objective Analysis notes the manufacturing cost of the 25nm chips will be approximately $.50 per gigabyte. This is a substantial decrease from the estimated $1.75 per gigabyte cost of 45nm flash. SSD storage sizes should double with the new 25nm chips. Be on the lookout for these new devices in Q4 2010.

To produce the 25nm chips, Intel and Micron (jointly known as IMFT) are utilizing immersion lithography. The resulting 8GB die consists of many smaller NAND chips, slashing the required amount of dies per device in half. For instance, a 256GB SSD can be built with 32 of the 8GB NAND dies instead of 64 dies with the previous 34nm technology.

Computerworld quoted Troy Winslow, director of NAND marketing at Intel, as saying the die is "small enough to fit through the hole in the middle of a compact disc, yet packs more than 10 times the data capacity of that CD." Winslow also noted the new chips have the same 5,000 write-erase cycles as the 34nm technology.

Report a problem with article
Previous Story

Beware of the tax refund phishing scam [UK]

Next Story

Windows 7 on the new Apple iPad will be possible

27 Comments

Commenting is disabled on this article.

What a great pair Intel and Micron make. Intel currently in court for anti-trust violations, already given a billion dollars to AMD to 'settle' it and Micron involved in Price Fixing of DRAM and fined 180 Million dollars.

Vice said,
What a great pair Intel and Micron make. Intel currently in court for anti-trust violations, already given a billion dollars to AMD to 'settle' it and Micron involved in Price Fixing of DRAM and fined 180 Million dollars.

(snipped) Micron is the last North American dram maker and is holding on for dear life as they're unable to compete with prices set by asian competitors seeing government subsidies and selling below costs. Just because someone was fined, does not mean they were in the wrong, were guillty of the accusations, or they were not justifiable. I'd say if I was trying to protect tens of thousands of jobs and the competition was "cheating", price fixing might be a reasonable route to put price where they should be, not that they even did this.

Edited by Eric, Feb 2 2010, 3:18am :

thornz0 said,

(snipped) Micron is the last North American dram maker and is holding on for dear life as they're unable to compete with prices set by asian competitors seeing government subsidies and selling below costs. Just because someone was fined, does not mean they were in the wrong, were guillty of the accusations, or they were not justifiable. I'd say if I was trying to protect tens of thousands of jobs and the competition was "cheating", price fixing might be a reasonable route to put price where they should be, not that they even did this.

They did the crime they must pay the price. And they were working with those very same 'Asian Competitors' to fix prices so any moral high ground you may perceive them as having is gone. And I did see the (Snipped) comment before a mod got to it but lets leave that in the trash.

Edited by Vice, Feb 2 2010, 6:56am :

Vice said,

They did the crime they must pay the price. And they were working with those very same 'Asian Competitors' to fix prices so any moral high ground you may perceive them as having is gone. And I did see the (Snipped) comment before a mod got to it but lets leave that in the trash.

You can't dismiss an entire argument simply by saying they did the crime because they paid for it. If their competition is getting subsidies and selling below cost to shut out competition, how then could they be in league with that same competition. It makes no sense *sigh* And yeah, I stand by my initial comment, I'm glad you saw it. Micron is a good company, and if you did any research into them at all you'd know that, but you'd rather just attack them out of context to the news.

"Winslow also noted the new chips have the same 5,000 write-erase cycles as the 34nm technology."
5,000 write-erase cycles lifetime is extremely poor IMO.

g0dlike said,
"Winslow also noted the new chips have the same 5,000 write-erase cycles as the 34nm technology."
5,000 write-erase cycles lifetime is extremely poor IMO.
Thats not much of an issue IMO. Intel's older MLC drives had only 10,000 write cycles per cell and it is guaranteed to work/run fine without failure for 5 years writing at 100GB/day.

Now, with 5,000 write cycles, you are looking at either 2.5 years at 100GB/day, or 5 years at only 50GB/day.

If you are running your system on the SSD, browsing the internet and playing games and what not, you are likely writing no more than 5GB/day to the SSD. At a rate of 5GB per day a SSD with only 5,000 write cycles per cell, it is going to take you around 50 years to use them up (depends on the size of the drive and how many cells it has).

To add more to it, a 120GB drive with 5,000 cycles. You are writing at a constant 20GB/day (you install lots of programs and what not). Your SSD is going to rewrite a each cell approximately once every 6 days. At 5,000 cycles, thats 5,000 * 6 = 30,000 DAYS before you run out of write cycles. Is 82 years not enough time for you for a SSD? By that time we likely will be well beyond SSDs.

As for the 5,000 write cycles, I highly double that is for the entire drive, that is likely per cell (each 25nm cell has 5,000 write cycles) so the actual time the drive will last will depend on the number of cells (which increases as the drive space does), how much you write to it each day, and a few other factors such as how good its load leveling algorithm is.

Sources: http://www.anandtech.com/cpuchipsets/intel/showdoc.aspx?i=3403&p=4
http://www.overclockers.com/hdd-vs-ssd/

Edited by Nagisan, Feb 1 2010, 7:45am :

Nagisan said,
Thats not much of an issue IMO. Intel's older MLC drives had only 10,000 write cycles per cell and it is guaranteed to work/run fine without failure for 5 years writing at 100GB/day.

Now, with 5,000 write cycles, you are looking at either 2.5 years at 100GB/day, or 5 years at only 50GB/day.

If you are running your system on the SSD, browsing the internet and playing games and what not, you are likely writing no more than 5GB/day to the SSD. At a rate of 5GB per day a SSD with only 5,000 write cycles per cell, it is going to take you around 50 years to use them up (depends on the size of the drive and how many cells it has).

To add more to it, a 120GB drive with 5,000 cycles. You are writing at a constant 20GB/day (you install lots of programs and what not). Your SSD is going to rewrite a each cell approximately once every 6 days. At 5,000 cycles, thats 5,000 * 6 = 30,000 DAYS before you run out of write cycles. Is 82 years not enough time for you for a SSD? By that time we likely will be well beyond SSDs.

As for the 5,000 write cycles, I highly double that is for the entire drive, that is likely per cell (each 25nm cell has 5,000 write cycles) so the actual time the drive will last will depend on the number of cells (which increases as the drive space does), how much you write to it each day, and a few other factors such as how good its load leveling algorithm is.

Sources: http://www.anandtech.com/cpuchipsets/intel/showdoc.aspx?i=3403&p=4
http://www.overclockers.com/hdd-vs-ssd/

Thanx, Nice info

Thanks for that information, Nagisan. Before this I had no idea how long 5,000 write-erase cycles actually lasted in real terms. Cheers.

iamwhoiam said,
Lower manufacturing costs rarely, if ever, gets passed on to the consumer.

LMFAO. No, never. Prices just go down because, you know, they are nice and all.

Mikeyx11 said,

LMFAO. No, never. Prices just go down because, you know, they are nice and all.
You can bet your ass that the money they saved making these new chips won't equate to cheaper SSD drives. SSDs won't really be cheap until they're in widespread use.

iamwhoiam said,
You can bet your ass that the money they saved making these new chips won't equate to cheaper SSD drives. SSDs won't really be cheap until they're in widespread use.

I agree with you. SSD are going to be probably few dollars cheap but not yet for widespread use. I believe by 2012+ is when widespread use is going to start.

iamwhoiam said,
You can bet your ass that the money they saved making these new chips won't equate to cheaper SSD drives. SSDs won't really be cheap until they're in widespread use.

Umm...I can pretty much guarantee they will lower the price accordingly. WHy? Because at the current cost only enthusiasts with money to burn can afford them. They want the masses to adopt them, they have to become affordable. The cheaper they are, the more they sell. The more they sell, the more money they make. I can guarantee you that if they have to not make as high a percentage on each one sold in order to make them affordable, they will lower their per unit profit and have their product widely adopted. Until they can lower their manufacturing costs to a point they will be able to sell them at an affordable price, they will keep to a pre determined percentage profit per unit in order to make the most profit they can in order to further research and development.

They WANT these drives to be cheap. I'm not sure how you can't understand basic economics.

iamwhoiam said,
You can bet your ass that the money they saved making these new chips won't equate to cheaper SSD drives. SSDs won't really be cheap until they're in widespread use.

And what does widespread use mean for the manufacturers? It means economies of scale, which means CHEAPER MANUFACTURING COSTS which then results in CHEAPER PRICES. Widespread adoption comes as manufacturing prices go down through R&D, which causes consumer prices to go down, so more people buy the product, then they can produce more of them cheaper, then it results in even lower prices and so on. Is it really that difficult to understand?

"Objective Analysis notes the manufacturing cost of the 25nm chips will be approximately $.50 per gigabyte. This is a substantial decrease from the estimated $1.75 per gigabyte cost of 45nm flash."

Good news!

Pixil Eyes said,
"Objective Analysis notes the manufacturing cost of the 25nm chips will be approximately $.50 per gigabyte. This is a substantial decrease from the estimated $1.75 per gigabyte cost of 45nm flash."

Good news!

very :)

Pixil Eyes said,
"Objective Analysis notes the manufacturing cost of the 25nm chips will be approximately $.50 per gigabyte. This is a substantial decrease from the estimated $1.75 per gigabyte cost of 45nm flash."

Good news!

Does the manufacturing cost carry the cost of the R&D that went into developing this process? That cost will get passed onto the consumers as well, as it should.

Edited by ObiWanToby, Feb 1 2010, 4:06am :

ObiWanToby said,

Does the manufacturing cost carry the cost of the R&D that went into developing this process? That cost will get passed onto the consumers as well, as it should.

Assuming they used the same scale to compare the values then that is somewhat irrelevant. If the $1.75 per GB is the price tag covering the R&D that went into 45nm chips, and the $0.50 per GB is the price tag covering the R&D on 25nm chips, than its a substantial price drop. If both prices are without figuring in R&D costs then its still a pretty huge drop unless they jack the prices up extremely high to cover the cost. If one is with R&D cost and the other is not, than its not a good comparison.

Either way, a $1.25 price drop in price/GB is substantial no matter how you look at it.

My current 120GB drive was around $400 when I bought it, going off the current estimate of $1.75 it is worth around $210 right now......based on the prices quoted in the article, the same drive in 25nm tech would be only $60.

Edited by Nagisan, Feb 1 2010, 4:25am :

You're talking about shelf price and the $.50 they're quoting is related to costs. So using their numbers it is a $210 cost to manufacture today and sold for $400 (hell of a mark-up!). Using that same ratio then the drive should now cost $60 to manufacture leaving it at about a $114 shelf price which is still a whole lot better than $400.

Good news that we'll finally be seeing some substantial price drops in the SSD market!

Pixil Eyes said,
"Objective Analysis notes the manufacturing cost of the 25nm chips will be approximately $.50 per gigabyte. This is a substantial decrease from the estimated $1.75 per gigabyte cost of 45nm flash."

Good news!

Yeap... Really good news.