Are FinFETs too Expensive for Mainstream Chips?

One of the most common things I hear now is that the majority of the fabless semiconductor business will stay at 28nm due to the high cost of FinFETs. I wholeheartedly disagree, mainly because I have been hearing that for many years and it has yet to be proven true. The same was said about 40nm since 28nm HKMG was more expensive, which is one of the reasons why 28nm poly/SiON was introduced first.

Let’s face it, the mobile SoC business drives the leading edge semiconductor processes now and that leaves a large wake of capacity that has already been ramped and paid for. The logical next step is for the foundries to further optimize that capacity for cost and power to enable the next wave of cheap battery powered chips.

For example, TSMC is already working on a more compact version of 16FF+ (16FFC) aimed at the consumer marketplace. The power will be reduced by over 50% and the pricing is cost-competitive for mainstream markets (versus 28nm). 16FFC IP will need to be recharacterized to verify margins but other than that it is FinFET business as usual. 16FFC tape-outs are expected by the end of 2015 with production starting in the second half of 2016, just in time for the iPhone 7 by the way. No coincidence there as 16FFC has Apple juice all over it.

An updated version of the 2015 McClean Report somewhat supports my position:

The total pure-play foundry market is forecast to grow 6.1% to $44.9 billion in 2015. The entire increase in pure-play foundry sales is forecast to be due to sales of devices built using leading edge (<40nm) feature sizes. The <40nm pure-play foundry market is expected to increase 24% to $16.1 billion in 2015 compared to $13.0 billion in 2014 (Figure 1). In contrast, pure-play foundry sales of ≥40nm devices are forecast to decline 2% to $28.8 billion, based on findings presented in IC Insights’ September Update to the 2015 McClean Report.

Figure 2 shows sales of ≤45nm devices from the Big 4 pure-play foundries on a quarterly basis for 2014 and 2015. IC Insights forecasts that 63% of TSMC’s 2015 sales will be based on ICs built using ≤45nm technology. Furthermore, TSMC is forecast to have about $5.7 billion in sales of ≤20nm devices in 2015 ($5.1 billion from 20nm and about $0.6 billion from 16nm devices). That would be 2.7x more than the $2.1 billion worth of ≤20nm product the company sold in 2014. The company began volume shipments of its 16nm devices in 3Q15 and stated, “The ramping of our 16nm will be very steep, even steeper than our 20nm.”

According to the most recent TSMC Conference call, 16nm/20nm chips accounted for 21% of wafer revenue and 28nm chips accounted for 27%. That compares with 20% for 20nm and 27% for 28nm in the previous quarter. It was also said that 16nm is still ramping and is expected to contribute even more next quarter.

The bad news from the TSMC call is that in 2015 smartphone unit shipment growth is +10%, PC is -6%, tablet -14%, and digital consumer electronics -6%. The semiconductor industry growth is about 0%, fabless growth is -5%, and TSMC growth will be about 10%. Not all bad news I guess.

The most recent IC Insights Bulletin also lowered the Worldwide 2015 IC Market Forecast from +1% to -1% due to a slowing China economy, strong U.S. dollar, and falling DRAM prices. The good news however is that they expect in 2016 the IC market will again register mid-single digit growth.

Follow the adventures of SemiWiki on LinkedIn HERE!