A former engineer over at SA provided a technical analysis on why Intel will win vs. AMD. Also, as a result, $APPL will suffer.
There has been a great deal of discussion about whether or not Intel’s recent failure to continue its long train of increasing sales and profits – year to year – in each quarterly report, is critical. Many proponents have taken the position that Intel’s stutter is singular and reflects a temporary reduction in PC sales and/or a hiatus before new products become available for the PC market and newer growing mobility markets. Many opponents support a position that says that ARM (ARMH) has decimated Intel in the mobility markets and that there is no way Intel can take market share from ARM. And now, AMD is preparing to enter the fray, presumably to capture market share from ARM and compete with Intel.
The Intel proponents cite significant semiconductor process advantages and new products to be released soon. The ARM proponents cite better architectures and lower power, along with an abundance of design wins in existing platforms. The AMD proponents – well, they don’t have anything to cite except maybe low cost.
How is an investor to evaluate the rhetoric and determine the financial outcome of this battle of words?
My position as a technical expert allows me to evaluate the true capabilities of ARM/TSMC (really ARM/ anybody_who_can_build_ARM_processors) vs. Intel – with AMD nipping at everyone’s tail, when it comes to mobility applications. I am about to do that in a way that should settle this hassle for the near future.
Without further ado, I can state with certainty that the new mobile oriented processors coming from Intel later this year will capture the processor space for the leading smartphones and tablets.
There are two fundamental, related, reasons for this:
- These processors are faster than anything that ARM/TSMC (or AMD) can offer at the same level of power consumption.
- Their power consumption at the same level of performance is much less than anything ARM/TSMC (or AMD) can offer.
You have heard all this before, but many of you don’t believe this and are still arguing that moving to 28 nanometer or 20nm (still planar) will save ARM from the Intel bomb. Well, that’s just wrong! Read on!
These two statements, above, are corollaries to laws of physics relating to silicon FET-based integrated circuits. In current technologies – 32 nanometer and smaller – constructed in a planar (TSMC) fashion, the predominant use of energy occurs as a result of leakage in inactive FETs. The FETs leak because they are running at low voltage, and in order to do that, they must have low threshold voltages. When you construct a FET with low threshold voltage in a planar technology, you can’t completely turn it off. So it leaks.
When you construct a FET for the same use in a 3D or “finFET” technology, the gate (the valve that controls the flow of charge carriers) has much more effective control and is more nearly able to completely extinguish the leakage current in the inactive FETs. Voila, the battery charge lasts longer. And because the gate length is shorter, the FET uses less energy for each transition from one state to another. So you can squeeze more transitions from a given amount of stored energy in the battery, i.e. it runs faster.
It’s the laws of physics that controls the winner in the processor race for mobile applications, and Intel has already won this race. Anyone who disagrees is in denial, has some ulterior motive, or is just dense.
What all this means is that whoever builds phones or tablets using Intel processors can make them run faster and run longer on a given battery than by using an ARM (or AMD) processor.
This argument cuts off the other arguments about market share, design wins, better architecture and other objections by the Intel opponents.
Market share is a measure of the past. What was, not what is. What is, is that the 22nm finFET is a better transistor and it makes for a better processor. If you want the best processor in your next phone/tablet/mobility_device, it had better be an Intel processor.
Architecture is an abstract thing; the “goodness” of an architecture can only be measured by objective effects. How quickly can I do something? How long will the battery charge last? Intel wins again, by actual comparison in Motorola phones!
I have given you the facts that support my statement that Intel has already won this race. If any of you want to continue to argue about this, please supply the facts that support your argument. I don’t argue with people I don’t know unless I know the facts that they are trying to use to support their arguments.
Now, the impact of this win:
It’s clear that Intel will rapidly gain market share pitting phones with Intel inside against phones with ARM or AMD inside. As I have stated elsewhere, this is a case where word of mouth will sell Intel-based phones and tablets. Whoever buys one will be delighted and tell their friend about the long battery life and the great performance – and their friends and their friend’s friends will go out and buy “Intel inside.” Marketing 101!
This will likely have a strong negative impact on Apple (AAPL), TSMC, and AMD. It will take TSMC about 3 years to fully vet a 14 nanometer finFET process in preparation for full production. Intel should be there in a year. So Intel has a near-term 2-year lead; that’s a painful disadvantage in the marketplace.
As an aside, I should mention that the positive effects of the finFET process for Intel are not limited to the mobile market – which to my mind includes smartphones, tablets, and laptops/ultrabooks.
Depending upon the degree of emphasis Intel applies to the Foundry business, this can be very big for them also – not only will customers want to use these processors inside of most ASICs (Application Specific Integrated Circuits), but they will enable the Intel Foundry to gain customers who are unable to build their chips in TSMC processes. This class of customer includes anyone who is trying to build a very large chip, a chip that will melt as a result of its own leakage current when built by TSMC can turn into duck-soup for the Intel Foundry.
And finally, do a Wiki search on “Intel MIC.” There you will find some information about multiple integrated core processors and who is buying them from Intel.Stay tuned for more articles on Intel.