TSMC will eventually win given its high volume history with big chip companies. One stop is convenient for these types of businesses and what TSMC is offering makes sense.
How ever, I don't think this bodes well for smaller companies and startups of tomorrow who can really innovate with applications and required processors enabled by 3D stacking. They may not have the volumes high enough to be even get their emails replied to by TSMC!
TSMC has to compete against Samsung which already holds many of the cards
Samsung might come out the big winner in the 3D wars !
To gain the most ( speed, power saving, cost ) out of the pain involved in 3D TSV, co-ordination of die floor - plans for shortest interconnect length would be critical.
Samsung can control everything themselves. They already
- fab APs for Apple ( A... A6 ) now at 28 nm
- design and build their own ARM based APs
- have 40 % market share in DRAM ( down to 30 nm ) and NAND ( 20 nm )
- showed a year ago REAL 3-D memory stacks( not vaporware or computer graphics, unlike some other claims ), complete w/ bandwidth measurements & SEMs of x - section )
- builds their own substrates
Fabless wonders who depend on TSMC for high margins and market share in the Smart Phone / Tablet segment might be sweating.
Blog Doing Math in FPGAs Tom Burke 24 comments For a recent project, I explored doing "real" (that is, non-integer) math on a Spartan 3 FPGA. FPGAs, by their nature, do integer math. That is, there's no floating-point ...