GLG News by Hanni Eid

Implications: 1. SSD is gaining traction, especially MLC based. 2. Apple has shown a lot of confidence by using STEC 3. Will STEC have the bandwidth (mfg) to support? 4. What will be the impact on their stock price?

Analysis: This new design win STEC has achieved to support the Mac Book Air with SSD's is a huge win for them. This will propel their technology (and that of others in the same field) to a new level. I was quite surprised to hear that Apple put their eggs in the STEC basket (especially if it is true they are moving away from Samsung) considering the relatively small size and manufacturing capacity of STEC. Clearly, Apple saw benefits in their controller and overall SSD technology.

This is not a rumor, this is a fact.

Implications: NAND SSD's will replace traditional rotating HDD's in various environments due to several performance factors. Hard drive manufacturers such as Seagate are recognizing this and reacting accordingly.

Analysis: Recent comments by Seagate CEO Bill Watkins indicate Seagate is beginning  a transition, at least partial, to NAND based SSD's. Seagate seems to be interested in "enhancing" the technology utilizing their years of hard drive design/manufacturing experience.

Most HDD manufacturers are realizing that they will need to shift strategies to survive as the price of NAND continues to drop and more advances are made in technology required for SSD (controllers, ECC, wear leveling, etc). The first thing these companies should be doing is adding a solid state drive product line to their offerings, then observing the market and adoption rates. NAND pricing will come down, and SSD will become more attractive.

Implications: Notebook computers will not be the only systems utilizing Solid State Drives. SSD's have inherent advantages over traditional rotating HDD, well suited to the server/data center environment (e.g. lower power consumption, less heat production, no moving parts) Expect SSD's to play a prominent role in servers (blades, etc) as NAND prices continue to erode YoY.

Analysis: The data center environment is similar to an ecosystem. You have temperature requirements, humidity limits, power usage considerations and many other factors that play a role in server health. Of course, heat generation and power consumption are two of the key factors in a dense server populated environment that need to be considered. SSD's fill a perfect niche for server farms by reducing this heat production and also reducing per server power consumption. Some manufacturers of streaming video servers (e,g, video on demand) and avionics entertainment sytems currently use this technology for flash-based content storage and have been for years.

IBM is already offering a Sandisk 32GB SSD as an option on certain blade servers. As prices come down on NAND, these will become more and more attractive to the IT buyers of the world.

The article also brings up an interesting topic of adding SSD's to television sets to perform a Tivo-like function. It will be interesting to see if the TV OEM's can successfully integrate a solid state drive into a television without adding external hardware.

Implications: 1. Currently, NAND pricing is too high for the widespread adoption of SSD's in consumer notebooks. 2. NAND pricing must further drop for SSD's to become more attractive of an option for the PC OEM's. 3. Of course, some people will spend the money, but cost is a huge barrier as evidenced by Sony Vaio with 32GB SSD. 4. Article author only focuses on battery life and price.

Analysis: As mentioned in a previous analysis ( http://news.glgroup.com/cm/Analysis/PostDetail.aspx?pid=14709 ), NAND pricing is still too high for the widespread adoption of solid state drives by both the PC OEM's and consumers. The Sony Vaio discussed in this article contains a 32GB SSD at a premium of $700, well out of the reach of the standard computer user. This article supports the theory that their must be further flash price erosion for SSD's to become widespread.

Also, this particular Vaio received fairly poor battery life reviews in this article, although a SSD with no moving parts would be expected to extend battery life by longer than 30 minutes, as compared to a standard rotating HDD. However, the author of the article has no opinion of recorded increased boot times, increased data seek speeds, and increased ruggedization of the entire system as a result of using an SSD. His article is solely attacking the drive cost and endurance of the battery on one charge.

Implications: 1. 16GB of NAND in an iPod video will be very costly at this time. September will not be extremely different. 2. What effect on WW supply will this have? 3. Will this cannibalize iPhone sales amongst certain users, thus leaving more flash on the market? 4. Is this article simply another overreaction to the Samsung power outage?

Analysis: Very opportune time for publication of this article, immediately after Samsung experienced a hiccup in production and after iPhone sales/activations were not as rosy as expected. Component makers would be very excited if prices and supply were affected by market tightening, as they benefit from this, and are in pain during times of oversupply. This is a good enough reason for them to remain bullish even through uncertain times.  With the current 8GB iPhone commanding a price of $600, to whom exactly does Apple plan on marketing a 16GB iPod to and what will the cost be? If the release is anytime in the near future, expect the price to be extremely high.

The fact that the article contradicts its own thesis in the last paragraph by stating (which I believe to be factual) and conceding that tight supply is expected to ease by mid-September, although I anticipate that happening a bit sooner, really confuses me. Samsung, although responsible for a huge percentage of WW NAND supply, is not the only manufacturer out there.

Bottom line is Apple needs flash component pricing to decrease before the release of their 16GB video iPod, if that is the true capacity they will be releasing. They need this to continue to make their margins and produce a product affordable by the masses.

Implications: 1. HDD manufacturers are finally concerned about losing marketshare to SSD vendors, specifically in mobile markets. 2. More and more 3rd party flash manufacturers are developing SSD's that were not in this market a few years ago. 3. Prices still need to drop before widespread SSD adoption occurs.

Analysis: This article is quite accurate in some of its predictions and notes. The fact that HDD manufacturers are altering their mobile product roadmaps is a very significant positive sign for SSD's. As NAND prices drop, inevitably flash will replace rotating HDD's in most mobile applications due to lower power consumption, less heat production, faster data access times and increased ruggedization, amongst other benefits. The current problem is that SSD's are cost prohibitive for most applications.
Also, Patrick Wilkison of STEC is absolutely correct in his statement about "find the flash vendor"....many, many companies previously with no interest in SSD technology are jumping in head first at this time. I believe the HDD manufacturers should also be worrying about their 3.5" positions within servers in data centers, especially blade servers where real estate and temperature regulation are premiums.

Implications: 1. NAND flash based drives (i.e. SSD) are currently and will continue to become a mainstream replacement for rotating media in various devices (notebooks, embedded computers, networking hardware,etc) 2.NAND pricing must further drop in order for 32GB-128GB and higher SSD's to become cost effective for the end user. 3. Currently, Apple is not helping the situation due to their high demand/usage of NAND for current and upcoming products. It is good for the NAND market/semis but not for the consumer.

Analysis: NAND has been replacing traditional rotating hard drives in industrial applications for several years already, but currently SSD (mainly 2.5") is spreading rapidly. Engineers are excited to work with a non-volatile storage media that can greatly improve system performance (faster access speeds, extended battery life, lower heat production,etc) but currently prices still make the SSD cost prohibitive for many manufacturers.

Currently, flash pricing is on an upward trend, which is good for the chip manufacturers and card houses, but bad for consumers hoping to have a relatively inexpensive 32GB/64GB SSD in their laptop. This may last for a bit of time, thus possibly delaying the wide adoption of high capacity SSDs in the consumer/enterprise market. But, as the semis have proven time and time again, a shortage lead to inevitable supply excesses as more production kicks in. In several months, when parts are in excess supply, SSD's will once again really become attractive to the mainstream markets and not be limited to industrial apps.

Implications: 1. I agree that Apple is currently receiving a large portion of NAND (32Gb and 64Gb) from certain semis, but other cell phone manufacturers are starting to load more NAND into their handsets. 2. 50%+ of new cell phones sold will have a NAND expansion slot (MicroSD typically) for music, video, games, photos, etc. 3. High capacity SSD adoption will be hindered by current increasing flash prices. 3. Apple is not the only player out there in the handset market.

Analysis: It is difficult to argue against the fact that Apple has a large portion of WW NAND production currently allocated to them. This is causing other, smaller flash card/drive manufacturers to suffer at the hands of the semis. There have been verifiable rumors of supplier (i.e. semis) de-committing on orders to smaller customers, resulting in long lead times and increased backlog/canceled orders for some of these smaller, sub-$500M players in the market. I have heard verifiable rumors of certain niche SSD manufacturers increasing lead times on new orders to 12-18 weeks, more than an entire quarter.

I believe when Apple launches a flash-based video iPod, as speculated, they will demand even more of WW NAND production,, which will trickle down to the smaller card houses, but we also cannot discount the market influence other handset manufacturers. Most of these manufacturers are either embedding NAND into the phone, or offering an expansion card slot for MicroSD. Some are doing both. My Blackberry 8800 contains both onboard NAND and an expansion slot, which I immediately filled with a 1GB MicroSD. This will definitely cause an even greater tightening of certain NAND over the next several months.  

Implications: 1. DDR3 platforms will not be widely adopted in 2007. 2. DDR3 demand will increase in 2008 when server/notebook platforms become available 3. There is no reason to get excited about DDR3 making a large impact on the market this year, it will not

Analysis: DDR3 is a new DRAM memory technology appearing as a natural step in the evolution of DRAM memory. DDR3 uses less power, produces less heat and operates at faster bandwidths than its predecessors. However, currently, DDR3 is at an extremely high price premium compared to DDR2/DDR1 and not many system builders have made significant platform (mobo) purchases.

The cost of the memory is currently a very high percentage of total BOM (Bill of Materials) cost for a DDR3 based computer, so only enthusiasts and memory-intensive application users will actually spend the money to utilize this memory this year.

As ECC REG DDR3 comes to market in 2008/2009, the adoption rate will dramatically increase as the DDR3-based server market will create demand. Also, as we see with any new memory technology, prices will gradually but continually trend down after widespread introduction occurs and more memory manufacturers announce and release their DDR3 modules.

Implications: 1. Simpletech has been "bogged down" by the commercial division of the company for several years. 2. The OEM/Industrial division (now known as STEC) has historically had healthy margins and niche products. 3. HOWEVER, STEC now faces competition from unlikely sources such as Super Talent, PQI, ATP, PNY and the return of Sandisk/M-Systems. This will drive down their margins and marketshare.

Analysis: Let's face it, the commercial division of Simpletech was never able to compete with the big 3rd party players in the memory industry. They did not have the buying power (DRAM and NAND) to service the highly competitive markets they were targeting (system builder, white box, reseller, etc). They had no niche and nothing to differentiate them from the rest of the market, except for their exceptional line of external HDD storage.

On the flip side of the coin, the OEM division of Simpletech (now STEC) has always had a niche and has a long standing business model comprised of servicing major OEM's and military customers with Industrial grade flash solutions. The market may not be as large as the commercial flash market, but the margins are worth the extra engineering effort and support required.

However, as more and more 3rd party players start to enter the SSD market and the Industrial Grade flash market, the margins of STEC will be driven down as will their market share. Sandisk will undoubtedly once again rise to be a major thorn in their side, since they have decided to reenter the Industrial-ish market since their acquisition of M-Systems. Simpletech picked up a lot of business when Sandisk exited the Industrial/SSD market a couple years ago, but will the opposite occur when all of these companies enter (and some reenter) the market space?

Implications: 1. DRAM/NAND chipmakers do not have a formula to determine how many components they should be producing at any given time, thus the market is often in shortage/oversupply situations due to production mix issues. 2. Continued expansion in the sector (more fabs, more lines, etc) will continue to negatively impact ASP (Average Selling Price) and cause further market saturation.

Analysis: Demand typically does not eclipse supply in the DRAM/NAND sector unless there are mix issues occurring at the semiconductor mfgr's...which happen on a regular basis. These companies (e.g. Micron, Hynix, Elpida) can never figure out the exact, perfect mixture of NAND, DDR1 and DDR2 components that they should be manufacturing due to the unsteady demand in the marketplace. This is not fully their fault due to the difficulty in predicting the behavior of consumers. Thus, we are too often put into oversupply situations on one product, and left with a shortage on another. The semis can switch their lines to produce DRAM or NAND chips, but the art of it is deciding when to switch up the lines and for how long. For example, the current glut of DDR2 components in the market is a direct of result of overexpansion of capacity and the continuing inability to control production.

Every corporation wants to expand, and semiconductor co's are no different. Unfortunately, since we are in a commodity market, when these companies expand it not only produces more choice for the consumer but it also negatively impacts the market by further increasing an oversupply situation. The expansion problem has no solution in sight.

Implications: Cisco, Nortel, Juniper, F5, etc. have been using SSD technology for several years already in many of their enterprise products. The only difference now is that prices are much, much more favorable to the customer.

Analysis: Many major networking hardware/ruggedized computer manufacturers/avionics-think flight data recorder/military customers have been using SSD's in their designs for several years. Now, much more attention is being focused on this non-phenomenon due to the dramatic reductions in NAND flash pricing. As NAND prices continue to drop, the fact of the matter is more and more design engineers will begin to choose SSD over HDD. When it fits in their budget, they will do it. This is simply a reality..at this time the perfect storm of SSD adoption will only occur when flash prices drop further.

Implications: - NAND Flash will be in oversupply in Q1 07, based on some manufacturer forecasts

- Prices will continue to erode

Analysis: Flash prices will continue to drop as supply will be overwhelming and lack of a "killer app" to use up all these newly produced flash chips is still affecting usage. Soon to be released in the US mobile phones are supposed to be the next big driver for flash card sales, but the majority of these phones have yet to hit the US market.

Implications: - Flash drives utilize much less power than traditional rotating hard drives.

- Currently, flash CPM (cost per MB) is prohibitive for most consumer laptop applications.

- Anticipated oversupply of Flash in early 2007 should help speed the adoption of this technology by forcing costs down.

Analysis: - Anticipated oversupply of Flash in early 2007 should help speed the adoption of this technology by forcing costs down.

    As evidenced by the heavy over supply situation of NAND flash in the first quarter of 2006 and the resulting approximate 30% drop in flash card ASP's, this type of situation results in higher MB sales of flash products. This is due in part to the low cost for high capacity flash cards in the consumer market due to lack of a current "killer application" to utilize all the "extra" NAND components floating around.

As expected, there should be NAND oversupply in early 2007. This sets the stage perfectly for the next stage in flash evolution: the wide adoption of the 2.5" flash drive as a drop in replacement for hard drives in notebook computers. This is a natural progression in technology as flash drives have much better power usage characteristics, ruggedizing to the point of no risk of "crashing", and extremely fast access speeds. Imagine hitting the "on" button on your notebook and it boots up just about as fast as your television takes to turn on. Its coming.

Implications: - Is the worldwide demand for Industrial Grade flash worthy of increased R&D investment by flash manufacturers?

- Why does the article fail to mention read/write endurance, a critical aspect of an Industrial Grade flash card?

- Can Standard Grade flash cards serve the same purpose in most flash-using applications?

- What caused Sandisk to exit the Industrial Flash market?

Analysis:     What is the true world wide requirement for Industrial Grade flash memory cards (ATA, CF, SD, 2.5"/3.5" Flash Drive,etc)? By general consensus, to be "Industrial Grade" a flash card should have, first of all, an operating temperature range of -40 to +85 C...At the top end of the range that is 185 degrees Fahrenheit. That is a very wide temperature range, and excepting field applications in harsh environments (think data collection in the Arctic or Sahara), it is not widely required.

    The next main attribute of a true Industrial Grade flash card is the number of read/write cycles, or endurance. This should be in the 1M+ R/W cycle and above range. These endurance cycles are achieved through methods such as Advanced ECC and wear leveling algorithms to move the data around as needed. Now, to require this level of R/W endurance, an application must be doing very, very frequent reads and writes both from and to the flash card.

My next question is why did Sandisk exit the Industrial Grade flash memory market? Although support costs are higher due to engineering requirements, margins are also higher for these product lines. Did Sandisk see more value in supporting the retail market with its mainly retail-centric product offerings? I cannot answer this, but it leads me to believe that the Industrial grade flash market is shrinking, and a few niche players will reap the benefits.

Implications: - Microsoft recommends 512MB to 1GB of memory for systems running Vista.

- Hardware manufacturers such as Dell recommend 2GB of memory for systems running Vista.

- Shipping PC's with 2GB of RAM will increase system cost quite significantly.

- Will memory manufacturers reap the benefits and will memory go into shortage as a result?

Analysis: With the advent and release of any new or upgraded operating system, comes the inevitable requirement for ever increasing amounts of system memory. However, there has been confusing and conflicting messages from Microsoft and PC OEM's regarding the recommended amount of DRAM to run Vista optimally. However, the PC OEM's will be increasing their system cost dramatically by increasing shipping memory to 2GB, and this will be passed on the consumer. PC OEM's will surely work every angle possible to keep their memory cost down.

DRAM manufacturers traditionally reap the benefits of new OS version releases, as new options integrated into the OS always require larger amounts of memory to run them at an acceptable and optimal speed. However, with the expected glut of memory (DRAM) chips in 1H 2007, the market should be just fine to withstand a sudden increase in demand from new Vista users without extreme price fluctuations or shortages. Prices may increase slightly, depending on adoption rates, but most likely will remain quite stable.

Implications: - DRAM spot market pricing will be down at least 5-9% in November.

- Supply has normalized, except for certain US based manufacturers.

- Potential oversupply looming based on some Taiwanese forecasts

- NAND oversupply/price erosion may be a good thing

Analysis: Following a 20-30 increase in DRAM module spot market pricing since August 2006, the price trend is currently down and experiencing a well deserved correction. Prices jumped amidst tight supply of 64x8 and 32x8 DDR 1 and DDR2 components. Parts now appear to be being released by the semis, resulting in a lifting on the tight supply situation. Anticipate this price downtrend to continue for at least the next 30 days, barring any unforeseen circumstances.

Based on the forecasts of some Taiwanese DRAM manufacturers, supply may outweigh demand in the first quarter of 2007, resulting in a potential oversupply situation with the almost certain ensuing drop in memory module ASP's. This will result in higher GB per system in both servers and desktops, and may be beneficial to the Vista launch, considering at least 1GB of system DRAM is recommended for running of the OS.

Even in the NAND industry, at least an oversupply situation would help the industry move towards ever higher flash device capacities for new applications (e.g. 2.5" flash drive) since Cost per MB (CPM) will drop dramatically. Consumers will benefit from the drop in prices while the manufacturers will be awaiting the next killer flash app that will act as the next driver for flash usage.