It has always been assumed that, outside the foundries, the three classes of people who will still own fabs in ten years time will be 1: Memory manufacturers; 2,: Specialty manufacturers e.g. mixed signal, power, MEMs etc; and 3: Intel. If Spansion and TSMC can show you can compete in leading edge commodity memory while using a foundry, then that class of three reduces to two. Of course TSMC and the other foundries have been making memory, both discrete memories and embedded memories, for ages, but, until now, so far as is known, they have not attacked the discrete memory business at the leading edge. For instance, TSMC has been producing 110nm production flash wafers for Spansion since Q2 last year, and is ramping up 90nm. But 110nm and 90nm are some way behind the state of the art as practised by flash memory market leader Samsung, which is in volume production of 8Gbit and 16Gbit flash on 50nm. Now Spansion and TSMC have signed an agreement to develop a 40nm process for flash. Unfortunately the timescale is not disclosed. But if a commodity memory player like Spansion can compete at the leading edge using a foundry, then that’s a significant catalyst for fundamental industry reorganisation.
Can Foundries Compete In Commodity Memory?
The news that Spansion and TSMC are getting together to push flash process technology to the limits raises the interesting question of can a foundry compete with an IDM in a mass commodity market?