Reflections on Life and Death in the Memory Sector: Spansion and Qimonda, Long on Technology, Have Too Few Friends in High Place$

Hammered by market events, two significant memory suppliers suffer in Chapter 11 of bankruptcy. For one, Qimonda, it is almost over, as its assets are being liquidated just as the DRAM market shows near-life again. For the other, Spansion, the final chapter is yet to be written, but whatever emerges from its Chapter 11 bankruptcy later in 2009, it will likely be a far smaller and less potent player that it was in 2008 and 2007. Its highly regarded NROM/CTF technology, with numerous and far-reaching advantages over the more traditional Floating Gate flash technology, merits hardly a mention in its current survival and reorganization discussion.

This essay is a (or should we say, 'another'), discussion of important factors which enable some memory makers to survive while some are forced to withdraw, merge or disappear. It also strikes at the general naviete of calling our beloved 'memory segment' a part of 'the technology sector', when, in times like these, technical prowess plays a distant second, or third role is the quest for survival and success. Ultimately, technology is secondary to timely management right-decision-making, to product portfolio and positioning, to investment and mis-investment, to having a banker's prudence, and to having 'friends in high places', who can keep you on life support until a better day. Companies who considered themselves live-or-die-standalone businesses have mostly died; those who considered themselves de facto arms of government or larger business entities, or parts of 'banking groups', are still around...not well, but not dead, either.

Consolidation is nice, but..: Everyone calls for 'industry consolidation', but no one can describe exactly how it happens, and every instance is different from every other one. Even attempts to orchestrate mergers run into resistance. Creditors who have much to gain by tough and realistic decisions to foreclose enterprises, remain passive. Gaining consensus among all interested parties is nearly impossible, so long as anyone has any leverage...and everyone has enough to stall decisiveness and tough choices.

It is rarely a matter of two companies waking up one day and saying, "Today I will merge with so-and-so." Independence dies hard, as we have seen in the nearly-a-year-long shotgun (or BB gun, in this case) consolidation that has so far failed to combine ANY of Taiwan's four DRAM makers (or six, if Inotera and Rexchip are counted...or 10, if one includes Taiwan's technology partners Elpida, Hynix, Micron, and Qimonda). While partnerships have been made and other alliances have been broken, precious little 'consolidation' (and those only piecemeal and ragtag) have happened in ways that improve the industry’s efficiency and the survivability of its players.

The foremost two casualties of the present memory downturn who, so far, appear to have taken the biggest hits, are NOR/NVM maker Spansion, and the residual of Siemens-Infineon's memory group, Qimonda. Both have market and technical roots that are decades old...Spansion was a "Junior Intel" in EPROMs as long ago as the mid-1970s, as they were in Fast SRAMs and MPUs. Siemens, almost always a solitary outpost for EuroDRAMs, put its DRAM stake in the ground about the same time. So, it's been 30+ years for each of them. Each has had ups and down over the long haul and many 'cycles', but neither Qimonda nor its predecessors when it was a part of Siemens or Infineon, was ever the market leader that Spansion was, in NOR flash, until it finally cried "Uncle" for protection from its creditors. Indeed, it was not until the mid-1990s that Infineon showed its independence and original contributions in DRAM technology, until then relying mostly on licensed designs and technologies from Toshiba, IBM and Mitsubishi.

In 2007, before the memory downturn grabbed hold, Spansion was a $2.5B NOR flash player, and top of the game in the wireless space, with more than 35% share of the NOR flash market. Qimonda was the #3 DRAM maker, with revenues of more than $4.1B., with more than 14% of the DRAM market and a strong G-DRAM market position.

What counts, or seems to count, in today's market: Whatever their market standing, however, they have both been remarkable assemblages of technical talent in a world that is steadily moving to one of "Right Management Makes Right", easy access to friendly capital, and little mistakes with big consequences. Taking one's eye off the ball for so much as a few minutes can spell disaster...and in this business, many lethal balls are in the air, in surprising places, at all times. Still, that Powerchip and ProMOS, or even Hynix...each with far less technical talent that Qimonda and Spansion (IMHO)...can out-survive these companies tells us something about what it takes to stay alive in the economic downdraft and onslaught we have faced this round. The "Technology Sector" might rightly be named the "Friendly Banker Sector", as that seems more to determine who lives and who dies.

Investment timing and asset management count for a lot, and preemptive cash flow actions count for a lot. Low cost manufacturing is important, as always. Pricing leverage with key customers, to avoid the intense pricing in the commodity market spaces has counted for a lot this time. Technology counts for something, but is not the be-all and end all of success. At the end of the day, management's appreciation for, and response to, the pending crisis...what they realized, what actions they took, and when they took those actions...has shown itself to be the difference between life and death.

Oh, yes, and sadly, who your friends (with money and patience) were and are, was critical. Those who had the capacity to expose huge risks to their shareholders and creditors, and keep spending when all hope was lost, and then, when those pathways were exhausted, go back to their bankers and governments, have consistently dominated the more timid and business-prudent competition. Not that what they ended up with at the other end was 'good'...with its persistent low margins, large-and-larger capital calls for new fabs and new technology, and providing "long-term care" for businesses that they themselves created, but then were forced to suffer with 'until a sunnier day'.

Near Death today...Death Tomorrow: Since the early 1990s in memories (and sometimes before that, too), there was no 'sunnier day' ahead for profits that even came close to making up for the losses suffered in such 'survival downturns'. Not even close. The exception, the early-to-mid-1990s, saw a wholesale retreat by Japan from the chip market, leaving a vacuum for anyone to fill, profitably. Since the end of 1995, the 'memory industry' has been a net sinkhole for money, by a wide margin, though some players with some products have sometimes made it a good business. But, even in the industry's good times, the seeds of later destruction were already planted, but had not yet poked through the soil. Good profits were added to outside capital and plowed back into the industry, ensuring a supply glut later on and also, at the same time, that R& D had to be maintained at a high level to cost-reduce to meet those low prices.

Wise men (who left) and a very few excellent business men (who stayed competitive) made the right decisions, in retrospect; the vast majority of memory industry players stayed way past closing time, and when the final tally was completed, could only have been deemed failed business ventures.

But, if economic theory tells us anything, it says "Do not do this." Let the failures fail, clear the market, free up the resources to make useful product with true EVA--Economic Value Add. Otherwise, the industry will persist in taking two dollars of resources to make one dollar's worth of end product, creditors and shareholders will lose their investments, as the eventual day of reckoning will come.