Intel Research Note
1 like135 views
1) The document discusses Intel's strategy in various consumer and embedded markets. It notes that neither Intel nor Infineon have leading solutions for both applications processors and modems. 2) Intel's Atom cores are still not as power efficient as ARM-based SoCs and are not as integrated, which are important for portable devices. Offloading tasks to hardware accelerators is important for power savings versus using general CPU cores. 3) The embedded market offers better margins than consumer markets, longer product cycles, and more persistent revenue streams once designed in. Intel's ecosystem gives it an advantage here over competitors like TI and Freescale.
Intel Research Note
- 1. © 2011 Karthik Ethirajan, all rights reserved Consumer segment may make headline but won’t help the bottom line Neither Intel nor Infineon have an industry leading solution for both apps processor and modem side. Competition is stiff in this segment that encompasses mobile and internet connected devices. There is also a lot of pressure on the gross margin in this business. Intel’s Atom core is still not power efficient compared to the incumbent ARM based SoCs, and neither are they as integrated – two key metrics for portable devices. It is not clear where Infineon stands with their 4G solution or adoption by any major OEMs yet. Historical performance does not bode well for Intel in their argument that Infineon acquisition will help them penetrate Smartphone market in a meaningful way. Prior acquisition of DSPC and JV that resulted in Xscale processor ended up as abject failures. SoCs that serve consumer market is a kitchen sink of IPs licensed from various specialist shops. Imagination and Vivante for graphics core. Apical for camera sensor. Intel has historically chosen to develop all its IPs in-house unlike its competitors. This strategy limits Intel from using the best of the breed IPs available in the industry and increases its time to market. With SoCs, it is all about freeing the core processor by offloading data intensive tasks (such as graphics, codecs and media) to hardware accelerators. This goes against the grain of Intel philosophy where the primary focus has been on CPU frequency. The power savings come from executing tasks on hardware rather than software running on general purpose core. The usage models of newer devices such as eReaders and Tablets are being experimented by consumers. These are fundamentally consumption devices, eReaders more so than tablets. As the market evolves, consumers will settle on a few form factors, likely to serve specific functions such as eReaders, gaming, education, and others. Although laptops stand a high chance of being cannibalized by these newer devices, it will remain as the primary device for content creation. Further, such cannibalization will be far less impactful on the enterprise side where laptops will continue to rule for the foreseeable future. Intel makes a very persuasive argument about the growth of laptops, especially from consumers, in emerging markets. They argue that when the consumer affordability of a laptop reaches 1-2 months of their earnings, the penetration rate of PCs will double. China, Brazil and many Eastern European countries are already at this lynch point. This expanding TAM will more than offset cannibalization of laptop by new form factor devices. If this argument holds, Intel might have found their next major thrust in PC sales with the potent combination of Sandy Bridge and Windows 7 in 2H11 and 2012. Embedded market, on the other hand, holds promise Embedded market composes of many sub-segments – industrial, smart energy, medical, automotive, home appliances, etc. Processors that cater to this segment ranges from low performance microcontrollers to high performance microprocessors. These markets have long product cycles, typically 5-10 years, along with a 1-2 years design in period. As such, high quality (low defect rate) and maintaining product SKUs over a decade are basic requirements for embedded customers. These characteristics coupled with strong channel presence is essential to penetrating this market. Direct customer relationship is still key to winning with major OEMs such as Denso or Honeywell. However, vast majority of lesser known customers can only be reached and serviced through distribution channel partners. Once designed in, the revenue stream from these segments tend to be more persistent.
- 2. © 2011 Karthik Ethirajan, all rights reserved Intel’s established ecosystem of hardware (IDH/ODM), software (ISV) and system (SI) partners gives them a leg up in general embedded market. This market is served by a myriad of processor architectures and well entrenched players like TI, Freescale, Renesas, Atmel, and others. The margins are usually better than consumers market, although it has to be shared with distributors. Digital home group is a trinket This segment is composed of smart TV popularized by the joint venture with Google. This is a growing, albeit nascent, segment that has gained some traction with European telecom operators. Intel claims that ~10K devices are sold every day. Intel will ride the cloud computing wave Unlike the consumer segment, where Intel is trying to catch the tail wind of the rise of smartphones and tablets, Intel is in the middle of the cloud computing wave. Intel is well positioned to capitalize on the infrastructure spend on the build out of the cloud. Adding to its traditional presence in enterprise servers and servers for data centers, Intel is also winning designs in core network equipments like routers and switches, storage and others. Where multi-core is just a marketing tag line for consumer segment, they are absolutely essential and margin adder for networking business. You get paid for incremental performance per core under a certain power envelope. x86 is not the only architecture vying for the infrastructure dollars. Power and MIPS architectures are fairly prevalent in enterprise networks and data centers. Many including Cavium, Netlogic, Freescale and others are also reaping the benefit of the cloud computing wave. The TAM and growth rate for this segment is impressive enough that it offers the best hope to offset softening of the core PC client business. So, where should Intel put its money ? Intel’s R&D program in transistor development is ‘revolutionary’ in their own words. Intel has not only led the industry in terms of transitioning to the next smaller geometry in process technology, but has done so while innovating on the transistor design. The latest 22nm move and 3D structures for transistors will widen the gap in process technologies between Intel and the rest of the world by 2 years. For example, HPC servers can pack a lot of MIPS in each core with a higher performing transistor resulting in improved ASP. A smaller transistor allows producing more chips per wafer driving cost down. As such these R&D and capital investments will support Intel in achieving its new gross margin target of 55-65%.