As we posted last week, DRAM cyclicity is at it again. With a commodity like DRAM, that is susceptible to numerous market dynamics, there are renewed questions as to where exactly this segment is headed? As we've noted, while July witnessed price reductions, the third quarter demand is still expected to pull DRAM into another shortage situation, with iSuppli forecasting "[…] bit growth of approximately 11 percent." Other analysts, such as Hans Mosesmann, an analyst with Raymond James & Associates, continue to push 13-15% bit forecasts for the remaining two quarters of 2010, as reported here by EETimes Europe.
The question of undersupply is still on the table. The reasons go back not solely to the question of increased demand, but rather, back to the issues around how CAPEX is being spent. Rather than adding lines, most manufacturers have moved their attention to the next architectures below 50nm, which require new equipment (now necessitating immersion tooling) and also introduce immersion yield questions, as explored in the iSuppli forecast. Those DRAM manufacturers which have already overcome these immersion issues (tools and yield) will be in the best market position.
A significant variable contributing to the tooling problems is equipment allocation – equipment manufacturers for immersion tools (esp. Nikon scanners, for example) are still in a supply shortage situation. Additionally, the tools are becoming increasingly more costly as the architecture shrinks for DRAM, further limiting the number of manufacturers willing to expend CAPEX on a volatile, yet fundamental, commodity such as DRAM. Whether or not 3-D approaches utilizing through-silica-vias (TSV) will provide solutions that are more feasible for memory is yet to be proven in the wider marketplace, due to the lack of standards, testing, and cost models (see EETimes Europe here for additional discussion).