14 DAY TRIAL // Although one of the foundry's top customers, Qualcomm, is frustrated about still not being able to meet demand for its 28nm chips, TSMC seems to be taking it “as it is” and is already making plans for next-generation manufacturing technology.
Most industry insiders know that there is no real problem at TSMC, and that yields are good and the foundry is doing its best to increase capacity.
The only “problem,” if we can call it that, is on Qualcomm’s end.
Practically, it’s Qualcomm’s own “fault” that they designed such a powerful and efficient ARM Cortex A15 architecture that is wanted by every tablet, netbook and mobile phone manufacturer on the globe.
Maybe it wasn’t obvious before, but now, seeing that Qualcomm reports that they’re still not able to meet demands despite working with TSMC, UMC, Samsung and GlobalFoundries, it’s clear that the problem is the incredible demand and not TSMC’s lack of achievement.
TSMC is reportedly already working on 20nm technology, and they have been working for more than a year.
The company is reportedly saying that they’ll start making the first chips in 2013, but they caution any eager customer that this will be a very low production level.
Likely, less than 1% of TSMC’s total production will be on 20nm next year. The company is calling this “risk production” and the normal volume production will only start in 2014.
There will only be one single 20nm manufacturing option offered by TSMC, unlike today, when the foundry is offering four different 28nm options with specific optimizations for performance or power efficiency.
The company is also reporting that they will start 16nm FinFET production ramp in the second half of 2015, which means that risk production will begin in 2014.