The market for compound semiconductors has more than doubled in two years to be worth $18 billion. Tom Hausken of market research firm Strategies Unlimited explains why.

Components market reaches $18 billion The market for components based on compound semiconductors reached nearly $18 billion in 2000 more than doubling in just two years. Although the industry is still dwarfed by the market for components based on silicon, this is a remarkable increase. And while growth is slowing down from these extraordinary levels, sales in 2001 are expected to show a strong increase. presents the total market for compound semiconductor components for the last seven years. The overall market grew nearly 60% from 1999 to 2000. The market is still dominated by optoelectronics 2000 sales hit $14.2 billion, a growth of 67% compared with 1999. Electronics represents $3.6 billion in sales, a growth of 34% since 1999. Demand in the communications sector, both wireless and fiber optic, led to most of the growth last year (see ). Communications applications amounted to more than 60% of the market for compound semiconductor products. The market for the visible LEDs used in lamps and displays is also growing rapidly, making up more than 16% of the total. The remaining applications include laser diodes for optical storage (DVDs and CDs), office machines, industrial, medical, aerospace and solar panels. Optoelectronics presents a breakdown of the sales in 2000 of optoelectronic components by the type of device. This excludes image sensors and detectors based on silicon rather than compound semiconductors. The market is a mixture of high-volume and low-volume products.On the high-volume and low-price end are lasers for optical storage and visible LEDs. On the relatively low-volume but high-priced end are telecom modules. Telecom The telecom components market (lasers and detectors) had the most prominence in 2000, growing more than 160% to $6.3 billion. While the overall unit sales of telecom products are still lower than in other sectors, the unit sales of some products have swelled by orders of magnitude. And although prices of individual products decline rapidly, innovation happens so quickly that the mix of products rapidly evolves to higher performance products that bring high prices. One of these developments is the phenomenal success in 2000 of 10 Gbit/s transmission, leading to surging sales of expensive components for transmitters and receivers. An emerging development is wavelength tunable lasers that enable flexible provisioning of wavelengths in all-optical networks. Another key development is the integration of the optoelectronics with ever more electronics and passive optical devices. This integration removes a design and manufacturing burden from the customer, and reduces the physical footprint of the overall ensemble of devices. As the industry migrates to the use of these modules over discrete components, the selling price rises with the value of the module, even as the prices of the individual sub-components fall dramatically. Datacom The market for so-called datacom components grew 31% to $1.1 billion in 2000. The datacom market differs substantially from the telecom market even though both are used for data communications. The datacom market is one of transceivers chiefly designed for Ethernet or Fibre Channel protocols for short-range links, usually at short wavelengths (850 nm). The transceiver product, while now evolving to data rates as high as 10 Gbit/s, represents little integration with the surrounding elements on the board. The end user is an enterprise rather than a carrier, and the all-round emphasis is much more on cost than performance. The mix of datacom products is migrating toward higher value mainly due to increasing data rates but overall sales are more subject to falling prices. Consequently, while datacom transceivers represent higher unit sales than pairs of telecom sources and detectors, revenue and growth in revenue are less in the datacom market. Datacom suppliers have often threatened to offer a "disruptive technology" to the telecom sector by introducing low-cost manufacturing principles into the high-end communications business. The convergence is not simple: greater aggregation of data, longer distances and carriers demands for very high reliability drive the performance requirements on each component to the very limit of today s technology. Put another way, telecom products must exhibit virtuosity in sheer performance, while datacom products must exhibit virtuosity in manufacturing. What is now happening is that the two sectors are naturally migrating closer using principles borrowed from each other. For example, datacom suppliers are looking toward long wavelengths and even WDM, while telecom suppliers now have the sales volumes to implement lower-cost manufacturing methods. It is becoming impossible to identify some products as fundamentally datacom or telecom. This development is not so much disruptive as it is a matter of the industry maturing to reach the next stage of volume production. Visible LEDs The visible LED market grew 24% in 2000 to $2.9 billion for applications in lamps and displays. The strongest contribution is from high-brightness LEDs, which now form a rapidly growing market, with such emitters appearing in car tail, turn and brake lights, exit signs and green and red traffic signals. Large-screen displays, such as the one used at Times Square in New York City, are becoming more prominent. These displays offer brilliant color resolution even in bright daylight. High-brightness LEDs are also seeing strong growth from applications as backlights for the liquid-crystal displays (or LCDs) in cellular phones and car stereos. Other lasers Laser diodes used for non-communication applications grew 60% to $1.5 billion in 2000. These are used for a wide variety of applications including optical storage (such as various types of DVDs and CDs), laser printers, barcode scanners, laser pointers, materials processing and instrumentation. Infrared LEDs The infrared LED business grew a healthy 22% in 2000 to $1.4 billion. Infrared LEDs are used in remote-control units, rotary encoders for printers, computer mice, photointerrupters and a wide variety of other applications. This is a mature business but has large growth for reasons that are not totally clear. Optocouplers Sales of optocouplers also grew at 22% to reach $960 million in 2000. These devices are used to electrically isolate two circuits from each other. The market is characterized by continuing pricing pressure and volume production. There is a persistent movement toward surface-mount packages and simpler configurations in 4-pin packages. Optically coupled solid-state relays (OCSSRs) are increasingly replacing electromechanical relays for a variety of applications in the areas of communications, modems and industrial instrumentation. Solar cells Sales of compound semiconductor solar cells were practically flat in 2000. These solar cells are mainly used in satellite energy panels for their efficiency and durability (due to higher radiation hardness) compared with silicon solar cells. This business grew rapidly from the mid-1990s and, after dropping off with the brief lull in satellite deployments, is now expected to grow modestly again in the coming years. Electronics Compound semiconductor electronics is still almost entirely based on GaAs, although there is emerging business in InP, SiGe, SiC and GaN. At $3.6 billion, and expected to grow 19% to $4.3 billion in 2001, this market is holding its own by filling niches that silicon cannot, although silicon is still pushing its performance to ever higher data rates. presents a breakdown of the market for compound semiconductor circuits by application. Communications dominates the market, with a 91% share in 2000. Of that, 69% is for wireless handsets and 18% is for the circuits used to support fiber-optic transmission. Emerging applications in communications include wireless LANs and fixed wireless radios for broadband and Internet access. Other markets for compound semiconductor electronics include cable and satellite TV terminals and modems, together with computing, industrial and military/aerospace applications. shows the forecast for worldwide cellular handset shipments. Each handset uses multiple GaAs ICs and field-effect transistors (FETs) as power amplifiers, receivers and RF switches. Sales in 2000 were up 43% on 1999 s figure to 400 million units. China and other non-US markets offer significant growth for handset sales, as does the commercialization of new Internet-enabled 2.5G and 3G mobile phones. For fiber-optic systems, compound semiconductor ICs are used as laser and modulator drivers, receivers, serializerdeserializers (muxdemux) and crossbar switches. These circuits must be able to operate just beyond the nominal telecom line rates of 2.5, 10 and, soon, 40 Gbit/s. At 10 and especially 40 Gbit/s, device and package design enters a different realm, one more familiar to engineers who have worked in the military industry than to conventional circuit designers. While unit sales in this market will remain orders of magnitude below those for wireless handsets, the value and growth rates are both very high. It appears that heterostructure bipolar transistors (HBTs) based on InP will be the choice technology for new 40 Gbit/s telecom systems. InP HBTs outperform GaAs HBTs, but at the moment few suppliers are prepared to offer products at this data rate. presents a breakdown of the compound semiconductor electronics market by device type. While analog ICs for the wireless industry remain the dominant device type at 66% of the market, the market for digital ICs is growing rapidly. Sales of discrete FETs are relatively static, and are expected to decline slightly in the years ahead. Transistors based on epitaxial substrates now generate more revenue than those that are based on ion-implanted substrates. HBTs, pseudomorphic high-electron mobility transistors (PHEMTs) and epitaxial MESFETs account together for about 54% of shipments, with ion-implanted MESFETs and diodes making up the remainder. The price of epitaxial wafers has been declining. They can now provide higher performance or lower voltage operation or both where they are needed more cost effectively. At data rates below 2 GHz, however, silicon-based ICs are now penetrating applications where ion-implanted GaAs circuits have previously been considered superior. Device convergence in modules The major GaAs fabs are moving to 6 inch wafers to increase the number of dice per wafer, thereby reducing chip cost. Chip sizes are likely to remain about the same over the next few years, but packaged devices will incorporate more functions by integrating GaAs, silicon and passive elements within multi-chip modules (MCMs). This move to hybrid integration mirrors what is occurring in optoelectronics for telecom applications. For years, optoelectronics researchers aimed for the holy grail: devices that monolithically integrated optoelectronic sources and detectors with electronic circuits onto a single compound semiconductor chip. Such optoelectronic integrated circuits (OEICs) remain impractical for commercially successful products. Today the industry employs hybrid integration: optimizing manufacturing processes for each type of device then integrating the chips on a common platform. Such an integration of dissimilar sub-components, each optimized for best performance at lowest cost, has proven more effective than seeking monolithic integration onto a single chip. While customers will demand more of these compact and integrated modules, the nature of hybrid integration is that the sub-components will continue to be manufactured separately for years to come.