The continuing combination or "integrating" of multiple integrated circuit functions onto fewer and fewer and ultimately a single integrated circuit has been termed "creative destruction" by the electronics industry.
So what does that mean?
In the beginning of PCs, for example, it took several large chips to make a functioning computer. Intel (INTC), of course, made the original CPU (Central Processing Chip), but it took several other chips to make a workable system. It took a memory controller chip and a video controller chip and an I/O controller chip, and others. Some of these "Super" I/O chips were made by manufacturers other than Intel. Functions like very high performance graphics controller chips are still made by companies like Nvidia (NVDA) and AMD (AMD).
Slowly but surely, Intel began putting these functions on the CPU chips to the point where today the Intel Haswell CPU chip is basically a computer SoC (System on a Chip). When all of the functions of a Super I/O chip were on the Intel chip, the previous Super I/O chip manufacturer's business went to zero. It didn't decline like in other competitive business situations -- it went to zero, and very quickly. The Super I/O business was DESTROYED. In its place, Intel CREATED a more complex CPU chip, enabled by the continual process shrinks that we still discuss today. Because Intel sucked the Super I/O chip onto the CPU chip didn't mean that it was now free; oh, no -- Intel was legitimately entitled to the value of that Super I/O chip and it was reflected in higher CPU chip prices.
This creative destruction has been going on in the semiconductor business since Robert Noyce and Jack Kilby invented the Integrated circuit, and is still going on today. The world heavyweight champion of creative destruction is, of course, Intel. It is as violent a game as is played in business.
An ongoing example of C/D activity today is the three-way struggle going on over the function of the Graphic Processing Units or GPUs. AMD bought a GPU company, ATI, in order to diversify away from total dependence on competition with Intel. Nvidia was founded to design and build GPUs. And then there is Intel; with every new Intel CPU architecture comes a new and better embedded GPU section that is actually part of the CPU chip. With every new improvement by Intel, a certain percentage of AMD and NVDA GPU business goes away. ZNET thinks the GPU in the Haswell will lead to more OEMs leaving discrete GPU chips behind altogether in the interest of lower cost with sufficient performance. In the highest end, Haswell embedded DRAM improves performance enough to join discrete class GPU solutions.
Both companies fight fiercely to retain their GPU customers, but it is inevitable that Intel's GPU will soon get good enough to obsolete all discrete GPUs from AMD and NVDA. It's just a matter of time. Every time Intel improves the functionality of the GPU, it charges a little more money for those parts that have the improved GPU.
Another example of creative destruction happening today involves the FIVR or Fully Integrated Voltage Regulator found on the new Intel Haswell processor. The FIVR is a bunch of voltage regulators that have been sucked onto the processor chip, supported by a separate chip of inductors mounted next to the CPU chip. This FIVR is one of the things that makes the Haswell much more power efficient. The more important thing to shareholders is that these voltage regulators and power control circuits used to be parts that were mounted on the PC motherboard. The parts were made by companies like Texas Instruments (TXN), Linear Technology (LLTC) and other analog semiconductor companies. This business will go to zero for those companies as the conversion to Haswell is completed. The scale of this business is somewhere in the $10 per computer area. Since 300 million PCs will be built, it is a $3 billion piece of business that is DESTROYED for TXN et al, and about $10 more per chip that Intel can charge for its CPU chips.
That is how Intel will grow, even in a flat to down PC market.
In the mobile market, a future candidate for creative destruction will be the LTE radio function that will be required for all mobile phones worldwide.
In the beginning, cell phones were referred to as cellular radios because they were more radio than anything else. Even today the bill of material for an iPhone contains $39 for radio components, twice as much as for the Samsung (GM:SSNLF)-made application processor and only behind the $44 for the display and touchscreen. The main radio has been the domain of Qualcomm (QCOM), but the Intel D10 caterpillar is headed for QCOM and the radio business by way of Digital Radio. Intel has a real and active program to convert the largely analog LTE radio section to about a half billion tiny digital transistors (about five times the head of a pin) and suck it on to the Intel application processor. A year ago Intel demonstrated Rosepoint, a completely digital solution to WiFi and showed slides of the implementation of digital radio for mobile baseband and LTE. It's on the way. With it will come the destruction of Qualcomm, just as surely as those super I/O companies that used to exist.
Now that you know what creative destruction is, I have to ask some questions: Who in the world of semiconductors can stop this Intel process? Is there anyone who can suddenly suck a 1.6 billion transistor Haswell onto, say, a GPS chip? Or an ARM chip? Obviously, the answer is no. Therefore, we know how the game ends. Intel uses creative destruction to Hoover up any and all business at the leading edge of technology. It may not all happen overnight, but it will happen. The leaders of Qualcomm know it; the leaders of NVDA, AMD, Broadcom (BRCM), and Marvell (MRVL) all know it.
With regard to creative destruction, Intel is in a unique position. Some companies combine WiFi and Bluetooth and GPS into a single chip, but Intel does creative destruction on a grand scale. Eventually, even those functions will wind up on an Intel CPU or application processor and those companies will be gone.
You can take it to the bank.