Optoelectronics market sales to triple by the next decade

Optoelectronics market sales to triple by the next decade

Expanding at 20% a year into the turn of the century, sales of fiber-optic optoelectronic components are anticipated to jump sharply from $1.2 billion in 1996 to $3.8 billion by 2001

Stephen Montgomery

electronicast Corp.

North American sales of fiber-optic optoelectronic components in 1996 are estimated at $1.2 billion. This market is expected to increase strongly at about 20% per year over the next five years, and to reach $3.8 billion in the year 2001. The rapidly rising quantitative growth, however, is projected to be substantially offset by falling average prices. Nevertheless, over the years 2001 to 2006, optoelectronic component sales are projected to nearly double to $7.0 billion.

North American production of optoelectronic components is predicted to expand from $1.2 billion to $6.5 billion from 1996 to 2006. The difference between sales and production--the trade balance--is foreseen to shrink as more North American producers establish foreign-based production facilities to serve this market, and foreign producers, in turn, become more competitive.

Optoelectronic products are being produced and sold at the chip, submount, packaged device, and complete transmitter and receiver levels. The values in this market study encompass packaged emitters and detectors, plus all intermediate circuits and other devices required to achieve the complete transmitter and receiver functions, including assembly and test costs.

Many equipment producers generally buy complete functional transmitters and receivers. Others prefer to purchase packaged emitters and detectors and then add integrated circuits from other suppliers, either in a single final package or distributed among other devices on one or more signal-processing boards.

Telecommunications networks ac counted for 65% of fiber-optic optoelectronics sales, or $0.78 billion, in 1996 (see table). This telecommunications market share is projected to decline markedly to 42% by 2001 due to the inroads forecasted for premises network applications. Use of such networks is expected to jump because of the accelerating deployment of fiber-in-the-loop and the continuing demand for network upgrades to accommodate rapidly increasing bandwidth requirements.

Market growth

The fiber-optic optoelectronics market for the years 1996 to 2001 divides into two major segments: electronic equipment, as shipped to customers; and optoelectronics purchased by customers to increase or upgrade the performance of installed equipment.

For example, Synchronous Optical Network (Sonet) and Asynchronous Transfer Mode (ATM) switches, hubs, concentrators, and similar electronic equipment is typically shipped to customers with empty slots in the interface rack. When network needs ex pand, network planners and providers can add system capacity at moderate cost by installing input/output boards into the vacant rack slots, by removing initial low-data-rate (such as T1 or T3) boards, and by substituting higher-data-rate optoelectronic boards. Consequently, the upgrade or growth optoelectronics segment represents a major share of the total market.

Many local area network suppliers design their equipment to accommodate certain network interface cards or boards, but not every possible type. These boards handle token ring, Ethernet, ATM, or sub-ATM network standards. Proprietary interfaces and telecommunication interfaces, such as DS-1, DS-3, and isdn, are less common. However, new interface boards are expected to proliferate as the market expands.

From 1996 to 2001, growth of the optoelectronics market due to network upgrades is expected to rise moderately because of the rapid expansion of new networks (ATM, Gigabit Ethernet, and Fast Ethernet, plus Fibre Channel interconnect). All these networks will initially have excess capacity. By the late 1990s, however, the loading of these systems is gauged to approach capacity. The sales value of upgrade network interface cards is then projected to accelerate. The upgrade of optoelectronic boards usually occurs in one of three ways:

Equipment shipped with empty board slots comes equipped with plug-in card-edge connectors wired into its central system. The system is then expanded by filling these empty slots with upgrade boards.

Existing optoelectronic boards, such as sub-ATM interface types operating at 25 Mbits/sec, are replaced by higher-data-rate boards, such as OC-3 (155-Mbit/sec) types.

Network electronic/copper interface boards, such as those for Ethernet and token ring, are replaced by optoelectronic boards, coincident with upgrading other support equipment.

Laser diodes dominate

Historically, the fiber-optic optoelectronics market has been dominated by laser-diode-based transmitter/receiver sets. This dominance prevailed because the early deployment of fiber optics was dominated by long-distance, high-data-rate installations whose performance requirements could not be met by light-emitting diodes (LEDs). In these installations, the optoelectronics cost reflected only a small part of the total system cost; therefore, little market pressure evolved to push for optoelectronics cost reduction.

Presently, the network trend is toward greater use of short cable lengths. Many cables are being operated at modest data rates, especially in premises networks. Correspondingly, in these systems, cable interconnections represent a substantial share of the total installed system cost. In these applications, fiber-optic cables are competing with twisted-pair copper wires and coaxial cables for market share. The higher cost of fiber-optic links has been the main impediment to fiber penetration of this market.

LEDs over the past decade have achieved high production volumes and, subsequently, low chip cost. Long-wavelength laser diodes, conversely, have been produced in smaller quantities and continually evolving designs, resulting in unit costs orders of magnitude greater than LEDs.

In the future, LEDs are expected to gain a substantial share of the low-to-moderate data-rate range (up to OC-12 or 622 Mbits/sec), which represents the large quantity market. The performance of LEDs also is being aggressively improved; for example, these diodes are now available in long wavelengths for use in low-loss fiber.

Conversely, laser-diode cost is decreasing as production quantities increase by orders of magnitude, and as performance requirements are relaxed for short-distance applications. In addition, the move of the network application market to massively parallel optical interconnects for inter- and intra-enclosure applications is expected to drive down the average price of transmitter/receiver pairs.

Over the next decade, the average price of both LED-based and laser-diode-based optoelectronic components is predicted to drop by 10% to 15% per year, and more in some applications. Historically, LEDs have been used in low-data-rate applications; however, they have been improved over time to handle 20-Mbit/sec rates in token ring and Ethernet networks. Laser diodes, on the other hand, have been applied to the telecommunications digital transport market, where data-rate application demands have escalated, from 45 Mbits/sec in 1974 to 10 Gbits/sec in 1994.

LEDs, however, were inexpensive, costing dollars to tens of dollars. Conversely, laser diodes were expensive, priced at hundreds to thousands of dollars. Compared to applications in local area networks, optoelectronics has achieved a lower value-added share in telecommunications networks; therefore, the price pressure has not been severe. With the coming move to fiber-to-the-curb networks, however, market ratios are expected to change markedly.

Over the past five years, as the use of Fiber Distributed Data Interface networks expanded and other potential markets came into view, LED performance has been markedly improved by manufacturers such as Hewlett-Packard Co. For example, LED transmitter/receivers suitable for ATM OC-3 applications are now commercially available at attractive prices.

Meanwhile, laser-diode suppliers are focusing heavily on cost reduction, especially for 155-Mbit/sec to 1-Gbit/ sec networks. The costs of laser diodes and LEDs are both quantity-sensitive. In production, LEDs have achieved high volumes (millions per month)--a major factor in their low price compared to laser diodes. The use of compact-disk-type lasers in such applications as Fibre Channel is expected to boost laser-diode quantities, and subsequently, push prices down. Fiber-in-the-loop networks are foreseen as an expanding market for both laser-diode and LED-based optoelectronics.

Massively parallel optoelectronics

Another factor that is anticipated to drive down the average price of transmitters and receivers, both LED- and laser-diode based, is the movement to massively parallel transport applications. These networks use fiber ribbon cables, such as those containing 32 or 64 fibers, where each fiber interfaces to a diode in an array transmitter/receiver. The commonality of die fabrication, packaging, and signal processing is projected to achieve major cost reductions in transmitters and receivers, as well as size, compactness, and packaging advantages.

Optoelectronics technology is moving rapidly into a higher level of product integration for two massively parallel transport network types:

links assembled from discrete emitters and detectors, which are individually packaged (in these links, four or more waveguide channels are used per transmitter/receiver set),

links assembled by incorporating line array emitters and detectors, which are integrated into the same module.

Massively parallel optical interconnect components comprise optoelectronics, optical waveguides, and connectors. In application, they might include an array of four or more transmitters, receivers, or both components located at one endpoint, which is connected to an equivalent array, via optical waveguide, to the other endpoint. They are also used within a single network, system, or piece of equipment.

Massively parallel optoelectronics includes both synchronous and asynchronous links. However, its value lies in new applications (markets), as opposed to conventional premises wiring and local-exchange, outside-plant interconnect applications.

Applications defined as massively parallel are confined to use inside an equipment enclosure or between units of a multi-enclosure system such as a telecommunications crossconnect switch made up of 10 to 25 racks, clustered workstations, or a workstation-memory combination.

In 1994, massively parallel optoelectronics found widespread usage in crossconnect switch intra-enclosures, such as the Tellabs Titan 5500. Other digital Sonet-capable crossconnect switches produced by Alcatel, DSC Communications, and Lucent Technologies from 1994 to 1996 were major users of intra-equipment fiber-optic links at 622 Mbit/sec and 1-Gbit/sec rates. However, these optoelectronic devices were emplaced in a fanout rather than an array/modular format.

Another rapidly emerging optoelectronics niche is optical-fiber amplifiers. Developed for underwater use, optical amplifiers are widely used by both landline telephone companies and cable-TV operators. To date, telephone companies have been the predominant users of optical amplifiers. The major North American optical amplifier suppliers include Lucent Technologies, Pirelli, Hewlett-Packard, Sumitomo, and Corning. u

Stephen Montgomery is vice president and chief operations officer at ElectroniCast Corp. in San Mateo, CA.