- Worldwide Production of Semiconductors
- Worldwide Consumption of Semiconductors
- Military Electronics
- The Business of Making Semiconductors
This chapter is from the book
This chapter is from the book
This chapter is from the book
Worldwide Consumption of Semiconductors
Where do all these chips, laser diodes, and sensors go? They end up in quite a few places, as you might imagine, and some that you probably can't. Semiconductors are used in many different kinds of systems, from the obvious ones like computers and televisions, to the obscure ones like dog collars, telephone wiring, and greeting cards. Here we'll take a look at some of the major categories, as shown in Figure 5.8.
)
Figure 5.8 Relative consumption of semiconductors by industry. Courtesy
of In-Stat. Used with permission.
Computers and PCs
First and foremost among consumers of semiconductors are computers. Computers are almost all semiconductors, with a little metal or plastic wrapped around the outside. A 30-pound bag of electronics would not be an inappropriate description of the average PC. Because we're measuring consumption by revenue, not units, PCs are even more heavily represented in the total because of their expensive main microprocessor. PCs also have a lot of memory chips, the second-most expensive class of semiconductors. All this makes PC sales a good proxy for semiconductor sales, at least in that segment.
Processor Pricing Battles
For years, AMD and Intel have battled over the PC-processor business. Intel has always commanded the lion's share of the market, yet AMD's chips are sometimes considered technically superior and are almost always cheaper. How can AMD build its chips so inexpensively, and why don't customers flock to the less expensive processors?
To answer the first question, it's important to understand that price and cost are not the same thing. They're only loosely related. Both Intel and AMD sell their PC processors for far more than they cost to make. Like perfume and luxury cars, the cost of the materials and labor is almost irrelevant. The price is set by market conditions, not by cost overhead.
The second question hinges on the small part the processor plays in the total PC price structure. Mainstream PCs generally sell for around $800 and laptop computers can easily be priced at double that amount. The cost of the microprocessor is only a small part of the cost of building a whole PC. A suicidal competitor could give its chips away for free, but the complete PC would only be about $35 less expensive, not enough to sway most customers.
Enter brand-name marketing. It's fair to say that most people buying a PC don't have the faintest idea of how they work or what all the specifications mean. This is the perfect environment for marketing tactics to flower. Years of "Intel Inside" ad campaigns never mentioned anything about the chip's technical features. They were pure brand-awareness ploys, imprinting a particular brand name on a largely nae public. Microprocessors are now sold like perfume.
It also explains why PCs are so price competitive. Virtually no PC makers also make chips. (IBM is one of the exceptions.) That means PC makers have to buy their chips from the same semiconductor vendors that are also supplying their competitors. Volume discounts are about the only concession PC makers can wring out of chip makers. With little value to add other than the color of the plastic box, PC makers fight for every penny, which depends on keeping their volumes up. The advantage to consumers is that prices keep going down as chips get cheaper and PC makers struggle to remain competitive.
Communications and Networking
After PCs, communications equipment is the biggest consumer of electronics. For our purposes, communications means telephones and telephone equipment, computer networking, cellular phones and their infrastructure, and anything to do with satellites, television, and radio transmissions (although not the TVs and radios themselves). This segment has been growing rapidly for a number of years. The actual percentage you hear depends on whom you ask and when you ask them. It dips and rises with market conditions, of course, as telephone companies and network companies first invest, then retrench, depending on regulatory and market conditions.
Networking equipment consumes high-end microprocessors and DSP chips, lots and lots of memory, and special-purpose communications chips developed especially for one or two customers. Network and telephone companies also buy lots of laser diodes and optical sensors for their fiber-optic networks. Cellular telephone makers consume vast quantities of DSP and microcontroller chips—there's typically one of each in every cell phone—as well as mountains of tiny RF components such as resistors, capacitors, and inductors.
Consumer Electronics
Consumer electronics gives communications systems a good run for the money, consuming about one-fifth of all the world's semiconductor value. Consumer items can be televisions, DVD players, electronic toys, and also "white goods" such as refrigerators, washers, and dryers, all of which now include microcontrollers to mange power consumption and add exotic features.
Running Rings Around Sega Saturn
In the late 1990s one of the most popular home video game consoles was Sega's Saturn. The Saturn was an extraordinary system in many ways. The system was so advanced and high-end that, ironically, it led to Saturn's collapse and Sega's eventual withdrawal from the hardware market.
Saturn had no fewer than four 32-bit microprocessors, three from Hitachi and one from Motorola. These were partnered with six custom-designed ASIC chips and several megabytes of memory. The entire system was considerably more complex than the average PC of its day. It was so complex, in fact, that most game programmers couldn't exploit its features well. Under pressure to meet deadlines, most game programmers took shortcuts and used only one or two of Saturn's four processors. Although many Saturn games were good, few flexed its considerable hardware muscle. After it cratered, Saturn left a vacancy for Microsoft's Xbox and Sony's PlayStation 2.
Home video games are a big consumer of electronics in the home. Nintendo, Sony, and Microsoft (and Sega, Atari, and Commodore before them) have all created very high-end computers that sell for very little money. In fact, these companies sell their game consoles at a loss. A new PlayStation 2, for example, might cost Sony $350 or more to manufacture, yet sells for $200 to $300 when new. Sony makes up the money on game (software) sales. Unlike PC software, PlayStation software must be officially licensed and "approved" by Sony, and royalties apply. (The same is true for other game consoles.) In this razor-and-blade business model, the game console is merely an enticement for consumers to buy games. Each game brings in a royalty of a few dollars to the maker of the console, in this case, Sony. Over the life of the console, Sony will make more than enough money from software royalties to offset the cost of effectively taping a $100 bill to every PlayStation.
The problems with this business model are obvious, yet the concept itself is a very old one. If game players don't buy enough games, the game maker loses money. The break-even point for most video game systems comes after consumers buy three to five titles. That makes it a safe bet, as statistics show that most video game owners buy more than a dozen games over the useful life of the system.
The second problem is that of deferred revenue versus instant gratification. The game maker must spend the money up front to manufacture and market millions of game consoles, generating a huge financial loss. Only after several months have passed will software royalties begin to make up these losses. Companies must have deep pockets, or very patient investors, to enter the video game market.
Finally, the entire scheme hinges on licensed software. There must be no "shareware" game titles, no pirated or copied games, and no independent or unlicensed game developers. In short, it needs to be the exact opposite of the PC software industry. To prevent this, game consoles include obscure and undocumented hardware features that independent programmers are unlikely to figure out. Officially approved and licensed programmers, however, are taught the secrets of the system in return for a licensing fee and a promise to pay royalties on every game they sell. In some cases (e.g., Sony's original PlayStation), the game CDs themselves are mastered and duplicated by the game manufacturer, which brings in additional revenue and helps control inventory. As a last resort, video game manufacturers can exert legal pressure on unlicensed programmers producing "rogue" software that doesn't generate royalties.
Industrial Electronics
The industrial uses of semiconductors are many and varied. Chips show up in robots, vision-inspection systems, alarms and security systems, and power generators, to name but a few. Large, expensive, high-powered semiconductors are used in dams, nuclear plants, and oil plants to regulate and control the electricity these plants generate.
Robots are full of electronics, of course. Heavy industrial robots have a half-dozen motors to move their joints, and each motor is usually controlled by its own miniature computer. Then there's one main computer (the robot's "head") that controls all of these. Heavy robots are amazing for their ability to pick up and move heavy loads, then set them down accurately to within fractions of an inch. That kind of accuracy calls for some exotic mathematics, called kinematics, to predict how and when the robot arm needs to speed up and slow down. This is all handled by low-cost microprocessor chips, along with dozens of memory chips, communications chips (for talking to other robots on the assembly line), and high-voltage chips to power the whole thing.
Robots with vision systems combine CCD image sensors with more miniature computers to analyze what they see. Some robots are nothing but vision systems, with no moving arms at all. Either way, these electronic eyeballs can look at parts moving by on a conveyor belt and instantly recognize any flawed or damaged ones. Other robots can then throw the bad pieces into the trash. Robots can also sort and straighten scattered parts so they're all turned the same way in nice, neat rows, making the job easier for the next robot down the assembly line. Robots with vision can assemble anything from vacuum cleaners to chocolates. A popular brand of sandwich cookie is made by robots that deliberately assemble the two halves slightly off-center, to make the cookies look handmade.
Automotive Electronics
The semiconductor content of automobiles has been growing steadily for years and shows little sign of abating. The average new automobile now carries about $200 worth of electronics, including almost a dozen microprocessors or microcontrollers. Large, late-model luxury cars can have well over 60 microprocessors. Some electronics are added for safety (collision-avoidance detectors, airbags, night vision), some for comfort or entertainment (in-dash CD players, electrically adjustable seats, air conditioning), and some to just run the car, replacing older mechanical designs (electronic ignition, antilock brakes, or automatic transmission). That doesn't even include the bits that actually look like computers, such as satellite-guided navigation systems and rear-seat movie players.
The electronic systems in cars are starting to communicate and interact in unusual ways. For example, in some cars the electronics controlling the position of the side-view mirrors communicates with the electronics in the automatic transmission. Why? So the mirrors will automatically tilt down and inward whenever you put the car into reverse, the better for you to see the rear of your car while backing up.
Other cars connect the in-dash radio or CD player to the antilock brakes. This seemingly bizarre combination allows the radio to adjust its volume automatically to compensate for road speed. (The antilock brake system has the best gauge of current road speed.) Cars fitted with satellite navigation systems and cellular telephones often connect these two systems together with the airbag controller. The purpose is to detect whether the car is involved in an accident serious enough to deploy the airbags. If so, the car automatically phones for emergency services and, using the satellite navigation system, transmits the exact location of the accident. Often this system is tied to a fourth system, unlocking the car doors automatically.