How will the fiber industry evolve?
With many restrictive patents about to expire, will we see a host of new entrants in the fiber-production market?
Arnab Sarkar Bular llc
From its inception, the U.S. optical-fiber industry has been very centralized. Corning Inc. (Corning, NY), Lucent Technologies (Morristown, NJ) and its subsidiaries, Alcatel (Richardson, TX) and Spectran Communication Fiber Technologies Inc. (Sturbridge, MA) now produce nearly 100% of the fibers used for telecommunications. Corning and Lucent (at&t at that time) had cross-licensing agreements for each other`s patents at the early stages of the technology`s evolution; Spectran and Alcatel (itt at that time) obtained patent licenses from Corning to produce fibers. Legal actions against companies attempting to produce optical fibers in the United States have prevented other entries.
The industry is also diverse in its structure. Shin-Etsu Chemical Co. of Japan sells preforms. There are integrated producers, such as Lucent and Alcatel, that not only produce preforms, draw fibers, and manufacture cables, but also produce other components of optical communications systems as well. Companies such as Corning and Spectran sell fibers but have cable companies as subsidiaries. There are also many cable companies that produce cable from purchased fiber. The health of this group has proven particularly vulnerable during the periods of fiber shortage that the industry has faced during its growth spurts.
Although not as prolific as other parts of the information marketplace, the fiber industry has a reasonably rapid rate of product introductions. Non-zero dispersion-shifted (nzds) fiber, developed to meet the needs of dense wavelength-division multiplexing systems, is still evolving. Lucent and Corning have 100% of the nzds fiber market; furthermore, these two companies are cross-licensed for each other`s patents. Yet the patent situation for nzds fiber is also dynamic, with many new patents being issued in recent years.
Another industry trend is the deeper penetration of fiber into the network. A consequence of this trend is an increased interest in optical-fiber cables by companies that are still predominantly producers of copper cable.
The industry`s track record on cost reduction, in terms of cost as a function of bandwidth distance, is impressive and probably matches or exceeds the performance of the semiconductor industry. Singlemode fibers are being sold at prices lower than anyone had predicted even a decade ago.
Most of the restrictive patents discussed above will expire in the next few years, raising the question of how the industry will restructure itself in the era of potentially open competition. How long will existing participants be able to prevent new entrants? Will cable companies integrate backwards into fiber production, and will penetration of fiber in premise wiring accelerate this?
Economics and market access will decide how the industry will restructure itself. In the semiconductor industry, materials companies produce ingots and wafers, and device companies produce chips. Will the fiber industry follow this example and have glass companies produce preforms, and cable companies draw and cable fibers?
Patent status
More than 500 optical-fiber patents have been issued in the United States. In addition to patents that can be considered restrictive for cost-effective production of optical fibers for a specific design by a specific process, many improvement patents must also be investigated in great detail to establish whether or not it is possible to produce optical fibers in the United States without infringing on existing patents. Furthermore, patents held by Lucent Technologies have not been considered restrictive, because these patents were obtained by at&t when it was still a monopoly, obligating Lucent to license these under terms defined by government decree. However, if history is any indicator, companies initially entering optical-fiber production must anticipate, budget, and plan to defend their position in a court of law.
The table shows the list of restrictive patents that are yet to expire. In compiling this table, no consideration has been given to whether or not a patent is valid. Singlemode-fiber production by the modified chemical-vapor deposition (mcvd) process is now unrestricted. However, Lucent patent 4,257,797, which claims sleeving and overcladding of mcvd preforms, may need to be licensed if one wants to manufacture and sell mcvd fiber before October 1999. Another Lucent patent, number 4,909,816, which expires in March 2007, claims the basic mcvd process, which was invented in the early 1970s and for which the company first filed for patent in 1974. If this patent is held valid in court, the assignee will have 33 years of protection for an invention made in 1974. Although that outcome is unlikely, companies planning to enter fiber production using the mcvd process should investigate this patent.
It is also worth noting that the recent action by Corning against Fibercore was based on patent 4,339,173, which expires in July 1999. From then on, multimode fiber production by mcvd will also become unrestricted.
In the case of the vapor-phase axial deposition (vad) process, the last remotely related Corning patent, sintering patent 4,165,223, expires in August 1998. The key vad patents for single- and multimode fibers (4,345,928 and 4,367,085, respectively) restrict their claims to sleeving as an integral part of the process. However, the soot overcladding process has been used instead of sleeving to produce vad fibers since the late 1980s. Thus, it may be argued that these patents do not restrict production of vad fibers in which additional cladding is added by overcladding. Similarly, an ntt sintering patent, 4,388,111, can also be circumvented, making vad fiber production unrestricted today.
The outside vapor deposition (ovd) process patents covering singlemode and multimode fibers will expire within the next six years.
In summary, it is evident that the era when existing participants can no longer restrict new entrants in standard singlemode and multimode fiber manufacture is fast approaching.
New patents in nzds fiber designs will continue to be issued over the next few years and already-issued patents are expected to remain valid through the next decade. Thus, it is possible that the market will become mostly unrestricted for standard singlemode and multimode fibers, but nzds will continue to be restricted and sold by a limited number of companies as a high-end product commanding a significantly higher price.
Effect of industry structure on competition
The structure of the industry at the start of the next century will probably be as shown in the figure. About 10% of cabled fiber will be produced from purchased preforms; approximately 40% will be from purchased fiber; and 50% of cabled-fiber output will be from fiber produced by fully integrated producers.
There are several potential drivers of this restructuring. Purchasers of optical-fiber cables are the most logical drivers for a competitive marketplace that would provide the best products at the lowest price. However, from the current buyer`s point of view, the cable market appears quite competitive. Because fiber accounts for only about 5% of the system cost, the fact that the fiber market is not competitive does not get noticed. Fiber is also a critical component for system reliability. Therefore, buyers of communications systems would rather have confidence in the quality of the product and the track record of the manufacturer than attempt to drive down price by soliciting new vendors.
Currently, the only visible drive for restructuring is by preform manufacturers such as Shin-Etsu Chemical Co., which has been selling preforms in countries where there are no patent issues and to companies in developed countries that have licenses to produce or have made optical fibers. Their efforts have created a small number of companies worldwide that have bought fiber-drawing equipment from established equipment manufacturers to produce fibers from purchased preforms. Globally, the share of optical fiber produced from purchased preforms can increase from nearly zero in 1987 to 10% in 2000.
However, the most active drivers of restructuring will be the cable companies that have large market share in distribution cables and have continued their emphasis on copper cables. As fiber makes deeper and deeper penetration into distribution networks for both voice and data, these companies have to become more aggressive in the optical-fiber-cable market in order to survive. They have to choose their level of integration: whether to buy fiber and cable; buy preforms, then draw and cable fibers; or be a fully integrated manufacturer.
Companies can most easily enter cable production with purchased fiber because they can use the name of an established fiber maker. This entry approach also has the lowest technical risk. However, because of purchased-volume-based pricing by fiber manufacturers that favor their own subsidiaries with an established volume base, with some exceptions, it has proven difficult for cable companies to build a long-term business at this level of integration.
The merchant fiber market is also smaller than most people consider it to be. If one considers Corning to be a captive supplier of Siecor, the merchant market comprises only Pirelli Cables plus perhaps 10 other companies, with a total market share of about 10%. Pirelli is a large-volume fiber producer in Europe, and the three largest after Pirelli are most likely Lucent-Fitel (a joint venture between Furukawa and Lucent) Sumitomo Cables, and Alcoa-Fujikura--all with fiber-producing affiliations. Consequently, if the patent situation clarifies itself, there is a good probability that all these cable companies will become captive to supply from their affiliated fiber companies. Thus cable companies now entering optical-fiber production have to understand that the merchant market for fibers may become less than 5% of the total market. At that volume, companies may not find it prudent to rely on fiber suppliers for their entire investment. Therefore, it is natural for cable companies to hedge their bets by producing a portion of their fiber needs, just as fiber companies have hedged their bets by establishing cable subsidiaries.
Another factor is the long experience cable companies have had in licensing patents from at&t in copper technology. These forces then make these companies the most probable new entrants in optical-fiber production.
As evidenced by Plasma Optical Fibers` relationship with Chromatics Technologies, it is likely that the established foreign fiber producers will want to expand their operations in the United States as well. These expansions will take the form of alliances, exports, and finally the establishment of fully owned operations such as Sumitomo Electric`s past efforts. Eventually, a portion of the cables used in the United States will be from these sources.
Possible new entrants
The effect of startups on the future fiber marketplace is expected to be minimal, at least at first. The magnitude of investment required to start optical-fiber production even at the minimum economic production level is significant. Estimates range from $30 million to $50 million per 1,000,000 km per year production capacity; and estimates of minimum economic production volumes range from 500,000 to 1,000,000 km per year based on the choice of technology. Startups will also have to factor into their cash flow considerations the time and cost to reach profitability and the cost of probable litigation relating to patents. Thus, it is unlikely that independent startups will be the first to break into fiber production in the United States.
A wild card in the restructuring scenario is Heraeus (Atlanta, GA), a tube producer that more or less supplies all the tubes necessary to produce optical fibers by the mcvd and plasma chemical- vapor deposition (pcvd) processes. Tubes represent approximately 95% of the glass in fibers produced by these processes, and tube price drives fiber cost. Thus if Heraeus can reduce tube price to the level at which Corning and vad manufacturers produce cladding glass--and there is fundamentally no reason that they cannot--integrated companies will continue to dominate the market for a long time.
Cable companies are likely to be the first set of new entrants to a restructured fiber marketplace. These companies are likely to lose the most if they are not more aggressive in the optical-fiber-cable market. Their advantages are established distribution channels and presence in the market. The hurdles they face are significant, the largest being the risk of technology acquisition. Equipment manufacturers are capable of providing equipment and process technology for fiber drawing and testing. However, the equipment and technology that can be obtained from equipment vendors for preform production is not adequate for cost-effective fiber production. Among the companies that have attempted to produce fibers independently or by use of technology provided by equipment vendors, only a handful have attained any level of success.
On the other hand, obtaining technology by alliances is expensive. Although the availability of trained technologists is improving and technical consulting companies are being formed, their track record is yet to be established. A second hurdle is to reach a scale of production at which producing is cheaper than buying--and then maintaining that position.
The best approach for these companies is to form a consortium company to produce fibers for their needs. This would give them the economies of scale needed for successful entry as well as a hedge against the eventuality of fiber companies consolidating with their subsidiary cable companies in the future.
Another possible entrant is the "mini-plant." Fiber production is a batch process, and the economies of scale are not like those of a continuous process. Also, fiber production has direct costs in the range of 20% to 30% if no tube is used in preform production. This creates a financial structure similar to that seen in the steel industry, where mini-mills have made a significant impact on the market. It is quite probable that with the expiration of patents, there will be a proliferation of mini-plants in fiber or preform production that will similarly affect the fiber industry.
Theoretically, a single-product mini-plant with a capacity of 1,000,000 km per year could sell singlemode preforms profitably at $15 per km or less. One scenario is the establishment of mini-plants by a consortium of cable companies; another is by startups after the patent situation is clarified in court.
Summary
In an era of open competition in fiber production, which in the United States is inevitable as the patents expire, the industry will restructure based on the actions taken by the dominant participants of the patent era. Will Lucent and Alcatel find it necessary to produce their own preforms and fiber, even when fiber is 2.5% to 5% of the value of the products they offer to their competition? Will Corning reduce prices fast enough to prevent new companies from entering the market or is it already too late for Corning to prevent backward integration? In other words, will cable companies buying optical fibers from Corning and Spectran set up fiber operations as a consortium or individually irrespective of what Corning does? Will Shin-Etsu drive preform prices down to a level that will make it unattractive for integrated cable makers to produce their own preforms? Will Heraeus drive down tube price to perpetuate fiber manufacturing amongst integrated producers? Will mini-plants make any impact in the optical-fiber industry?
It is hard to predict what will happen, but there will surely be a battle for turf. It is unlikely that the industry will remain as it has been during the initial patent era. u
Arnab Sarkar is the founder of Bular llc, a start-up company in the field of optical fibers. He has been involved in the industry for over 25 years and has experience in producing optical fibers using mcvd, vad, and ovd processes. He can be reached at [email protected].