Image via WikipediaPosting about the semiconductor industry has been a cheerless task lately so I am finding it satisfying to be writing a positive account of Nehalem, Intel's newest microprocessor.
By all accounts a great product, is it significant enough to truly make a difference to Intel or the industry as a whole?
Before we can answer that question, we should present a little background. Nehalem is the codename for the new Intel processor microarchitecture. Nehalem processors use a 45 nm manufacturing method, currently Intel's most advanced geometry though 35 nm is already in the works. What distinguishes Nehalem from its predecessors is that the new chip has a built-in memory controller for each CPU and there are four CPUs per chip. Another major improvement is the replacement of the old front-side bus with QuickPath, a new high-speed interconnect system that routes data between the CPU, memory, graphics controllers, etc.
So much for the geek-speak (yes, I love this stuff). The bottom line is that the outcome of these architectural decisions is blazing speed. Intel is leaving competitors; ie, AMD, in the dust. The Nehalem chips, now branded as Core-i7 and optimized for single socket systems and dual socket systems are now the standard against which other PC processors and Windows-based servers will be judged. Various spokespersons at Intel have said that Nehalem represents the biggest performance jump since the introduction of the Pentium Pro back in 1995
This is like a challenge to a lot of the folks who run tech blogs so there has been a flurry of benchmark testing and it appears that Intel's claims are accurate. Word is that Intel's partners are reporting performance improved by a factor of two and in some cases as much as three. Taking a shot at Sun and IBM, Intel further claims that the era of proprietary and RISC/Unix computing is over, declaring that a two socket box running the high end Nehalem processors can easily compete with the IBM Power server and Sun UltraSparc.
In addition, the new chips are far more power efficient than predecessors and competitors, drawing 70 watts less than comparable systems. Where this gets interesting is when you consider the additional performance of the new processors combined with their power efficiency, the performance-per-watt difference is big enough to act as a significant differentiator.
Climbing on the virtualization bandwagon, Intel has included features that are intended to reduce the overhead of running in virtual environments so that performance approaches that of running on native machines.
Intel's various customers are now faced with a situation where they need to immediately incorporate the new Nehalem chips in their products because their competitors certainly will. The speed advantage the chip offers can't be ignored so, in order keep up with their rivals, Dell, HP, Cisco and IBM, for example, will be forced to quickly roll out products with the new processor. Apple already offers a rackmount server utilizing two processors and claiming an 89% improvement in performance-per-watt.
IBM and SUN and the other vendors of Unix-based servers such as H-P will also have to answer Intel's claims that the speed and reliability advantage no longer belongs to Unix.
Interestingly, AMD pioneered certain aspects of the kind of architecture Intel has adopted but AMD is now in the position of having to play catch up.
So, game changer or not?
There are some who say that the raw power of these processors will eventually transform all kinds of computer tasks. What used to take hours can be done in minutes. This means hard things can be done quickly and really difficult things become reasonable to attempt. Since versions of these chips can be used in workstations and PCs, not just servers, we are now going to see a leap of computing power put into the hands of individuals. Where will that take us? Hard to predict but it will no doubt be somewhere no one expected.
As for Intel, it has to be a significant positive. The company, always the major player in the processor market, takes a clear leadership role and the new processor family becomes the flagship product in the Digital Enterprise Group business segment that generates more than half of Intel's revenue. While the Mobility Group has shown greater growth through the surge in notebook and netbook computing, the new Nehalem processor assures Intel that it's bread and butter business segment will maintain its strength.
More to the point, however, will the new processor allow Intel to generate superior growth? That is an important question. Looking at the following chart, it is clear that Intel, despite its strong position in the markets in which it participates, has actually been an under-performer over the last five years.
So Intel currently holds the high ground. Sun Microsystems is staggering, racking up losses and failing to nail down a deal with IBM. AMD is splitting itself up and trying to reduce losses. Dell is also suffering profitability problems. IBM is coasting on its services revenue. H-P seems to be more involved in cost cutting and integrating their EDS acquisition. Intel seems to be the company exhibiting leadership in R&D.
Can this leadership translate into earnings? Introducing this product during a recession implies sales in the near term may not set the world on fire but as the economy begins to improve and enterprise IT budgets loosen up, Intel is well positioned to grab share from AMD and Sun. Those companies who offer both Windows-based servers and Unix-based servers may see a shift in product mix toward servers with Nehalem processors. This means they could become even better customers of Intel. So the potential for significant earnings growth is there.
With Intel currently under $16, up 30% from its recent lows, and its PE just over 16 it would seem that the company is fairly priced at the moment. The new processor family has the potential to add value to the Intel franchise at a time when it needs a higher-margin driver of sales growth to offset the move toward netbooks.
The contest, then, is well underway and Nehalem might just be the game-changer that Intel needs.