Flash memory is standard fare in digital cameras and MP3 players. But the chips could soon become even more pervasive, spreading to a whole new crop of electronic devices as the next generation of flash technology takes root. Two companies are racing to bring competing versions of 4-bit flash memory to market. M-Systems ( FLSH) and Saifun ( SFUN), both based in Israel, say their technology will be available next year, and they project rapid industry adoption. The technology promises to bolster the fortunes of flash memory suppliers, whose double-digit sales growth rates have made it the star of the semiconductor industry. But the debut of 4-bit flash may not be the cakewalk its proponents expect, with everything from performance to manufacturing issues threatening the technology's ability to develop. "I'm going to take a wait-and-see attitude," says Jim Handy, a flash analyst at market researcher Semico Research. "I'm not going to build my forecasts around those chips hitting the market. On the other hand, I do believe that there's a strong need and a very strong motivation around the suppliers to make this technology work as soon as they can." The benefits of 4-bit flash are apparent. Flash memory, which retains data even when the power is switched off, is increasingly being integrated into consumer electronics devices. Today's flash memory chip consists of numerous cells, which can store either a single bit of data per cell (single-layer cell) or two bits of data per cell (multilayer cell). Because 4-bit technology retains twice as much data on each cell as does 2-bit technology, the result is a flash chip that can store twice the amount of digital music, photos, etc. Four-bit flash also promises to significantly reduce the cost per bit of producing chips, enabling companies to sell flash memory at a discount. M-Systems contends that manufacturers that license its 4-bit flash technology, called x4, can reduce their production costs by 30%.
Combine higher storage capacity with lower prices, and it becomes viable to put flash memory into all sorts of new places. Digital camcorders, for instance, could use flash memory rather than tape to store video footage, and personal computers could be equipped with flash chips instead of today's mechanical hard drives. There is even speculation that record labels might one day distribute albums on flash chips rather than on CDs, and that textbook publishers could stop printing on paper. All this is prompting some bullish predictions. M-Systems CEO Dov Moran has declared that the company's x4 technology will make up the majority of the flash market by 2010. M-Systems' plan of conquest involves licensing the technology to other flash makers in exchange for manufacturing capacity at their chip-fabrication facilities -- an arrangement designed to grow the company's revenue and the market share of its proprietary x4 technology. Saifun, which went public last year, focuses exclusively on licensing its technology, a variant of flash memory that it calls NROM. The company has seven licensees, including Infineon ( IFX), Spansion ( SPSN) and Sony ( SNE). The modest $79 million in revenue that Saifun generated in 2005 represented a 156% increase from the year before. If Saifun's so-called quad NROM technology takes off, the company's revenue and its roster of licensees might see a good deal more upside. But Saifun's spotty track record with its current generation of 2-bit technology suggests that the road to 4-bit flash could be bumpy. German chipmaker Quimonda (the recently spun off memory division of Infineon), for example, has been working with Saifun's 2-bit flash technology since 2003. This effort yielded a paltry $167 million in flash sales last year for Infineon, roughly 1.4% of the market, according to industry research firm Gartner.
Compare that with Hynix Semiconductor, which rang up a whopping $1.48 billion in flash sales in 2005, only a year after entering the market using standard NAND flash technology. Gartner analyst Joe Unsworth says the contrast raises some serious questions about how easy it is to manufacture chips in large volume with Saifun's technology -- and about the prospects of Saifun's next-generation 4-bit flash. "If you look at
Quimonda and how they've struggled with 2 bits per cell, I'm a little bit skeptical as to when 4 bits per cell is really going to hit the market," says Unsworth.
Antiestablishment ArchitectureA Saifun representative said in an email message that the company can't comment on the business details of its licensees, such as Quimonda, but writes that it is no simple task for a company to come to market with a technology that competes against the established flash architecture. With Quimonda having announced that it will sell quad NROM chips in the next twelve months however, Saifun says it expects the chipmaker to increase its market share. Moreover, as the flash industry adopts manufacturing processes that produce smaller chip circuitry, Saifun contends that the standard "floating gate" flash architecture will reach limitations. In order to provide 4-bit flash memory, Saifun says flash suppliers will have no choice but to adopt its quad NROM technology. Manufacturing issues could also blunt the impact of M-Systems' competing 4-bit flash technology. To offer a cost advantage over existing 2-bit flash, analysts say that x4 chips need to be produced with the most advanced manufacturing processes. But M-Systems has yet to provide any details about who its manufacturing partner will be and what size circuitry its x4 chips will be produced with. "We are familiar with the competitive environment, and we will target the competitive environment accordingly," says Noam Kedem, M-Systems' marketing VP.
And though M-Systems has flagged video storage and hard drives as two key markets that will convert to flash memory, thanks to the benefits of 4-bit technology, it's unclear how suitable x4 really is for such applications. True, streaming video requires the deep storage capacity provided by 4-bit flash. But video also requires fast performance, something which naturally suffers as flash memory retains more bits of data on each cell. Because M-Systems has been tight-lipped about the specifications of its controller, which plays a key role in determining flash memory speed, no one really knows how x4 will perform. "We know performance is going to take a hit, but how much?" says Gartner's Unsworth. M-Systems counters that x4's performance will be above the minimum requirements for video and hard drives, and that it can tailor its technology to perform optimally for different applications. In any case, even with poor performance, 4-bit flash could still provide a cheap way to cram capacity into mass-market products such as USB thumb drives. Revolutionary innovations are great, but there's always a market for old-fashioned evolution.