Seeking Alpha
Long/short equity, contrarian, independent research, tech
Profile| Send Message| ()  

My faithful readers (or people who just happened to stumble across my work) know that I believe that Intel (INTC) bears are about to get fried. While many in the tech press are simply touting Haswell as the chip that changes everything (it sort of does), it is important to understand that there is a long term fundamental change going on at Intel, and it's going to be a wonderful ride for shareholders. "Haswell" is just part 1 of the strategy.

Haswell Brings Substantial Platform And CPU Power Reduction

The Haswell story is rather complex, but I'll try to explain it as best as I can. Every year, Intel puts out a new chip. On odd numbered years, we get what is known as a "tock", or a brand new micro-architecture built on-top of mature process technology. A "tick", on the other hand, is more or less a port of the most recent "tock" to newer transistor technology, which means area and power savings, but not to the extent to which a new micro-architecture would allow (since performance/power characteristics are determined largely by micro-architecture and the quality of the implementation of said micro-architecture).

Haswell is a "tock" - that is, a brand new design on the 22 nanometer transistor technology first introduced with "Ivy Bridge". Clock-for-clock in code that has not been optimized for the new features of chip, Haswell brings roughly a 10% improvement on a per-clock basis. This is usually considered pretty good, particularly as the process technology has not changed, and area/power constraints are certainly in place for the kinds of form factors and price points that the company is trying to achieve.

However, the real story here, is that Haswell brings rather substantial power management features along with significantly lower platform power. While Haswell brings power efficiency up at full load, the key is that when the platform is doing nothing or very little (which is what most computers do, and why tablets and smartphones have long battery life), the processor and platform see a rather substantial reduction over the previous generation.

Battery Life Improvements, 6W "SDP" And iPad-Thinness

In a nutshell, these improvements all translate into substantial battery life gains while improving performance and keeping the die sizes rather small (quad core, 20EU part is only 177mm^2, not much larger than the quad core, 16EU "Ivy Bridge" counterpart). If Intel's claims are to be believed, then Haswell-based Ultrabooks will last substantially longer than Ivy Bridge-based Ultrabooks given similar displays and battery capacities:

(click to enlarge)

Additionally, while the previous Ultrabook chips based on Ivy Bridge required an external chipset that sucked down ~3W in addition to the 17W TDP (maximum power consumption) of the processor itself, the new Ultrabook-oriented Haswell processors are rated at a 15W TDP. So, performance goes up, maximum power consumption goes down, and idle power consumption comes way down for Ultrabooks.

More interestingly, though, is the "Y" series parts meant for systems that put the processor behind the screen, rather than in the base, see a real efficiency improvement. While the previous generation Ivy Bridge Y chips were rated at a 7W SDP ("SDP" just means that the chip restricts its maximum clock speed to hit this new power envelope), these required the 3W external chipset, for a total "kit SDP" of 10W. The Haswell-based "Y" series parts are rated at a 6W SDP - and that includes the chipset which is now brought on package. I expect that after "Broadwell" beginning with "Skylake" the "U" and "Y" series chips will actually be full single die solutions.

What does this enable? Well, according to Intel, these chips will be going into designs that can be as thin as 10mm thick; for comparison, the Apple (AAPL) iPad 4 is a mere 9.4mm thick. Haswell will be going into sub-11.6" tablets that can be nearly as thin as an iPad.

(click to enlarge)

What Does All Of This Mean?

While the "PC" market has been shrinking, the overall market for computing devices is clearly on the upswing, as tablet and smartphone sales have clearly shown. I believe that what we are seeing is simply a form factor shift; people want touch screens, all-day battery life, and more performance. The "good enough" argument doesn't really hold water since ARM (ARMH) and its licensees are feverishly getting into the "performance wars" and ratcheting up power consumption to get there. Device vendors know that more performance enables fundamentally new and better software experiences, and so a company that can provide higher performance chips can usually win the really good designs while commanding pricing power.

In short, while the traditional PC continues to lose share, a whole new wave of convertible, detachable, clamshell, and pure-tablet devices that optimize for a wide variety of parameters will hit the market, and users will be able to get substantially higher performance and responsiveness than they are used to with traditional "tablet" processors. No longer will people see the need to buy a 10" tablet and a notebook PC - they will buy a great tablet that is also a good PC (or a great PC that is also a good tablet).

Intel, with a micro-architecture advantage, a process technology advantage, a co-optimization of process and micro-architecture advantage, a scale advantage, and - maybe most often overlooked - a strong brand that resonates with consumers, is set to benefit heavily from this trend. The recent positive tech press and the surprising resilience in Intel's share price amidst a market correction only serves to fuel the expectation that Wall Street is finally starting to "get it", and that within 6 months, nobody will even remember anything about "the death of the PC" and "excess capacity charges".

Conclusion

I remain firm in my conviction that Intel is on track to hit $48/share by 2016, and the imminent roll-out of the new systems based on "Haswell" (as well as "Baytrail", and a slew of other things on the phone as well as server side of things) will finally put to rest the doubt and fear that Wall Street has about Intel's long term viability. Stay tuned for part 2 in this series, "Intel's Silvermont: The Must Read Guide".

Source: Intel's Haswell: The Must Read Guide

Additional disclosure: I am short ARMH