by Pete Singer, Editor-in-Chief, Solid State Technology
May 20, 2008 – Speaking at The ConFab in Las Vegas yesterday, AMD’s Doug Grose said that the timing is not right for a transition to 450mm wafers, and suggested that the concept of the industry historically moving to a new wafer size every ten years was flawed.
Describing the present macro-economic picture as “pretty bleak,” Grose gave a nod to the cost that equipment and material suppliers must carry in making a wafer size change a reality. “Distribution of profit across our industry is really not equitable. We all know that we can’t create shareholder value without the equipment and software and materials suppliers and neither can they,” Grose said. “So what sense does it really make to pursue an industry strategy that’s not really a win-win for everybody? The last thing in the world we need right now is to starve our suppliers. That’s obviously like shooting ourselves in the foot.”
Grose estimated that 4% of the industry’s 25%-20% annual gains in productivity really come from increased wafer size. The rest comes from smaller die sizes, process improvements and obviously improved yields, defect density, etc. “The cost to the industry of driving to get that 4% a year is going to be very, very large. It seems inevitable we’re going to have to make that move because the benefits are significant,” he said, and asked: “But when, and how do we get the most from 450mm once we do make that move?”
Taking aim at three of the world’s top chipmakers — Intel, Samsung, and TSMC — Grose questioned a key assumption in a recent announcement regarding the 450mm transition, in which they proposed a due-date of 2012 to have a tested and ready 450mm-wafer-capable pilot line. The companies said they chose the 2012 due-date because it keeps up with the industry’s “historical pace of growth” of a wafer-size transition every 10 years (while also hedging bets, adding that integration complexities will necessitate reevaluations of that timeline). Grose chastised the trio for their “spirit of ‘that’s always the way we’ve done it,'” noting that a closer examination at the relative timing of wafer size transitions from the start of production of the next wafer size indicates that the transitions actually have slowed down. “If you do a little bit of an extrapolation, you see that the projected 450mm transition date is actually around 2021, not 2012,” he told the Confab audience. “We’ve got all the right digits maybe we just have them in the wrong order.” Even extrapolating the 15 years it took to transition from 200mm to 300mm, “2016 looks like a much more realistic date than 2012,” he added.
If the currently bleak economic outlook is factored in, “it is pretty obvious” that the industry should not exacerbate the situation with such talk of Herculean shifts in fundamental systems and processes, Grose proposed. “Think about the industry as an ecosystem, where everything is related, where decisions in one area can obviously affect the other areas either positively or negatively. If we all step and think about the classic systems theory where economic systems should be in balance, healthy and obviously nurtured to survive and grow, then what sense does it make right now to really abuse the system?” — P.S.