What are the business aspects of 3D integration processes that need to be considered when deciding which approaches will be best for volume production?
This is a difficult question. From a business standpoint you want the most financially strong entities with the greatest experience and track record. True 3D integration experience is pretty thin among the typical semiconductor supply chain companies. Today it is also pretty hard to say who is strong given the current market difficulties. I think the business view must also look at the under pinnings of the 3D process desired. Can the process be brought up at another company if things go wrong?
Your company has been following the business trends of 3D integration. Which companies are well postioned to take on 3D? Are the bigger companies waiting to evaluate early adopters success before throwing their hat in the ring?
The first companies that have started 3D TSV integration are large companies (Toshiba, ST for example) that are able to spend the extra money for TSV process because they get decrease of cost in other places of the value chain. This is typicaly the case of CMOS image sensors.
At YOLE Développement, we see 4 type of players which are well positionned to take benefit of the TSV revolution: the IDM and the foundries (because TSV can help them diferentiate compared to copetition), OSAT (a new packaging business for them), substrate manufacturers (new substrates will be used to host TSV stacked die) and MEMS foundries (they are already in medium volume production of TSV stacked chips since more than 3 years for some of them). Which one will make the best ROI is still a question bt their is clearly a different business for these 4 types of companies.
As it's been indicated the TSV fabrication itself is best suited to the front-end foundries, and that packaging houses are most suited to handling back-end processes, won't it require both entities to get involved before real volume production can be realized? It seems unlikely that the foundries will invest in back-end processes, and vice versa.
Discussion Summary
While companies with strong financial postions, experience and track records are likely to be the most suited to take on 3D integration, few exist (Toshiba and ST Micro come to mind) that have absorbed the risk of investment thus far. Yole Developpment research indicates that IDMS, foundries, OSATS, substtrate manufacturers and MEMS foundries are the most likely candidates, but which will take up the gauntlet remains to be seen.
Print