SAN JOSE, Calif. - What will happen in the NAND flash market in 2011? Here are seven trends collected from various reports from analyst C.J. Muse of Barclays Capital:
1. Year of NAND
''2011 is likely to be the year of the NAND. While the majority of NAND production is at 3x-nm, bragging rights are being decided for 2x-nm fabrication. NAND benefits from the fact that two vendors control ~75 percent of the market, namely a Samsung (40 percent) and Toshiba/Sandisk (33 percent) duopoly. For 2011, for the first time, capex spending on NAND is likely to exceed capex spending on DRAM.''
2. Apple effect
''We think that 2010 tablet market will have a lot to do with the supply demand balance or lack there-of. Applied Materials Inc. is on record saying that in addition to shrinks, there would need to be a 200 kwspm wafer start fab to support expected demand. Apple's demand by itself is expected to double at the very least in 2011 over 2010 depending on who you talk to, but most likely triple because all the Macbook are now SSD based.''
3. Samsung: Leader, follower
''The move to 2x-nm. Overall, all the vendors are moving to various forms of 2x-nm. Samsung at 27-nm, Micron at 25-nm, Toshiba at 24-nm and Hynix at 26-nm.
Samsung is no doubt a production and profitability leader, but not necessarily a technology forefront leader in NAND. Samsung was a late entrant to the multi layer cell fabrication. Right now, Samsung is able to make NAND chips at both 3x-nm and 2x-nm nodes. Samsung will be at the 27-nm node in 2011, which brings down the cost by 35 percent from its 3x-nm node. Samsung expects that 3x-nm and below geometry at 80 percent by the end of 2010.''
4. Thanks Toshiba
The ''wind (is) at the back of flash due to the power outage at Toshiba in 1Q '11. Temporarily, the power cuts are at it again. This time it is at Toshiba. While the amount affected is small, nevertheless it will help with price stability in January and Febuary 2011 as likely 20 percent of the work in process production was in some way affected by the power outage.''
5. Toshiba tablet?
''Toshiba is doing a strong 32-nm/24-nm ramp and is also releasing its own tablet. Toshiba started 24-nm production in Aug 2010. In addition, it started construction of its Y5 fab in July 2010 and it will be ready by Spring. Important to remember is that unlike Intel’s 450mm D1X fab, this facility is not yet capable of supporting EUV tools.''
6. Resurgence of Hynix
''Typically, Hynix used to suffer from long intractable yield issues, for example at the 6Xnm and 4Xnm nodes. But recently their yield profile has improved drastically and they are closing in on the leaders in both 3x-nm and 2x-nm. Adversity and survival instinct and good management is working wonders.
Hynix is doing a 3Xnm transition followed by a 2X nm transition within a relatively short period of time. It is in M11, and while capacity is likely to go up by 30kwspm in CY10, CY11 will end up supporting a ramp of at least 30,000. However, a lot depends on Hynix’s ability to secure additional capital in light of its weakened balance sheet.''
7. Micron ramps (sans Intel)
''Micron recently guided FY11 capex in the range of $2.4-$2.9 billion, with ~2/3 of the amount on the NAND IM Flash Singapore (IMFS) ramp. Micron plans to ramp ~60 percent of IMFS capacity of a potential 100,000 wpm by year-end CY11. Recently from approximately 50-50 percent (joint venture with Intel), Micron upped its stake in IMFS to 71 percent. IMFS is using 2-bit MLC, probably better than the 3-bit MLC that both Toshiba and Samsung have developed. Micron expects 25-nm cross-over in 1Q11.''
SSD's seems to be too costly to replace the HDD's at this moment. But there is also advantages which will be considered only if the prices comes down. Now a days everyone talks about cost before anything else. But once the price comes down there will a huge surge in the NAND demand.
OneNAND is being used to replace NOR for XiP already. It is NAND with SRAM cache buffer. But it could have a prohibitive cache miss penalty. With major NOR makers already investigating alternatives such as PCM, their sense of future for NOR is questionable.
I disagree. While NAND has already caught up with NOR for bootloading, it doesn't support execute-in-place (XIP) (its architecture inherently doesn't support it). But, I do not know enough about how big a deal XIP is. Maybe someone can comment on that.
"but most likely triple because all the Macbook are now SSD based." Macbook & Macbook Pros still ships with HDD as the default option. To replace a 500GB HDD with 512GB SSD , Apple charges ~1500 US$ in addition to the normal costs. Unless the prices drop to reasonable levels, I dont see Apple Mac business being a driver for the NAND industry. But ofcourse all the iDevices can help triple the NAND usage.
On 5, Please dont count on Toshiba's tablet business. They have recently released an Android tablet "folio 100" and stopped sales because of the high return rates.
I think to rely on analysts for directions in semiconductor industry investment is very dangerous, as there are too few of them to have any peer review to keep them in line. I have noted that Nomura and Barclays have been particularly careless. It probably speaks more that few analysts have considered it safe to focus on semiconductors. With too much focus on cost instead of innovation out of the infrastructure box, the industry has become a larger investment risk than before.
I am looking forward to the increase in SSD usage and the corresponding lowering of the cost. I have seen desktop systems using a combination of SSD and HD storage with increased performance (at a cost premium). Can't wait to see what 2011 will bring!
Apple effect caught my eye and attention. By 2011 all the Macbook will be SSD based, which means other notebook will soon make a transition to the SSD memory due to advantages of lower noise, better speed, less wear-tear etc.
I knew the main hurdle to switch to SSD memory was cost. Quoting trend 3: Samsung: Leader, follower: price at the 27-nm node will be 35% lower, it is still way expensive than hard-drive. How much will be prices will be lowered in 2011?
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.