Electronic systems and components have been in a constant state of evolution for nearly 50 years. Moore’s Law—the doubling of transistor density every two years—started it all in 1965, and the trend is now expected to continue through 2015 and beyond. Current hardware technology development responds to growth in cloud computing, Internet servers, mobile computing, pervasive wireless, embedded everything, integrated power supplies, satellite-based communications, flexible circuits and displays, many-core processors, carbon nanotube circuits, printed circuits and more. Tomorrow’s technologies are likely to involve human implanted sensors, controllers, displays, and microprocessors. Imaging sensors have already been implanted in the human retina, allowing blind people to see. The bio-electronics age is just beginning.
Electronics manufacturing, design, and research was one of the first high-tech industries to be fully globalized. The continual downsizing and miniaturization of integrated circuits within electronic applications has created complex manufacturing and testing requirements that are more expensive than any other volume manufacturing system on the planet. A new manufacturing plant for state-of-the-art microprocessors can cost $5 billion in the U.S. or $3 billion in a lower cost Asian site, which makes offshore manufacturing by U.S. companies very attractive, even with added support requirements. As a result, substantial infrastructures have been established in many countries for even the most complex procedures.
The complexity of electronics manufacturing requires that a substantial amount of the company’s R&D be dedicated to development of the manufacturing procedures and often spent in and around the local manufacturing site. With 80% market share, Intel has become the largest manufacturer of microprocessors in the world. It recently opened a $1 billion chip assembly and testing facility in Vietnam and within two weeks opened a $2.5 billion semiconductor manufacturing plant in Dalian, China. The Vietnam facility is Intel’s seventh assembly and test site. Others are located in Penang and Kulim, Malaysia; Cavite, the Philippines; Chengdu and Shanghai, China; and San Jose, Costa Rica. Intel’s chip plant in Dalian uses 300-mm wafers, which cuts chip manufacturing costs by 30% over the previous generation 200-mm wafers. But of course, the equipment for 300-mm manufacture is considerably more expensive than that for 200-mm wafers.
Intel’s continuing R&D and large-scale manufacturing capabilities allow it to push manufacturing efficiencies and keep its competitors at a distance due to the expensive capital equipment and entry costs. Intel has not ignored its U.S. manufacturing base and recently announced investments of $8 billion to build a new plant in Oregon and upgrade four existing plants in Arizona and Oregon.
Intel has also used its strong market position in microprocessor manufacturing to open doors for other associated ventures. The company recently partnered with the government of Taiwan to set up a multimillion dollar Internet computing research lab. Intel announced that it would work with Taiwan’s National Science Council and a leading Taiwan university to establish a cloud computing research center at an estimated cost of $24 million over the next three to five years.
Advanced Micro Devices (AMD) has been attempting to challenge Intel for market share for a long time, but with 20% market share, its smaller size puts it at a disadvantage. Because Intel has more money to spend on R&D (nearly five times that of AMD for 2010), it can often develop and manufacture the next generation of more complex, smaller and more powerefficient chips ahead of AMD. Being first to market means a much larger initial and even later market share for the next generation of computing systems.
AMD has not conceded the microprocessor market to Intel. It acquired ATI four years ago for $5.4 billion and has been using its R&D to develop an accelerated processing unit (APU) that combines the central processing unit (CPU) with a graphical processing unit (GPU) into one chip of ‘super-silicon’. AMD’s first Fusion chips were introduced this fall, ahead of the Q1 2011 target date. This device is targeted at mainstream notebooks and desktop computers. It has twice the gaming performance of existing microprocessors and graphics cards/chips at half the power requirements. The ultra-small form factor also gives it 10 times the graphics performance over existing net-book computers. Production scale-up will allow the first products to appear in early 2011. To unload itself from the massive maintenance and capital requirements, AMD spun off its chip fabrication facilities into a separate company, Global Foundries, Inc., in 2009.
Agile R&D as a Strategic Advantage
While Intel and AMD spend their R&D investments developing new microprocessor devices and the processes for creating them, Apple, Inc., invests in developing products that use mostly existing components. The systems they’ve developed over the past several years—the iPod, iPad, and iPhone, along with their line of Mac portable computers—have established whole new product lines and created demand where none existed before. This R&D strategy has fewer risks than the capital-intensive Intel and AMD have, but is much more prone to competition. Apple’s response is to use the speed of R&D as a competitive advantage. It continuously develops upgrades that are announced on short time scales and that keep competitors one step behind. The tremendous success of this approach has allowed Apple to hire thousands of employees to support their operations (12,600 new hires over the past year) and to increase their R&D investments (up 34% or nearly $500 million over the past year) and development programs. A substantial part of this R&D investment is also targeted at software development and upgrades for each of its product lines.
It’s been noted that the payback on Apple’s R&D investments is substantially better than those for Intel, Cisco or even Microsoft. Apple, for example, spent $4.6 billion on R&D over the past four years. To the extent that revenue growth is an indicator of return on R&D, Apple’s concomitant revenue increased from $25 to $43 billion. Over the same period, Microsoft spent $31 billion on R&D, and its revenue only rose from $44 to $58 billion. Cisco spent $19 billion on R&D, while its revenue grew from $28 to $36 billion. Intel spent $23 billion on R&D, with flat revenues over a four year period at $35 billion/year.
Targeting Product Cost
Qualcomm is another chip manufacturer that focuses its R&D investments on mobile chipsets for cell phones. It spends heavily on reducing chip costs since its applications are cost-sensitive cell phone suppliers. Qualcomm also recently announced that it was establishing an R&D center in Taiwan to help it tap into China’s growing market for cell phones. The company is also creating a new data exchange format that could help it gain an edge for supporting 3D chips.
Dell Computer, which built its business without significant R&D investments, has decided to increase R&D spending to develop higher margin computer servers, data storage, networking gear and technology services. The margins on desktop and notebook devices have recently been shrinking and are not expected to return to previous levels.
In some situations, chip-fabrication technologies have become too complex and expensive for even the largest manufacturers. Over the past decade, these situations have required the formation of pre-competitive collaborations. In October 2010, Intel, Toshiba and Samsung announced they were collaborating to develop devices with 10-nm semiconductor feature sizes by 2016. In this initiative, Japan’s Ministry of Economy, Trade, and Industry is providing more than $60 million in initial funding for the R&D efforts. Toshiba and Samsung are expected to use the technology to make 10-nm flash memory chips, while Intel will make faster microprocessors that use less power. In a similar situation, IBM, Samsung, GlobalFoundries, and STMicroelectronics announced in June 2010 that they were configuring their manufacturing facilities for the production of devices with 28-nm processors developed by the four and Toshiba, Infineon and Renesas Electronics.