Flash Memory Stocks: Why I'm Holding M-Systems and Not Sandisk or Saifun (FLSH, SNDK, SFUN)

Includes: FLSH, SFUN, WDC
by: Shlomi Cohen

Two weeks ago, I wrote that I would explain why I decided to keep M-Systems Flash Disk Pioneers (Nasdaq: FLSH) in my portfolio, but not the other two flash companies that I had, Saifun Semiconductors Ltd. (Nasdaq:SFUN), and Sandisk (Nasdaq: SNDK). I will now try to explain as promised, despite several dramatic events that I never dreamed would happen, although as I will explain later on, my decision has remained unchanged.

As is known, about two weeks ago M-Systems abruptly announced that it had initiated an inquiry into its options allocation policy, and also said it was cancelling its planned massive $340 million share issue, which was meant primarily to finance supply agreements based on its new x4 technology.

Today, the stock is 16% below its price of two weeks ago when I took the other two flash stocks out of my portfolio. During this period, Saifun also suddenly lost 7%, while Sandisk dropped 12%, as a result of the general falls on the market. Sandisk lost ground despite having its rating upgraded to “Buy” by Bear Stearns.

The three flash companies, under the leadership of Israelis Dr. Eli Harari, Dr. Boaz Eitan, and Dov Moran, are aiming for the huge flash chip market for data storage [NAND] in mobile handsets, music/video players, digital cameras, including flash-based video recorders, laptops, USB drives, and many other applications. This market was worth $12 billion in 2005, and it is expected to triple by 2010.

Of the three companies, there is no doubt that Sandisk is the leader, with a market cap that has fallen from the January high of $15 billion to the low of $10 billion today. Investors have given Sandisk this market cap because it has managed to earn $300-400 million annually in royalties on its flash-based inventions, in addition to having annual sales in excess of $3 billion.

If M-Systems and/or Saifun's inventions for increased storage capacity become a commercial success in years to come, it will almost certainly be at the expense of the phenomenal success of Sandisk’s developments. If this does come to pass, Sandisk’s current market cap could be at risk of collapsing by tens of percentage points.

M-Systems reached its current market cap range of $1.1-1.5 billion thanks to a 20-fold increase in sales over the last five years, while completing a dramatic transition to substantial earnings per share. Like a skilled surfer who expoits every wave, M-Systems has so far been successful with every product it has developed and launched, although it has virtually no income from royalties (which go straight into the net profit line), as Sandisk does.

Investors have quite rightly given M-Systems a sales multiple that is only a quarter of the one they have give Sandisk, because of the former’s low gross profit and the lack of income from royalties. All this, however, could become history if the 'x4' product will really do what they claim it will do. For the present, x4 is priced at zero, if not negatively, because of the options issue.

As opposed to the other two companies, Saifun is a company that only sells IP in return for royalties from sales generated by customers with whom it sells licensing agreements. Saifun had quite a struggle during its first years, and failed to achieve significant success in the less attractive fields of flash. This cost Saifun’s partner, Tower Semiconductor (Nasdaq: TSEM) the painful loss of the dream of making the big leagues together with Saifun as the manufacturer leveraging its inventions.

Only in recent years did Saifun realize that the area of phenomenal growth was to be found in NAND, an area that Sandisk and M-Systems rightly focused on 10 years ago. Virtually nothing remains of the partnership between Saifun and Tower, who are now engaged in a squabble over pennies, while in the meantime, the dream of production lines that would churn out flash chips based on Saifun’s inventions has moved on, like everything else, to China.

The SMIC foundries set up in China in 2000 with government support are expected to begin manufacturing memory chips based on Saifun’s NROM technology by the end of the year. In addition, Spansion Inc (SPSN), the US manufacturer spun out of Advanced Micro Devices (NYSE:AMD), has big plans to take over market segments with its MirrorBit technology, which is also based on Saifun’s NROM technology.

Though it's considered a company that produces patents wholesale just like Sandisk, Saifun has so far failed to earn substantial sums from the licensing of its first invention for doubling storage capacity. Investors have based the bulk of its curent value, $1 billion fully diluted, on extensive implementation of its latest invention, Quad NROM, for massive information storage.

It should be understood that Sandisk’s MLC invention for doubling storage capacity has become a runaway success that has now been translated into multi-billion dollar chip manufacturing operations at plants run by Samsung and Toshiba. On the other hand, sales of chips based on Saifun's invention amounted to less than $100 million last year.

Exactly a month ago, M-Systems announced that it too had joined the playing field of inventions that bring about dramatic change in the basic cost structure of flash manufacturing. Sandisk has MLC, Saifun has NROM, and now M-Systems has x4.

M-Systems’ options review diverted attention from its dramatic announcement, though it should be understood that if the company can deliver on the promises made in the announcement, such as the claim that a substantial segment of the NAND market will eventually be based on x4, this is a potential breakthrough that could take the company’s market cap up to several billion dollars within a few years.

Most of the analysts in the field cover all three companies, and I found that none of them is willing to give investors their estimates of which company will succeed and which will fall by the wayside. It seems to me that these three technologies cannot possibly survive and coexist side by side.

Sandisk managers said at presentations last month that they were not worried by Saifun or M-Systems, since they felt the implementation of either technology on production lines would be difficult and complex, and that therefore they did not see any threat to MLC’s success. They added that Sandisk was working on improvements to the MLC technology, from double to triple storage capacity on one single memory cell.

I went over to Saifun to hear what they had to say about the new player on their field, M-Systems, which claims it has also increased capacity to four bits per cell. I came away in a state of shock. “There’s no such thing, it’s impossible,” was the reply I received during a 90 minute meeting with president Kobi Rozengarten, and after spending another quarter of an hour with founder and CEO Dr. Boaz Eitan (pictured).

Perhaps there might still be something in it, I insisted, since they announced that they had “working samples.” The answer from both managers was an emphatic no. “There is no way that they could do what we’re doing, using their methods.” They then went into technological details such as the “16 level barrier”, as if it were some form of sound barrier that I, a lowly infantryman, couldn't handle alongside decorated pilot Boaz Eitan.

At Saifun, they summed up their great potential for me in one word - Samsung, the world’s largest manufacturer of NAND chips. One of Sandisk’s best and most profitable agreements with Samsung will expire in 2008. They claim at Saifun that Samsung announced last year that current flash technologies such as MLC will no longer be suitable for its future production lines (due to be set up in 2009), which would have a geometry of less than 50 nano. Saifun’s managers are confident that they will then gain a strong foothold in Samsung, because NROM will become the only possible solution.

I went over to see my sources at M-Systems and told them I was confused. How is it that Saifun won’t give their inventions the time of day? M-Systems has, after all, come up with one or two cutting edge inventions in recent years, and has had a few sales here and there which will reach the $1 billion mark this year. “At Saifun, they’re silicon people, while we’re systems people” which is why they don’t get the picture, they explained politely. They reassured me that what they can’t see in Netanya, they can see clearly in Kfar Saba and in a few more locations across the globe, about which we will learn sooner or later.

To sum up, I am sticking with the M-Systems’ potential, because over the last six years since it raised itself up from the devastation of the bubble in 2000, it has delivered on all its promises to investors with the precision of a Swiss watch, with every single promise becoming a runaway success bought by companies worldwide within the space of a year to eighteen months. I therefore find no reason to believe that the implementation of x4 will be any different.

As for Saifun, it may well be that its technology has not really made the big breakthrough yet, despite the many years that have passed since it was first launched. Perhaps they are now finally on the right track and will enjoy mega success, provided they can get Samsung to trade in Sandisk’s MLC technology for their own.

For the time being, I prefer to observe the Saifun saga from the sidelines, on the assumption that I will have more than one opportunity in the future to reconsider my decision to leave it out of my portfolio for the time being.

The same applies to Sandisk, since as I mentioned earlier, there's a question mark about its ability to continue to rake in hundreds of millions of dollars a year in royalties. Furthermore, as opposed to the other two, Sandisk has spent billions of dollars in a joint investment with Toshiba in manufacturing plants, and it could be hit the hardest of all three companies by any surplus supply in the flash market.

Published by Globes [online], Israel business news - www.globes.co.il - on June 13, 2006
© Copyright of Globes Publisher Itonut (1983) Ltd. 2006