EDA Tech Forum Keynote Dataquest's Martin Reynolds offers up detailed predictions for future markets ...
Martin Reynolds, vice president at Gartner Dataquest discussed "Markets, Technologies, and Products: Finding the Killer Apps" in his keynote address delivered at the EDA Tech Forum in early November at the Santa Clara Convention Center. Reynolds started by enumerating the IT and semiconductor markets. The U.S. is the largest consumer of information technology with total sales of $852 billion - approximately 20 percent in hardware and the balance in services and software. Europe follows with $692 billion, then Asia/Pacific at $390 billion and Japan at $292 billion. The overall market is growing slowly, with compound annual growth between 3 and 7 percent from 2005 to 2008. The growth rates vary by region. Within the U.S. IT market, PCs and cell phones are the largest segments with personal computers accounting for $206 billion and cell phones for $119 billion. The fastest growing segments are networking with a CAGR (compound annual growth rate) of 7.3 percent from '05 to '08 and storage growing at 4.8 percent. The overall market is in a transformation phase and not expected to grow much in the next few years. A lot of the hardware purchases are replacements of older machines, a buying cycle that was distorted by the high spending at the turn of the century. Both PCs and cell phones are in transformation zones, adding new features and functions to both categories of product and serving as drivers for changes in technology. One shift in PCs is the change from desktop machines to laptop or mobile devices. Mobile devices will exceed 1 billion units in the next two years. Desktop purchases are replacement systems that are just sales taken out of future sales for computers. China is one exception to this replacement trend, where the purchase of desktop machines represents all new sales. Overall, enterprise spending is seeing a bubble left over from the end of 2004 and should show a decline in the last two quarters of 2005. Servers will grow in units and dollars in all regions with Internet and intranet functions as the primary drivers. Emerging trends One emerging trend is the replacement of RISC architectures with x86 platforms. The x86 machines have sufficient scalability and reliability to function as servers for non-critical functions and can produce as much as 60 percent gross margins. One exception to the reduction in RISC machines is the Intel Itanium, which has a growing share of the shrinking market. Semiconductor markets are going through their normal business cycles. Both 2005 and 2006 will have sluggish growth with total revenues of about $230 billion, followed by strong growth in 2007 and upwards of $300 billion in sales in 2008. The following year, 2009, will show a small decline followed by strong recovery in 2010 with sales expected to exceed $300 billion. Market percentages by device in 2004 showed ASSPs with 24 percent of the market, followed by memory at 23 percent and microprocessors at 21 percent. The other third of the market included all other product categories. Non-optical sensors are seen as the kings of growth with projected increases in sales of about 18 percent. In 2006, memory is expected to be out of sync with the rest of the markets, dropping as much as 21 percent while other markets are rising. One cause for the disparity in markets is the value of branding. Intel gets about $150 for a Pentium 4 versus the $80 AMD charges for its Athlon. The manufacturing costs are roughly comparable, so Intel gets about $100 for name brand. In comparison, memory is considered a commodity with essentially no brand recognition. The value of memory is in the silicon, which accounts for approximately half of the total price. In comparison, all other product categories have silicon costs which represent no more than a third of the total cost. Killer apps To get to the drivers for markets and determine the next killer application, one must look out to a 10-year horizon. That point of view perceives a massive increase in data volume and transmission, implying a need to build larger data systems. Expect, therefore, to see an increase from 109 to 1011 connected devices by 2009. This trend will cause increases in spending for IT hardware and a compensating decrease in the number of people operating those systems over time. Management is currently spending $70 billion for hardware support, but the next generation systems will have to migrate towards zero management. To reduce costs of operation, systems will need more virtualization in services and hypervisors to make the systems capable of automatically adding new network components and tolerating device failures without system failure. This new layer of software will enable the necessary reductions in manpower and will facilitate network self management to minimize system sensitivities to failures. In addition, the network will extend further out to include (smart) cell phones and other external sensors. Among the new functions necessary will be disc encryption that is operating system and network independent. The data needs to be protected because the entry points and weak links are pervasive. Systems will need the ability for remote, secure boot. There will be other "invisible" components in the IT systems that will handle other functions that are currently performed in a processor, such as virus detection, encryption, credential management, and user and system authentication. Other changes in the market include an increased demand for sensors. These devices will be small, but will include power supply, computing capabilities, the sensor function, and communications capabilities. The sensors will be able to self-assemble into a network configuration. Each node will be a sensor and also a communications link, so long distance data traffic can traverse the sensor space. IT applications with sensor networks will need to have location services and location data. You need to know where something is to use it. By definition, the sensors have to be auto-locating. In addition, the software needs to have event-driven state information automatically transferred across servers to storage. This data requires event monitoring and event notification capabilities that transcend server outages. In the data communications areas, wireless data on WiMax will be a major challenge. The nature of the medium makes signals and noise a problem due to intermediate distance symbol interference. The noise from one node can make bit signals in a multi-bit modulation scheme unintelligible. This interference is a function of the number of channels and number of bits per symbol and is exacerbated by high volumes of users is a local area. Possible solutions to the interference problems are to separate frequencies, time multiplex, reduce channel bit rates, or allow "dead" spots in the network. Some other technologies are emerging, but their costs may be prohibitive. Ultra-wide band networks will be used for short-range communications and are not expected to reuse frequencies. The applications will be in wireless USB or Bluetooth 3 applications. One problem is the two competing proposals for a standard - Freescale's is simpler, while the other from Intel and TI is more complex with higher power consumption, but possibly more robust. The marketplace will eventually decide. Consumer electronics (CE) is an area of interest. The PC makers and their sales channels have done poorly in consumer electronics, partly because the PC focus is on digital and is dominated by Intel, Microsoft, and Dell. The CE areas have high brand identities and vertical standards along with strong analog focus. The principle players in the market are Sony, Thompson, and Matsushita. Despite the differences in focus, the growth areas in CE are the digitally dominated TVs like HDTV and digital camcorders. Another high growth area is in smart appliances that take advantage of multitude of networked sensors and available compute power to enable and enhance functions. Medium growth areas are DVDs, games, digital audio players, and digital still cameras. Declining markets include CD players, analog TV, VCR, analog camcorders, and mini-disk players. Break out products Once in a while a category breaks out. The iPod has been very hard to beat because of branding issues and attractive design features. The product addresses a passive user rather than trying to be fully interactive. HDTV is currently starting to take off, but Microsoft has shown itself to be too late in entering the market with a usable set of capabilities. The company's digital rights management is perceived to be intrusive and user-hostile. Cell phones will have more functions and features in their next versions. The enhanced capabilities of the new phones are limited only by imagination, and will include various radio, music, photos, GPS, tag scanning, and will become a major part of the sensor network. The market will be about 1 billion units in 2006, while third-world countries are bypassing the wired infrastructure for cell towers. Even in the U.S., consumers are converting from wired to wireless phones, much to the dismay of the phone companies. Depending on the nature and extent of changes in core technologies, the hot areas are in zero management, high automation networks with hypervisors and hardware convergence, ultra wideband data communications, self-assembling system technologies, RFID and smart cards. Consumer wireless will grow to $200 billion per year and robust computing comprised of unreliable components will become widespread. The next largest product categories will be super smart cell phones, networked consumer devices, very smart appliances, modular servers, and sensor networks. ******************************** November 28, 2005 Long time EDA industry observer Tets Maniwa can be reached at maniwa_at_sbcglobal.net
|