Intel (INTC) held a webcast today during which it disclosed the micro-architectural details of its upcoming "Silvermont" microprocessor architecture. The presentation began with a broad overview of the market segments that Intel is targeting with the new Atom -- phones, tablets, micro-servers, networking, and in-vehicle infotainment -- but the real "bomb" is in the performance and performance per watt comparisons that the company provided at the end of the presentation. The days of "Intel can't do mobile" nonsense are over.
Performance and Performance Per Watt: Nuking the Competition
As I've said in numerous articles, Intel's current processors based on the five-year-old ones are actually quite competitive with the ARM (ARMH) based competition today, albeit nothing special in terms of performance. I've also said that Intel needs to have a marked improvement in performance per watt over the competition in order to really rack up the design wins. Well, Intel finally gave some numbers and they are devastating to the ARM competition:
Click to enlarge images.
Intel's dual core Silvermont-based chips apparently lay waste to the competition's quad core processors in smartphone power envelopes (1W). At the same performance, Intel uses on average 2.4x less power, and at the same power, two Silvermont cores are 1.6x the speed of the competition. This is not something that can be understated. This is much more dramatic of an advantage -- if true -- than Intel had on Advanced Micro Devices (AMD) when it finally put an end to AMD's leadership in the PC market.
In tablets, the picture looks even better for Intel and worse for the ARM guys, with Intel's quad core beauties either giving twice the performance at the same power level or 4.3x less power at the same performance:
Thanks to a brand new micro-architecture and the world's best process technology, Intel is finally acting like Intel in the smartphone and tablet space -- a winner.
What Does This Mean for Investors?
If these claims are true, it means that Intel will be winning a lot of designs. Why? Because if one tablet/phone vendor uses an Intel SoC, then the other vendors -- using inferior products -- will be at a significant competitive disadvantage whether it be in performance or power consumption. When you are a component vendor, and you have a markedly better component, then you tend to win all the marbles. Ask Qualcomm (QCOM) why it currently dominates the phone space -- it has by far the best modem technology on the planet, and as a result no phone vendor wants to go with anything but Qualcomm for high value devices.
Windows RT's death is now all but assured as there's no way Win RT devices can compete on performance or power/battery life, and Intel should make some serious headway into the Android tablet market. I still think that phones may be a little thorny as modems are arguably more important in this space than apps processors and Qualcomm is still the leader there, with Nvidia (NVDA) and Broadcom (BRCM) also offering strong solutions.
ARM's Outgoing CEO Dead Wrong, but He Won't Suffer the Consequences
ARM's soon-to-be-retired CEO had a tendency to go on about how Intel would never be a leader in power efficiency. This may still be true in the micro-controller space where Cortex M-series chips reign, but it's certainly not true in the phone space and above. Intel's engineer at the presentation boldly referred to "big.LITTLE" as an "abomination" and pointed out that not only could Intel's Silvermont reach higher performance levels at lower power consumption, but Intel could also scale down to lower power levels at the lower performance levels:
All of this also comes without the need to have extra, wasteful cores on the die, or complex software routines to switch between the "big" and the "LITTLE" cores. This is elegance incarnate, and it is clear that both Intel and Qualcomm -- the big dogs in this space -- don't buy into "big.LITTLE" style designs for very good reason. It's a hack.
ARM apparently severely underestimated Intel, and it is likely that a lot of the "hype" that ARM was magical and insurmountable in the low power processor space will fade to black as Silvermont shines brightly during the second half of 2013. It will be interesting to see how ARM's share price reacts to this potentially huge shift in the low power landscape.
Intel is here to win, and it seems that with a new micro-architecture, coupled with yearly refreshes/updates of its "Atom" line, the company will lead a valiant charge into the ARM-dominated smartphone and tablet space, while preempting ARM's attack on the micro-server space. Everything is playing out according to the plan first laid out two years ago at the 2011 investor meeting, and I couldn't be happier with how things are going for Intel in this space.