Saifun – Is it the little flash company that could?
One market segment that could see even stronger growth than these separate products we mentioned, and include other growth products, is the flash memory market. Flash is a root component used in all the above products and more.
Based on history we are forecasting that flash is the memory medium of choice for a plethora of devices in the consumer electronics in wireless devices and that flash will grow faster than the wireless devise market. It appears that in the past, memory for computing devices has grown faster than the device that utilizes the memory. Memory of the Personal Computer (PC) and the Internet has grown faster than their supporting platform. With the PC creating tremendous growth and history as our guide the demand for both memory and disc drives for the personal computer was often the impetus of many upgrade cycles. The Internet with the many millions of new web pages created a tremendous growth in storage. I've seen in many reports that forecasted storage of the internet has been one of the fastest growing subsets of the internet as a whole.
With a decrease in price per gigabyte (GB) of more than 80 percent over the past three years and with the high growth in wireless data the need for new and addition memory could exceed the growth of the hardware device market that uses flash for its memory. The current market in flash memory is about $25 billion annually and its forecast is about 40 billion by 2010.
With each new product cycle the advantages of flash have become more disruptive allowing it to become about 30-40% cheaper every year. Many experts are forecasting this disruptive curve to replace the disc drive market for PC's. Flash has already replaced hard drives in most MP3 players.
Currently the flash memory is designed to support two types of flash memory. One type of memory supports your machines internal usage or operating system, the other type is for more external storage needs. The internal memory often uses the architecture of NOR, which has been established for years and Intel (INTC) considered by many as the market leader. The NOR technology is a more complex technology and is starting to see the market mature.
Often you will find both NOR and NAND in the same mobile device.
The much faster growing market is for external memory market needs or NAND and the one of the leaders is SanDisk. SanDisk Corp. (SNDK), founded and managed by president and CEO Dr. Eli Harari. SanDisk and Toshiba jointly launched the multi-level cell (MLC). This technology made it possible to divide the cell and store two bits of data on the same piece of silicon (x2, as it were), which significantly improved the profitability of manufacturers and fabs, basically doubling the price performance curve.
This process has become the leader and allowed NAND MLC to become disruptive to the predecessor NOR architecture and in 18 months penetration has been so great that MLC is becoming dominate force in flash.
We believe that this new curve of double captivity on a single cell technology will become the single most important factor for next generation flash memory, and it will become essential as flash is staring to see possible limits in the reduction of its die size as many experts are starting their forecasting. If flash is going to continue on its curve of lowering the price of a gigabyte by 80% over the next three years, it is my opinion they will need an architecture that's designed specifically to establish this goal. There is a proprietary NROM architecture that has many advantages toward increasing capacity of bits per cell. The NROM is close to production of 4 bits of memory in each cell or quad flash.
The company we believe has a unique position and leads the NROM approach in the flash memory market is an Israeli based company called Saifun (SFUN).
Saifun is an intellectual properties company which its revenues come in three forms: licenses, royalties and support. This type of model has been very successes for our model portfolios in the past. The three previous companies that had core business from intellectual property we investment into our portfolio's were Qualcomm (QCOM) in1997 at 3.31 per share and still holds a position. Arm Holdings (NASDAQ:ARMHY) in 9/29/1999 @ 9.60 and holds half a position and Rambus (RMBS) in 1998 which appreciated about 350% in 2000 and we sold the position in the model portfolio when Intel stopped supporting the Rambus architecture late and 2000 and in 2001.
Even though it is very early is Saifun publicly traded history we are excited by its new form of flash memory architecture, it appears that Saifun's approach has many advantages over the more established NAND and especially NOR. The single most important part is their technology curve. They have the ability to double the bits per cell allowing for a second compounding curve. The other architecture they are working hard on is to shrink their size and increase density, but we believe that Saifun with its simpler model should achieve a smaller die than the others but the real advantages with Saifun is the ability to allow 4 bits of memory in every piece of silicon (x4). Doubling again the events of MLC while at the same time reducing their size thus possibly leading the new flash architecture. Another advantage is NROM's ability to work both as an operating system and memory component being able to supply both markets that individually NOR and/or NAND has target.
A second company has just announced that in 2007 they will start producing a 4 bit cell in NAND. The company making this announcement is M-Systems (FLSH). They claim they will have a product on the market some time in 2007. Even though they have achieved this tremendous breakthrough we believe that because they use the whole cell instead of a fraction of the cell for this doubling process, the whole cell's ability to double again may become geometrically tougher. On the last review M-Systems has not explained their business model to (make at own fabs or licenses) and delayed the secondary offering.
It is has been our opinion that companies that form successful royalty models resemble gutters and the fab companies have the appearance like shingles when looking at a roof. When it rains the gutter can create a stronger stream receiving income and achieve a much higher level of profitability. The delay of M-Systems secondary offing might reduce the chance of more fab developments.
Either way this looks like a marathon race and since this is such a very large market it will be about a $40 billion market when quad flash is widely available, that means that any of the top three or four should benefit.
Saifun already competes extremely well with NOR but early 2007 when it doubles the number of bits from 2 bits to 4 per cell it should be able to show advantages over MCL NAND currently the price performance leader. Saifun has a chance of repeating the same step that, in our opinion, allowed SanDisk to lead the last cycle.
There are many new technologies looking to replace flash but at this point there are a few that are close to achieving mainstream volumes. You should know the Saifun technology hibernated for about twenty years. This is very common, the Internet incubated for about 30 years and electricity for 100 years. New technologies often hibernate longer than people anticipate, and then it seems that they often almost explode onto the seen very quickly.
Even though Saifun's approach is about 20 years old, the technology they have just started to achieve is commercial feasibility.
The true advantage is since they only use points in the cell versus in the more convention approach such as NOR or NAND that uses the whole cell. This simpler usage allows for higher data retention and also provides a faster response time, and hopefully more density, and less power.
This is a tremendous advantage having 4 times the bits in competitive cells. Saifun also believe future that future cells could expand to possible to 8 or even 16 bits per silicon.
Saifun only has a handful of clients, if they loose Infineon Technologies (IFX) Saifun largest client, they would impact their business tremendously. On a side note, it looks like it will pick up UMC out of Taiwan.
Saifun has basically signed many very large vendors like Sony (SNE) and Spansion (SPSN) a spin off Advanced Micro Devices (AMD) / Fujitsu (pink sheets) these based solely on the flash market are small in the market, since the production volume is small this could make it harder to be designed into leading volume products.
Even though we believe NROM offers a simpler cell structure with several layers, we believe it will be easy over time to reduce or migrate to a smaller form factor, but this has not been completed in high volume production. If and/or until they can compete in a smaller form factor this company will be, based on unit size, be at a significant disadvantage. Experts believe in 2007 this disadvantage should be at most minimal and Saifun believes in late cycles this will be come a true advantage.
1) If Saifun continues to lead the flash market with more bits per cell with NROM flash architecture.
2) If Saifun if achieves the forecasting of smaller die than comparable flash.
If Saifun achieve either of these goals it could become an architecture leader in the flash memory market. If they are able to achieve both they would attain a real architecture leadership position.
According to several of our monopoly theories, available at www.durig.com the stock market value of the companies that lead architecture often grow faster than all the combined companies stock market values that utilize the architecture.
Thus, if Saifun become the dominant architecture with the smallest die size in my opinion it will probably attain the leading stock market value in the flash memory market.