SanDisk in license talks with Toshiba; adds NEC as supplier

Electronic News, July 14, 1997 by Jim DeTar

"PCMCIA tried for a long time to develop an alternative interface, like the PC card bus, 32-bit bus, and it has really had very little success in that. So, if there's any direction, it's to simplify and reduce the cost."

EN: We've got NOR architecture, we've got NAND and Mitsubishi and Hitachi with DINOR; and are we going to see that type of flash architecture continue to diversify? Or is it something that's not a concern to SanDisk because you're able to work across all the levels?

Dr. Harari: "We studied a great deal on the alternative architectures, the DINOR, the NAND, the NOR; they are different architectures that are best suited, more optimized for specific applications. I mean, I'm a device guy, so I feel pretty comfortable with any one of these architectures. You can make things work, you can skin the cat more than one way. We do so with our technology, which is really just absolutely driven to achieve the lowest cost possible. We don't worry to much about questions such as: can we use our technology for embedded flash, and can we use it for socket flash, can we use it also for what we want to do. We say to achieve the lowest cost, that's where the biggest market potential is. And we think we will optimize it to the greatest degree.

"Our technology is a triple-poly (three-layer) process, which to the best of my knowledge, no one in the industry uses because it's perceived to be a difficult technology. Triple poly, on the other hand, has the advantage of extremely high density, very, very small cell size.

"If you recall, the DRAM industry and the SRAM industry used to be--in 1980 or so--used to be double poly and the United States players, Intel and Mostek and so on, refused to take the leap to the triple-poly DRAM and triple-poly SRAM. The Japanese did, and the U.S. DRAM and SRAM suppliers basically fell by the wayside.

"The reason is because triple poly gives you one more degree of freedom to shape the cell. That's what we have with our technology. We think by adding a third level of poly, we have a very small and very scaleable cell. We think that our technology is more scaleable than DINOR, more scaleable than NAND and of course more scaleable than NOR, and also is more amenable to multiple density, double density. So, we think it's best optimized for what we're trying to achieve. It's, however, not the most optimum technology for what, you know, if you were trying to build a four-megabit standard flash. I would not use our technology."

EN: On the card form factor, do you think there will be a fourth competitor to Toshiba, Intel and Sandisk's architectures? Can anybody else come up with a different one, or is the market established in terms of the card form factors?

Dr. Harari: "Or beyond that. I think that there are different applications and different markets that require different kinds of form factors and different kind of interfaces. So it's more a question of co-existence than a question of competition or replacement. We don't see CompactFlash as displacing PCMCIA. There are applications where PCMCIA is the best solution application or CompactFlash is the best solution. There are applications where CompactFlash is too large, and Miniature Card is too large."