Spansion, which is owned 60% by AMD and 40% by Fujitsu Ltd, filed for an initial public offering in April. Spansion moved closer to completing the deal on Friday, when it filed papers with the US Securities and Exchange Commission outlining the terms of the offering.
The business unit, which makes flash memory used in mobile phones, expects to net $561m, after underwriting fees, by selling 35.3 million shares priced between $16 and $18. The Sunnyvale, California-based also plans to offer eight million shares of convertible preferred stock at $25 each. Separately, Spansion said it would offer $400 million worth of unsecured notes in a private placement.
On completion of the deal, AMD would own 40% of the outstanding Spansion shares.
AMD has been under pressure from Wall Street to ditch money-losing venture so it can focus on selling server and computer chips, where lately it has been gaining market share against rival Intel, particularly with PC microprocessors. The Spansion IPO could also be a major boost for AMD stock during the next couple of years.
Of course, this very flash memory business once helped keep AMD afloat. AMD would probably not exist today if it didn’t have that healthy flash business in the late 90s’ and early 2000’s, said iSuppli senior analyst Mark DeVoss.
But Spansion has weighed on AMD’s profit recently, as falling prices and capacity gluts plague the worldwide NOR flash memory market.
For the year to date, Spansion has had $250m in operating losses. However, its revenue has had a minor revival during the past two quarters and AMD expect the unit to pocket in the fourth quarter more than the $516m it achieved in the most recent September quarter.
Content from our partners
The type of flash memory that Spansion makes, NOR (Not And) was outpaced last year by another type of flash, called NAND — a trend that is likely to continue. Independent researcher iSuppli projects that NAND sales will grow 13% annually until 2009, versus a 4% rate for NOR.
Because to date we have sold only NOR-based Flash memory products, this trend could materially adversely affect us if we are unsuccessful in executing our strategy, said Spansion in its SEC filings.
NOR is optimized for code storage and guarantees not have any data integrity issues for about 20 years, which makes it suitable to support applications on mobile devices, such as the power-up feature on a smart phone, for example. After all, if NOR were comprised in this example, the mobile phone user potentially wouldn’t be able to turn on the device.
NAND, however, is optimized for data storage and is shipped with known bad bits, in which a one may turn into a zero, for example. Over time those bad bits will compromise what it’s storing, DeVoss said. But NAND has a far greater storage capacity than NOR and a bad bit here and there isn’t noticeable in large files, such as MP3s.
NAND’s greater storage capacity is becoming increasingly attractive to mobile device markers. The largest NOR-density devices available are 1 gigabit, versus a 16-gigabit NAND flash from Samsung, DeVoss said.
Increasingly, OEMs are using a controller with NAND, which monitors the status of bits in a device. So now NAND is also beginning to be used for code storage, DeVoss said. But it’s a more complicated design [than NOR], obviously, he said.
NAND also is much cheaper than NOR. Probably if it costs a dime for NOR flash megabyte then it costs a penny for NAND flash, even a sub-penny, DeVoss said.
But it would be cost prohibitive to embed NAND flash into a mobile phone, he said. High-density NAND flash probably exceeds the budget for the phone.
For NAND makers, however, mobile phones are their fifth or sixth largest market behind various multimedia application markets, such as USB cards, for example. The cell phone market is, by far, the sweet spot for NOR.
At least for the time being, NOR is likely to remain the embedded memory supporting key mobile phone applications. And the advent of smart phones bodes well for NOR makers such as Spansion, because more applications on the device means more code, and NOR is best suited to support that code, DeVoss.
Depending on how smart phones are architectured, DeVoss reckons NAND and NOR can co-exist.
Still, NAND does a pose a threat, especially because it can be used in NOR-type applications using a controller.
And as mobile phones become more advanced, they require higher density Flash to meet increased data storage requirements, said Spansion in its SEC. Because storage requirements will increase to accommodate data-intensive applications, OEMs may increasingly choose NAND-based Flash memory products over NOR-based Flash memory products for their applications, read Spansion’s SEC filing.
Analysts agree the growth outlook for NOR is limited.
Still, things have been looking up for Spansion lately. Price wars in the NOR market have subsided somewhat, DeVoss said.
Intel also is Spansion’s greatest rival in the NOR market, having wrested the top spot from Spansion with lower-priced product last year. But Intel continues to struggle with tight capacity, which can only be to Spansion’s advantage.
Spansion may also have, in the long-term, an upper hand with its version of double-density flash. That is, technology that doubles the storage capacity of a single Flash cell. Spansion’s version, called MirrorBit, promises to realize 35% cost savings in manufacturing efficiencies, DeVoss said. Unlike Intel’s version, called StrataFlash, MirrorBit involves a new type of manufacturing.
But, while Intel is on its fifth-generation StrataFlash, Spansion just this year released its second-gen MirrorBit product, so it likely will take some time before the bugs are ironed out and Spansion recognizes those savings.
No date for Spansion’s IPO has been set. The company hopes to list its stock on the Nasdaq under the symbol SPSN.
