by James Montgomery, News Editor
“Food for thought” served up during a SEMI breakfast panel near Boston suggested memory firms’ wild overspending and capacity increases may in fact be shrewd strategic moves to shore up leading-edge 300mm capacity before an oncoming technology “brick wall” makes their existing 200mm fabs obsolete.
Gartner’s Bob Johnson showed data tracking 200mm and 300mm DRAM and NAND capacity, showing how much 300mm capacity is being rushed into the memory segment. With the onset of advanced technology requirements where 200mm/75-65nm just won’t cut it, that means 30% of total capacity will start to become obsolete in the next year, he noted [see slide, below]. Replacing that capacity amounts to about 600K 300mm wafers/month, costing roughly $35-$40 billion (based on an average fab output of 50,000 wafer starts/month, meaning 12 300mm fabs). Of course some of that 200mm capacity will be moved over to other devices (as Micron has done with image sensors, for example). But turning that capacity usage over as a foundry operation isn’t a likely scenario, nor is sending it overseas to China or India. And 150mm capacity is still sticking around (~25% of worldwide capacity) where upgrades really aren’t necessary, he said. Thus, piles of 200mm capacity are likely to simply be retired.
Which brings up an interesting “what if” scenario, Johnson pointed out. What if memory firms, instead of blindly overinvesting in 300mm capacity in a race to stay ahead of each other and not be squeezed out of the future market, are in fact smartly building up 300mm capacity ahead of a need to retire all this 200mm capacity? If — or more accurately, when — memory markets hit an oversupply situation, this 200mm capacity can be shut down to preserve a supply/demand balance, ASPs, and profits, he suggested. And isn’t it interesting, he posed, that the timing of this 200mm obsolescence coincides with an anticipated memory market crash?
Johnson admitted, though, that this scenario was mainly a mental exercise, and Bill McClean quickly discounted the notion. Samsung et al. probably do have people in-house who sit and think ahead several years strategically like this, he told WaferNEWS in a post-presentation discussion. But they are vastly outnumbered by those with an overriding drive to take market share, ship units, and win at all costs. — J.M.