Article

Déjà view

A recent study by Bain & Company states that the future of optical networking remains bright.

Par Scott Bender and Vince Tobkin

First published on octobre 24, 2001
min


Not long ago, a second-place technology company discovered a new product with the potential to revolutionize data handling. The company's decision to focus on this new technology fueled rapid earnings growth and propelled its stock price upward 230% in less than five months. The market greeted the new technology with seemingly limitless demand. Clamoring customers beefed up research and development budgets, which attracted a flood of new suppliers to the industry. Venture capitalists, eager to cash in, lined up to bankroll every entrepreneur promising the next breakthrough innovation. Customers rushed to acquire the newest technology, upgrade existing products and services, and thereby boost their own company valuations. The gold rush had begun. ...the future of optical networking remains bright. [There will be] growth in the optical components sector as the demand for bandwidth continues to increase and copper-based electrical networks are replaced with faster, more robust fiber optic systems. Then the mother lode petered out. An economic slowdown tempered demand, and the component maker's glittering prospects turned murky. Disillusioned investors scorned the innovator's bloated inventories, sagging stock, and lowered prospects. Business Week's prior proclamations of "boom times"[1] yielded to "a gloomier forecast"[2] for the sector. You're probably guessing that we're talking about optical-age star Nortel Networks and the roller coaster optical components industry. Guess again. This is the story of chip maker Intel, whose maniacal focus on semiconductors revolutionized computing in the early 1970s, drove the company to market leadership, then to a screeching halt—for a time. However, the story could have easily been that of today's optical components vendors Nortel and JDS Uniphase. The contours of today's crisis for optical component makers, and yesterday's for chipmakers, look close enough to constitute, perhaps, a crystal ball. If we peer into its depths, we may see that optical component makers are merely taking a historic ride, which could soon regain momentum. Current Uncertainty Investors and managers in the optical components sector (and the fiber optics industry in general) are searching for such a crystal ball. Former stock market darlings JDS Uniphase and Corning are both down over 90% from their previous 52-week highs[3]. Revenues for optical equipment providers like Lucent and Nortel have shrunk as telecom capital expenditures have plunged, with carriers seeking to optimize the performance of existing networks instead of building new ones. Yet, a recent study by Bain & Company states that the future of optical networking remains bright. Bain projects growth in the optical components sector as the demand for bandwidth continues to increase and copper-based electrical networks are replaced with faster, more robust fiber optic systems. Nonetheless, if history is our guide, today's setbacks will change the competitive landscape. Rocky times will restructure the industry, reset key purchase criteria and shake out weaker competitors. What Can History Teach Us? Following the invention of the transistor in 1947 and the integrated circuit (IC) in 1958, the semiconductor industry experienced an explosion in growth (see figure 1) similar to recent developments in the optical components industry. In both cases, improvements in their respective price/performance ratios accelerated. Gordon Moore, semiconductor industry pioneer and co-founder of Intel, observed in 1965 that the performance of ICs doubled every 12 months (later increased to every 18 months) at equivalent price points. The optical networking industry has experienced its own version of Moore's Law as the performance of optical components has doubled approximately every nine months at a given price. The similarities between the evolutions of the semiconductor and fiber optic industries extend beyond market and technology growth rates. In fact, both tales include a similar cast of company characters (see figure 2). Both started with just a few key players, but their growth prospects have sparked explosions of new competitors. National Semiconductor (founded 1959) and Intel (1968) achieved significant success as startups focused on semiconductor development and manufacturing. Similarly, their modern-day counterparts, JDS Uniphase and Avanex, attained early success by sticking to their core business: optical components. In both cases, these industry-focused "pure plays" competed head-to-head with well-funded divisions of more diversified corporations. The impact of a recession Then, semiconductor makers experienced the 1974-75 global recession, where leading industry competitors fell with the market downturn (see figure 3). This mid-70s slowdown profoundly affected the sector's competitive dynamics. Today, as U.S. market indices remain well below year 2000 levels, optical component vendors face similar straits. Over the last twelve months, lowered projections for overall telecom equipment expenditures coupled with stingy capital markets have led to a decline in valuations for several leading component producers (see figure 4). The next problem? Price. In the semiconductor industry, Fairchild Semiconductor saw its order backlog drop by almost 40% from 1973 to 1974.[4] The drop in demand drove prices down as much as 60% on several major product lines during the 1974-75 recession. Recovery was not immediate either, with several semiconductor plants still running at only 50% capacity as late as 1976[5]. Fast-forward to today: Oversupply of optical networking products has already driven bandwidth prices down significantly. And prices should continue to fall as manufacturing processes become more automated and a host of contract manufacturers (CMs) enter the space in hopes of providing build and assembly services similar to those available in the electronics industry. Combined with the general softening in telecom network spending, those production factors have put component makers under pricing and margin pressures they hardly experienced a year ago. Industry Maturation What will the optical component industry look like going forward? What will be the new competitive dynamic? Again, insight can come from hindsight. When the semiconductor industry emerged from the 1974-75 recession, vendors found strong underlying demand that restored industry growth and lifted them (and their stock prices) out of the doldrums. However, the industry emerged as a more mature, price competitive sector, which was quite different from its previous incarnation. The optical components industry is staring down this same path. Here are the parallels: Industry stabilization and stratification. In the latter half of the 1970s, the U.S. semiconductor industry largely stabilized around the "Big Five": Texas Instruments National Semiconductor Motorola Intel Fairchild. These broad-line vendors accounted for about two-thirds of U.S. factory sales and nearly all of the industry's pre-tax profits. A group of 35+ second tier players struggled to achieve the scale needed to effectively compete[6]. Driven by the need for large R&D budgets, increased product standardization, and the need to be price-competitive, a similar stratification is happening in optical components. A handful of large vendors—among them JDS Uniphase, Agere and the captive component division of Nortel—account for the majority of sales and are driving the industry's technological agenda. Higher entry barriers. The ability of new startups to enter the semiconductor industry declined dramatically in the late 1970s due to the increased cost of R&D and manufacturing equipment and a lack of venture capital. The capital required to compete with the Big Five simply became too great for a newcomer. After a surge of new entrants in recent years, the optical component industry should expect to see a decline in startup activity. Much like the semiconductor industry, the push towards the holy grail of automated manufacturing will require large amounts of cash, which only the large-scale vendors can access. In addition, small players will lack the customer leverage needed to secure product placement as supply begins to balance demand. Intense price competition. Semiconductor price competition intensified in the late 1970s, driven by increased product standardization and changes in end-use demand. The move away from customization to standardized memory and microprocessor designs focused the customer purchase decision more on price. At the same time, the demand for semiconductors became less and less dependent on the technology-focused military and more driven by price-sensitive consumer and industrial markets (see figure 5). The optical component industry is currently undergoing a similar shift as vendors see less opportunity in custom designs and growing demand from the designs applicable to more price-sensitive metropolitan-area markets. The metro market will be a large driver of growth in the optical networking space as service providers turn to optical systems as a means of unlocking the bottlenecks developing in local data traffic. Penetrating that market requires much lower cost equipment than that previously required for the long haul and submarine markets. Current estimates are that DWDM prices are twice what they need to be to gain wide acceptance in the metro market. Looking Ahead Flashing back to the semiconductor industry's ride of 30 years ago offers insight on the road ahead for optical components. Driving toward success will require the following: KEYS TO SUCCESS Dramatically improve price/performance Maintain strategic focus Compete aggressively for talent Dramatic price/performance improvements. With only slight modification, Moore's Law has held true in the semiconductor industry for over 35 years. All indications are that the rate of innovation in the optical networking industry will not plateau any time soon. Significant improvements in the price/performance ratio are necessary if fiber optic networks are to continue their rapid replacement of copper. This is especially true as the price-sensitive metro and access markets become the primary engines for growth. MORE INFO Vertical-Cavity Surface-Emitting Lasers Edited by Carl Wilmsen, Henryk Temkin, Larry A. Coldren Available November 2001 Laser Types - VCSELs vs Fabry-Perot Diode Lasers Component manufacturers must be continually searching for the technological breakthroughs, which will redefine the optics market in much the same way that Intel's microprocessor revolutionized the semiconductor industry. One promising development is that of vertical cavity surface emitting lasers (VCSELs) capable of transmitting signals over long distances at the conventional telecom wavelengths of 1310 and 1550 nanometers. VCSELs cost significantly less to produce than standard Fabry-Perot and distributed feedback (DFB) lasers, but several technical hurdles have limited them to short-haul 850-nanometer transmission. Several well-financed startups have already made announcements about long-wavelength VCSEL technologies whose potential order-of-magnitude cost reductions could alter the market's economics enough to threaten the source laser revenues of multiple Fabry-Perot and DFB laser manufacturers. Maintain strategic focus. Many semiconductor vendors were burned by ill-advised forays into consumer products such as digital watches and calculators. This move downstream meant that semiconductor vendors were competing directly against their customer base. Today, some competitors in the optical component sector are displaying a similar lack of focus. Nortel's optical components business unit competes against both JDS Uniphase and Corning, but Nortel is also both firms' biggest customer. Such relationships may be tolerated in today's environment, but it's doubtful they can continue as competition intensifies and the optical networking value chain disaggregates. Industry participants must identify where they can add the greatest value and move quickly to capitalize on it. Once they have defined and focused on their core businesses, optical component vendors must learn from the mistakes of their semiconductor predecessors and avoid unprofitable expansion into unattractive, though seemingly related, businesses.[7] Compete aggressively in the battle for talent. The semiconductor division of Fairchild Camera and Instrument Corporation was the brightest of many stars in the early development of the semiconductor industry. Founded in 1957, Fairchild developed the revolutionary planar diffusion process (which became the fundamental means of producing transistors and ICs) and was one of the early pioneers in penetrating the industrial and consumer markets. Despite these early successes, Fairchild experienced a tremendous amount of "brain drain" in the late 1960s as former employees departed to form over 45 new companies by the early 1970s[8]. Many of these new firms (including Intel) contributed to Fairchild's eventual decline. Top optical engineering talent is a limited resource and the ability to attract and retain a skilled technical staff is of critical importance. Indeed, several firms have utilized non-compete agreements and lawsuits to keep talent from straying to competitors. To avoid extreme measures, firms must develop targeted recruiting programs and compensate staff at a level to discourage their moves to competitors. Incumbents need to pay particular attention to their retention efforts as new entrants often look to poach skilled engineers using potentially lucrative stock options as a major enticement. Scott Bender is a Vice President in Bain & Company's Chicago office, and leads the optical components efforts within Bain's Telecom & Technology practice. Vince Tobkin directs Bain's Telecom & Technology practice. For more information contact Scott.Bender@bain.com.

Not long ago, a second-place technology company discovered a new product with the potential to revolutionize data handling. The company's decision to focus on this new technology fueled rapid earnings growth and propelled its stock price upward 230% in less than five months.

The market greeted the new technology with seemingly limitless demand. Clamoring customers beefed up research and development budgets, which attracted a flood of new suppliers to the industry. Venture capitalists, eager to cash in, lined up to bankroll every entrepreneur promising the next breakthrough innovation. Customers rushed to acquire the newest technology, upgrade existing products and services, and thereby boost their own company valuations. The gold rush had begun.

...the future of optical networking remains bright. [There will be] growth in the optical components sector as the demand for bandwidth continues to increase and copper-based electrical networks are replaced with faster, more robust fiber optic systems.

Then the mother lode petered out. An economic slowdown tempered demand, and the component maker's glittering prospects turned murky. Disillusioned investors scorned the innovator's bloated inventories, sagging stock, and lowered prospects. Business Week's prior proclamations of "boom times"[1] yielded to "a gloomier forecast"[2] for the sector.

You're probably guessing that we're talking about optical-age star Nortel Networks and the roller coaster optical components industry. Guess again. This is the story of chip maker Intel, whose maniacal focus on semiconductors revolutionized computing in the early 1970s, drove the company to market leadership, then to a screeching halt—for a time.

However, the story could have easily been that of today's optical components vendors Nortel and JDS Uniphase. The contours of today's crisis for optical component makers, and yesterday's for chipmakers, look close enough to constitute, perhaps, a crystal ball. If we peer into its depths, we may see that optical component makers are merely taking a historic ride, which could soon regain momentum.

Current Uncertainty

Investors and managers in the optical components sector (and the fiber optics industry in general) are searching for such a crystal ball. Former stock market darlings JDS Uniphase and Corning are both down over 90% from their previous 52-week highs[3]. Revenues for optical equipment providers like Lucent and Nortel have shrunk as telecom capital expenditures have plunged, with carriers seeking to optimize the performance of existing networks instead of building new ones.

Yet, a recent study by Bain & Company states that the future of optical networking remains bright. Bain projects growth in the optical components sector as the demand for bandwidth continues to increase and copper-based electrical networks are replaced with faster, more robust fiber optic systems. Nonetheless, if history is our guide, today's setbacks will change the competitive landscape. Rocky times will restructure the industry, reset key purchase criteria and shake out weaker competitors.

What Can History Teach Us?

Following the invention of the transistor in 1947 and the integrated circuit (IC) in 1958, the semiconductor industry experienced an explosion in growth (see figure 1) similar to recent developments in the optical components industry. In both cases, improvements in their respective price/performance ratios accelerated.

Gordon Moore, semiconductor industry pioneer and co-founder of Intel, observed in 1965 that the performance of ICs doubled every 12 months (later increased to every 18 months) at equivalent price points. The optical networking industry has experienced its own version of Moore's Law as the performance of optical components has doubled approximately every nine months at a given price.

The similarities between the evolutions of the semiconductor and fiber optic industries extend beyond market and technology growth rates. In fact, both tales include a similar cast of company characters (see figure 2).

Both started with just a few key players, but their growth prospects have sparked explosions of new competitors. National Semiconductor (founded 1959) and Intel (1968) achieved significant success as startups focused on semiconductor development and manufacturing. Similarly, their modern-day counterparts, JDS Uniphase and Avanex, attained early success by sticking to their core business: optical components. In both cases, these industry-focused "pure plays" competed head-to-head with well-funded divisions of more diversified corporations.

The impact of a recession

Then, semiconductor makers experienced the 1974-75 global recession, where leading industry competitors fell with the market downturn (see figure 3). This mid-70s slowdown profoundly affected the sector's competitive dynamics. Today, as U.S. market indices remain well below year 2000 levels, optical component vendors face similar straits. Over the last twelve months, lowered projections for overall telecom equipment expenditures coupled with stingy capital markets have led to a decline in valuations for several leading component producers (see figure 4).

The next problem? Price. In the semiconductor industry, Fairchild Semiconductor saw its order backlog drop by almost 40% from 1973 to 1974.[4] The drop in demand drove prices down as much as 60% on several major product lines during the 1974-75 recession.

Recovery was not immediate either, with several semiconductor plants still running at only 50% capacity as late as 1976[5]. Fast-forward to today: Oversupply of optical networking products has already driven bandwidth prices down significantly.

And prices should continue to fall as manufacturing processes become more automated and a host of contract manufacturers (CMs) enter the space in hopes of providing build and assembly services similar to those available in the electronics industry. Combined with the general softening in telecom network spending, those production factors have put component makers under pricing and margin pressures they hardly experienced a year ago.

Industry Maturation

What will the optical component industry look like going forward? What will be the new competitive dynamic? Again, insight can come from hindsight. When the semiconductor industry emerged from the 1974-75 recession, vendors found strong underlying demand that restored industry growth and lifted them (and their stock prices) out of the doldrums.

However, the industry emerged as a more mature, price competitive sector, which was quite different from its previous incarnation. The optical components industry is staring down this same path. Here are the parallels:

Industry stabilization and stratification. In the latter half of the 1970s, the U.S. semiconductor industry largely stabilized around the "Big Five":

Texas Instruments

National Semiconductor

Motorola

Intel

Fairchild.

These broad-line vendors accounted for about two-thirds of U.S. factory sales and nearly all of the industry's pre-tax profits. A group of 35+ second tier players struggled to achieve the scale needed to effectively compete[6]. Driven by the need for large R&D budgets, increased product standardization, and the need to be price-competitive, a similar stratification is happening in optical components. A handful of large vendors—among them JDS Uniphase, Agere and the captive component division of Nortel—account for the majority of sales and are driving the industry's technological agenda.

Higher entry barriers. The ability of new startups to enter the semiconductor industry declined dramatically in the late 1970s due to the increased cost of R&D and manufacturing equipment and a lack of venture capital. The capital required to compete with the Big Five simply became too great for a newcomer. After a surge of new entrants in recent years, the optical component industry should expect to see a decline in startup activity.

Much like the semiconductor industry, the push towards the holy grail of automated manufacturing will require large amounts of cash, which only the large-scale vendors can access. In addition, small players will lack the customer leverage needed to secure product placement as supply begins to balance demand.

Intense price competition. Semiconductor price competition intensified in the late 1970s, driven by increased product standardization and changes in end-use demand. The move away from customization to standardized memory and microprocessor designs focused the customer purchase decision more on price. At the same time, the demand for semiconductors became less and less dependent on the technology-focused military and more driven by price-sensitive consumer and industrial markets (see figure 5).

The optical component industry is currently undergoing a similar shift as vendors see less opportunity in custom designs and growing demand from the designs applicable to more price-sensitive metropolitan-area markets. The metro market will be a large driver of growth in the optical networking space as service providers turn to optical systems as a means of unlocking the bottlenecks developing in local data traffic.

Penetrating that market requires much lower cost equipment than that previously required for the long haul and submarine markets. Current estimates are that DWDM prices are twice what they need to be to gain wide acceptance in the metro market.

Looking Ahead

Flashing back to the semiconductor industry's ride of 30 years ago offers insight on the road ahead for optical components. Driving toward success will require the following:

KEYS TO SUCCESS
Dramatically improve price/performance
Maintain strategic focus

Compete aggressively for talent

Dramatic price/performance improvements. With only slight modification, Moore's Law has held true in the semiconductor industry for over 35 years. All indications are that the rate of innovation in the optical networking industry will not plateau any time soon. Significant improvements in the price/performance ratio are necessary if fiber optic networks are to continue their rapid replacement of copper. This is especially true as the price-sensitive metro and access markets become the primary engines for growth.

MORE INFO
Vertical-Cavity
Surface-Emitting Lasers
Edited by Carl Wilmsen,
Henryk Temkin, Larry A. Coldren
Available November 2001
Laser Types - VCSELs vs
Fabry-Perot Diode Lasers

Component manufacturers must be continually searching for the technological breakthroughs, which will redefine the optics market in much the same way that Intel's microprocessor revolutionized the semiconductor industry. One promising development is that of vertical cavity surface emitting lasers (VCSELs) capable of transmitting signals over long distances at the conventional telecom wavelengths of 1310 and 1550 nanometers.

VCSELs cost significantly less to produce than standard Fabry-Perot and distributed feedback (DFB) lasers, but several technical hurdles have limited them to short-haul 850-nanometer transmission. Several well-financed startups have already made announcements about long-wavelength VCSEL technologies whose potential order-of-magnitude cost reductions could alter the market's economics enough to threaten the source laser revenues of multiple Fabry-Perot and DFB laser manufacturers.

Maintain strategic focus. Many semiconductor vendors were burned by ill-advised forays into consumer products such as digital watches and calculators. This move downstream meant that semiconductor vendors were competing directly against their customer base. Today, some competitors in the optical component sector are displaying a similar lack of focus. Nortel's optical components business unit competes against both JDS Uniphase and Corning, but Nortel is also both firms' biggest customer.

Such relationships may be tolerated in today's environment, but it's doubtful they can continue as competition intensifies and the optical networking value chain disaggregates. Industry participants must identify where they can add the greatest value and move quickly to capitalize on it. Once they have defined and focused on their core businesses, optical component vendors must learn from the mistakes of their semiconductor predecessors and avoid unprofitable expansion into unattractive, though seemingly related, businesses.[7]

Compete aggressively in the battle for talent. The semiconductor division of Fairchild Camera and Instrument Corporation was the brightest of many stars in the early development of the semiconductor industry. Founded in 1957, Fairchild developed the revolutionary planar diffusion process (which became the fundamental means of producing transistors and ICs) and was one of the early pioneers in penetrating the industrial and consumer markets.

Despite these early successes, Fairchild experienced a tremendous amount of "brain drain" in the late 1960s as former employees departed to form over 45 new companies by the early 1970s[8]. Many of these new firms (including Intel) contributed to Fairchild's eventual decline.

Top optical engineering talent is a limited resource and the ability to attract and retain a skilled technical staff is of critical importance. Indeed, several firms have utilized non-compete agreements and lawsuits to keep talent from straying to competitors.

To avoid extreme measures, firms must develop targeted recruiting programs and compensate staff at a level to discourage their moves to competitors. Incumbents need to pay particular attention to their retention efforts as new entrants often look to poach skilled engineers using potentially lucrative stock options as a major enticement.

Scott Bender is a Vice President in Bain & Company's Chicago office, and leads the optical components efforts within Bain's Telecom & Technology practice. Vince Tobkin directs Bain's Telecom & Technology practice. For more information contact Scott.Bender@bain.com.