In a press release on Oct. 17, Samsung (NASDAQOTH:SSNLF) announced that it had begun mass production of its first system-on-chip based on its 10-nanometer LPE ("low power early") technology. The company boasts that this technology delivers an "up to 30% increase in area efficiency with 27-percent higher performance or 40-percent lower power consumption."
Although this press release might initially be viewed as a positive for Samsung, it really demonstrates how behind its main competitor -- Taiwan Semiconductor Mfg. Co. Ltd (NYSE:TSM) -- Samsung is.
Less than a full generation shrink
Notice that Samsung claims that its 10-nanometer process delivers "up to a 30% increase in area efficiency" -- that's far less than the typical 50% shrink that a new manufacturing generation is supposed to bring. I suspect that Samsung sacrificed area scaling in order to hit a more aggressive rollout schedule.
TSMC, in contrast, is claiming that its own 10-nanometer technology delivers a 0.52x scaling factor relative to its prior generation 16-nanometer FinFET technology (via EETimes). The company says its 10-nanometer process will generate revenue in the first quarter of 2017 -- likely in support of Apple's (NASDAQ:AAPL) A10X chip.
Although TSMC's 16-nanometer technology was slightly less dense than Samsung's 14-nanometer technology to begin with, TSMC should still wind up with its 10-nanometer technology solidly ahead on this front relative to Samsung's using the gate pitch times minimum metal pitch metric (a common proxy for process density).
14/16nm Density (nm^2)
10nm Density (nm^2)
Technology gap further widens in a year
Samsung also said in its press release that its next stop after 10-nanometer LPE would be a process technology known as 10-nanometer LPP -- a higher performance version of its 10-nanometer technology. Although performance should improve, there is unlikely to be any area/cost reduction.
In contrast, TSMC is planning for mass production on its 7-nanometer manufacturing technology in early 2018. TSMC's 7-nanometer technology isn't expected to deliver a full generation area shrink relative to its 10-nanometer process, but the company is saying that it expects 7-nanometer to deliver 1.63x the density of its 10-nanometer technology (via EETimes).
It would appear that when performance, power consumption, and area are all factored in, TSMC will open up a lead over Samsung that could very well be permanent.
TSMC's confidence not surprising... for Samsung
On a prior earnings call, TSMC claimed that during the 10-nanometer generation, it would capture more than 70% market share. Then, recently, TSMC's executives said that at the 7-nanometer generation, the company expects to enjoy share "higher than" what it's predicting for the 10-nanometer generation. I expect TSMC's lead in terms of technology to be a critical factor in achieving these share goals.
TSMC's confidence doesn't exactly bode well for Samsung's efforts, considering that it is currently TSMC's only viable competition at leading edge manufacturing technologies in the world of contract chip manufacturing.
Samsung's future as a foundry looks questionable
Over the long term, I have serious doubts that Samsung will continue to invest in trying to be a leading-edge logic semiconductor manufacturer. According to IC Insights (via EETimes), Samsung is expected to generate just $2.79 billion in revenue from its contract chip manufacturing efforts.
TSMC is, according to analyst consensus, on track to hit nearly $29 billion in sales this year -- more than an order of magnitude greater than Samsung's contract chip revenue.
Assuming a gross profit margin of 40% and an annual research development budget related to these efforts of about $1 billion (half that of TSMC's), Samsung's foundry business would barely be breaking even. Barring a substantial uptick in revenue, the increasing research and development intensity that this business necessitates may simply make it an unattractive one to the company longer-term.
Ashraf Eassa has no position in any stocks mentioned. The Motley Fool owns shares of and recommends Apple. The Motley Fool has the following options: long January 2018 $90 calls on Apple and short January 2018 $95 calls on Apple. Try any of our Foolish newsletter services free for 30 days. We Fools may not all hold the same opinions, but we all believe that considering a diverse range of insights makes us better investors. The Motley Fool has a disclosure policy.
More from The Motley Fool
Why Taiwan Semiconductor Manufacturing Co. Stock Jumped 26.4% in 2017
The chip builder wins when cryptocurrency miners need more of its specialized mining chips.
Taiwan Semiconductor Mfg. Co. Ltd. Is Expected to Build Apple's A12 Chip
Another day, another report claiming that this supplier will keep on winning Apple's business.
Here's Why Apple Won't Switch Away From TSMC
I don't think Samsung will be the winner here.