When it comes to mobile chips, there has always been something "off" about Intel's (INTC 1.32%) execution. This is the company with the world's largest semiconductor research and development budget, a rich legacy of manufacturing technology leadership, and the highest revenue level of any semiconductor company on the planet. Yet making chips for cell phones has been a problem for the Intel, even as much smaller semiconductor players like Qualcomm (QCOM -0.60%) and Broadcom (NASDAQ: BRCM) succeed in the sector.
Clearly something was wrong internally, and at this month's London Analyst Summit, Intel Vice President Josh Walden, general manager of the company's platform engineering group, offered some perspective on why the chipmaker has struggled and what organizational actions it has made to remedy this apparent deficiency.
Reusable IP blocks -- it's like Legos
According to Walden, the "old" way of doing things within Intel was to essentially "handcraft" each CPU. Since Intel's product lineup has traditionally been very narrow, and variants on a given product have usually been limited, this was a successful strategy. However, as the world transitioned to a multitude of different devices, each perhaps requiring different functionality/IP blocks, Intel has needed to become much more flexible in cranking out variants on a given chip.
IP reuse is the name of the game at Intel. Source: Intel.
The idea here is that the company will have teams that develop a broad swath of IP blocks, such as the CPU, graphics, image signal processor, and so on, depending on the project (a smartphone's requirements will be different from those of, say, a microserver processor). Then, the various integration teams would go ahead and put together the right IP blocks (either internally developed or -- in the case of more commoditized IPs -- an externally licensed IP) to spit out the right system-on-a-chip for the right market segment.
This is something that Qualcomm, Broadcom, and others have been great at
What Intel is proposing is a pretty fundamental shift from how it does business, but it's nothing new within the industry. For example, Broadcom -- which develops everything from high-end network processors to chips for set-top boxes -- has had such a methodology in place for some time now, as it was keen to illustrate at its recent analyst day.
Source: Broadcom.
Qualcomm, too, has demonstrated that it is quite adept at building various flavors of its chips for various market segments. It has a robust top-to-bottom, targeted product stack for everything from the lowest-end handsets to the highest-end "hero" phones. Many companies can compete with Qualcomm on a few IP blocks or in a few segments, but no other company has been able to match its portfolio thanks to this very refined engineering methodology.
It could finally be time for Intel to succeed in mobile
Intel certainly has no problem investing huge sums of money on the R&D side of things to make inroads into mobile. If the corporate/organizational structure becomes as system-on-a-chip oriented as Intel has pledged, then perhaps we are seeing the light at the end of the tunnel for the company's mobile struggle. Certainly, the competitive landscape is fierce, and competitors like Qualcomm won't make it easy for any rival to win easily, but Intel might finally have the structure to compete properly in this market.
Will Intel succeed? While there are no guarantees in the world of high-tech, it is clear that Intel is in this for the long haul. We'll see how the company manages to do during 2015, when the fruits of the reorganization that CEO Brian Krzanich initiated about a year ago can begin to be appreciated in real, shipped products.
