Last month, I argued that microprocessor giant Intel (INTC +4.49%) didn't have much of a chance of winning the manufacturing contract for Apple's (AAPL 0.49%) A-series processors, which serve as the brains of the iPhone, iPad, and Apple TV.
At the time, I cited Intel's spotty track record in recent years in transitioning to new chip manufacturing technologies, as well as potentially limited benefits to Apple from switching away from trusted suppliers to a supplier with said spotty track record.
Image source: Intel.
In reviewing Intel's recent Technology and Manufacturing Day in China materials, I came across some information that strengthens my belief that Intel's technology simply isn't good enough to be worth the risk to Apple.
Intel's best bet was to win on performance
Intel has historically spent a lot of time talking about the relative densities of its chip manufacturing technologies. In a nutshell, Intel claims that a chip designer can integrate more stuff in a certain amount of silicon by using Intel's latest technologies than by using technologies from Intel's main competitors, Taiwan Semiconductor Manufacturing Company (TSM 0.28%) and Samsung (NASDAQOTH: SSNLF).
However, technology density isn't everything. The performance of the technology, which ultimately influences the performance and power efficiency of chips built using that technology, is just as important, if not more important, than the density of a technology relative to its competition.
During Intel's Technology and Manufacturing Day, the company -- much to my surprise and delight -- provided a competitive comparison of its upcoming 10-nanometer technology (as well as its third-generation 10-nanometer technology, known as 10nm++) with TSMC's and Samsung's currently shipping 10-nanometer technologies.
Here's that comparison:
Image source: Intel.
At first glance, this looks really good for Intel. The company's first-generation 10-nanometer technology is clearly ahead of the 10-nanometer technologies from its main competition, and things are supposed to get much better when 10nm++ rolls around.
There's just one problem here: This slide doesn't represent the competitive landscape that Intel's 10-nanometer and 10nm++ technologies will face, particularly when we're talking about the contract chip manufacturing landscape.
Realistically, Intel's 10-nanometer technology will need to face off with TSMC's upcoming 7-nanometer technology, as well as Samsung's 10-nanometer LPP technology (a performance enhanced version of its currently shipping 10-nanometer LPE technology).
Samsung claims a 10% performance boost for its 10-nanometer LPP technology compared to 10-nanometer LPE, and TSMC claims a 20% speed improvement for its 7-nanometer technology compared to its 10-nanometer technology.
This means that, if Intel's graph is representative, its 10-nanometer technology should only have a roughly 10% performance edge over Samsung's 10-nanometer LPP and no real edge over TSMC's 7-nanometer technology (which is what Apple is likely to use for its 2018 iPhone lineup).
Moreover, TSMC claims that its 7nm+ technology, which I'd peg as the most likely choice for Apple's 2019 iPhone processors, will deliver a 10% improvement in performance compared to its original 7-nanometer technology.
That would mean that TSMC's 7nm+ should deliver similar performance to Intel's 10nm++ technology -- at least in the power ranges relevant to Apple's needs -- but would arrive earlier than Intel's technology.
This doesn't bode well for Intel's odds of trying to snatch away Apple A-series chip manufacturing from TSMC anytime soon.
