DRAM Market Problems Escape All Solutions So Far

DRAM Market Solution? You Won’t Find It Here!:
If you are reading here, expecting to find a solution to the “DRAM Crisis” of red ink, low prices, huge product development costs and ‘next-generation’ process costs for fabs, to the slowing of DRAM demand, and arresting the more-than-a-year now DRAM supply excess...then stop here. I do not have the solution. There is no quick fix, and maybe there is no real fix at all, especially with everyone wanting (no, insisting on) being one of the survivors, and eventually finding the funding for life extension from governments, parents, or compliant bankers. One loan 'at risk', begets another to save the first.

This is not the first time I have watched as the DRAM makers and their fellow travelers marched lemming-like to their own destruction, but it may be the last time (or maybe not).

They are a hearty breed, to be sure, but, though I do not have the numbers in-front of me, I doubt that DRAM makers collectively have made one cent...no, in actuality they have lost billions of dollars, at least since the end of 1995, and maybe even by looking further back to 1985. Along the way, in the downturns of 1985-6, 1989-90-91, 1998, 2001-2-3, and again today, I have heard every argument as to why ‘we have to be in DRAMs’: its role as a process driver, or as a new tool-prover, or its the high volume runner, or the idle capacity fab filler (depreciation absorber with a limitless market) with the sense of ‘imperativeness’ has been able to justify all manner of financial excess, bad business decisions, and postponements of the inevitable in which one survivors’ ‘survival solution’ invariably inflicts huge pain on all others.

Recall, I was at IBM when they pulled the plug on DRAMs in 1999...a final agony that started probably on 1 January 1996, when prices dropped from an anomaly of a historically unique sustained DRAM pricing level that had been characterized by “13 and 52” (dollars per unit for a 4M and 16M DRAMs respectively.) since about late 1992. Of course, I fought the decision to leave the market, “If only we would TRY harder,” adopt best practices, invest in scale, etc. Better minds with broader visions prevailed, and eventually, but not after an unnecessary $2.5B operating loss over those four years it took them to pull the plug. Never had the “dollars loss/unit of market share” been higher, as IBM left with maybe 3% DRAM share. The industry, especially that which is centered in Taiwan today, is locked in the same decision-prolonging agony. Each quarter brings billions of additional dollars in losses, and the proposed solution, to build up a home-grown DRAM capability, will only keep the market oversupplied and keep the red ink flowing for longer.

So, when I see the posturing, the attempts to slow down capacity expansion, then take capacity off line, then decommission the vast majority of fully-functional 200mm DRAM lines, the chatter of this company talking to that one about merging operation to achieve scale(to reduce R& D and expense cost/unit shipped), and governments trying to keep employment and their attendant income taxes high, unemployment benefits low, to see banks forgiving and rescheduling debt...and Taiwan (reportedly) cajoling the Island’s DRAM makers (PSC, ProMOS, Nanya) to have a workable technology roadmap plan, and maybe join forces with one another. Not that those companies do not know already have about $13B (US) in debt, the government is willing to put another $6B (US) on top of those funds to ensure their survival.

Every day, we see floated another ‘solution’ to solve the problem. For certain, the DRAM supply reductions have had a substantial effect in stabilizing prices (at some significant cost that we will see more of when they report in coming weeks). But the JV-think and ‘producer rationalizations' that have been implemented, at Hynix through more bank loans, at Qimonda with infusions of government money and shedding some capacity, with Micron-buys-Qimonda's-stake in Inotera and redoes its business plan to present to Taiwan Government, though useful, is not a sustainable solution unless the mindset changes and significant capacity is taken down and investment slows. Literally, not in decades has the DRAM industry made enough profits in good times to begin to make up for losses in bad times. For better or worse, they are working against fundamental industry characteristics of fixed costs, huge scale economies that favor very large producers, and never-say-die mindsets that pour good money after bad and harm everyone. It may be good for DRAM consumers, but most would argue that, ultimately, it is bad for bankers and their shareholders and for government investors. Today, we are also faced with DRAM demand that is weakening, largely from PCs...and an imbalance that is exacerbated by the steady progress to finer line-geometry DRAMs that yield more DRAMs/300mm wafer than ever before; for sure, 'cost reduction' brings its own problems with increased output. DRAM production might grow 40-50% in 2008 without adding a wafer. Today, die-per-wafer start is probably 20-25% more than only six months ago, as 6F2 cells (or 4F2 cells tommorrow) replace 8F2 cells, and line geometries converge in the 58-65nm zone, down from an average of about 70nm+ in 3Q08...enroute to 50nm processes in 2010.

In the Great Depression in the US, the government bought apples from the growers and crushed them into the Columbia River, to support the prices farmers got; one solution proposed for the US housing surplus and its attendant downward price pressure, only half in jest, was for the US government to take just 1/3 of the TARP $750B bailout funding, and buy 1 million houses off the market and burn them down, reducing supply and stabilizing prices. The Taiwan proposal has no such solution embedded in it, but the cost is still high. It will bring more competition, and more losses, both before and after it is implemented. Surely, we can use those DRAM fabs to make something useful and profitable, with a little rearrangement of the manufacturing facility, and of our thinking. Today, the world is awash in excess production capacity, and not just in DRAMs; life sustaining investments in ventures that should be folded up, serves no useful purpose.

So, please, back to the drawing board, to find a workable solution in which the seeds of its own destruction are not already planted. And the sooner we do, the sooner DRAM prices will stabilize at profitable levels, and we can get on with things.