Back to GetFilings.com

Table of Contents

UNITED STATES SECURITIES AND EXCHANGE COMMISSION

Form 10-K

000-27999

Finisar Corporation

1308 Moffett Park Drive

Registrant’s telephone number, including area code:

Securities registered pursuant to Section 12(b) of the Act:

Securities registered pursuant to section 12(g) of the Act:

      Indicate by check mark whether the registrant (1) has filed all reports required to be filed by Section 13 or 15(d) of the Securities Exchange Act of 1934 during the preceding 12 months (or for such shorter period that the Registrant was required to file such reports), and (2) has been subject to such filing requirements for the past 90 days.     Yes þ          No o

      Indicate by check mark if disclosure of delinquent filers pursuant to Item 405 of Regulation S-K is not contained herein, and will not be contained, to the best of registrant’s knowledge, in definitive proxy or information statements incorporated by reference in Part III of this Form 10-K or any amendment to this Form 10-K.     o

      Indicate by check mark whether the registrant is an accelerated filer (as defined in Rule 12b-2 of the Act).     Yes þ          No o

      As of October 31, 2002, the aggregate market value of the voting stock held by non-affiliates of the registrant was approximately $95,910,000, based on the closing sales price of such stock as reported on the Nasdaq Stock Market on such date of $0.75 per share. Shares of common stock held by officers, directors and holders of more than ten percent of the outstanding common stock have been excluded from this calculation because such persons may be deemed to be affiliates. This determination of affiliate status is not necessarily a conclusive determination for other purposes.

      As of June 27, 2003, there were 208,534,940 shares of the registrant’s common stock, $.001 par value, issued and outstanding.

      Portions of the definitive proxy statement for the annual meeting of stockholders of the registrant are incorporated by reference into Part III.

Table of Contents

INDEX TO ANNUAL REPORT ON FORM 10-K

1

Table of Contents

PART I

Item 1.     Business

      This report contains forward-looking statements within the meaning of the Private Securities Litigation Reform Act of 1995. We use words like “anticipates,” “believes,” “plans,” “expects,” “future,” “intends” and similar expressions to identify these forward-looking statements. We have based these forward-looking statements on our current expectations and projections about future events. These forward-looking statements are subject to risks, uncertainties and assumptions about us, including:

	•	uncertainty regarding our future operating results;
	•	our ability to introduce new products in a cost effective manner that are accepted in the marketplace;
	•	delays or loss of sales due to long product qualification cycles for our products;
	•	the possibility of lower prices, reduced gross margins and loss of market share due to increased competition; and
	•	increased demands on our resources due to the integration of several companies and product lines that we have acquired and cost reductions which may further reduce our available resources.

      Other factors that could cause actual result to differ from expectation are discussed in “Factors that Could Affect Our Future Performance.”

      In light of these risks, uncertainties and assumptions, the forward-looking events discussed in this report might not occur. We undertake no obligation to publicly update or revise any forward-looking statements, whether as a result of new information or future events.

Overview

      We are a leading provider of fiber optic subsystems and network performance test and monitoring systems which enable high-speed data communications over local area networks, or LANs, storage area networks, or SANs, and metropolitan access networks, or MANs. We are focused on the application of digital fiber optics to provide a broad line of high-performance, reliable, value-added optical subsystems for data networking and storage equipment manufacturers. Our line of optical subsystems supports a wide range of network applications, transmission speeds, distances, physical mediums and configurations. We also provide network performance test and monitoring systems to original equipment manufacturers for testing and validating equipment designs and to operators of networking and storage data centers for testing, monitoring and troubleshooting the performance of their systems. We sell our products to leading storage and networking equipment manufacturers such as Brocade, Cisco, EMC, Emulex, Extreme Networks and Hewlett-Packard Company.

      Since October 2000, we have acquired six privately-held companies and certain assets from two other companies in order to gain access to new technologies which can be used in conjunction with our existing core competencies to develop new and innovative products. During our fiscal year ended April 30, 2001, we acquired Sensors Unlimited, Inc., Demeter Technologies, Inc., Medusa Technologies, Inc., and Shomiti Systems, Inc. During our fiscal year ended April 30, 2002, we acquired Transwave Fiber, Inc. and certain assets, including equipment and intellectual property, of AIFOtec GmbH in Germany. In May 2002, we acquired certain assets, including equipment, inventory and intellectual property, from New Focus, Inc. In April 2003, we acquired Genoa Corporation. These acquisitions have broadened our product offerings and provided us access to advanced optical component technologies that we believe will enable us to develop innovative and more highly integrated subsystems while accelerating the timeframe required to develop such products.

      The principal strategic goal of most of our acquisitions to date related to our optics business has been to gain access to leading-edge technology for the manufacture of optical components in order to improve the performance and reduce the cost of our optical subsystem products. We have also sold a limited amount of

2

Table of Contents

Industry Background

      The ubiquity of computing by businesses, organizations and individuals worldwide and the need to interconnect multiple computing and storage devices to enable widespread communications has given rise to the multi-billion dollar computer networking and storage industries. The rapid growth in the number of corporate and residential users accessing communications networks and the proliferation of new applications designed for electronic commerce, communications and entertainment has resulted in the digitization and accumulation of enormous amounts of data. In addition, the value of much of this data has and will become increasingly mission-critical to business enterprises and other organizations which must ensure that it is accessible on a continuous and reliable basis by employees, suppliers and customers over a diverse geographic area. The need to quickly transmit, store and retrieve large blocks of data across these networks in a cost-effective manner has resulted in large-scale equipment expenditures by enterprises and service providers to expand the capacity, or bandwidth, of their network and storage infrastructures using fiber optic transmission technology.

      Data networks are frequently described in terms of the distance they span and by the hardware and software protocols used to transport the data. The major network segments are frequently referred to as MANs, LANs, SANs and wide area networks, or WANs. The technologies used to build these networks are continuously evolving but retain a common thread — the growing use of digital fiber optics and multiple wavelengths to increase capacity and performance in most network applications.

      Digital Fiber Optics. Digital fiber optic transmission technology was originally developed for use in WANs to increase the capacity and performance of long distance telecommunications networks using equipment which supports the Synchronous Optical Network (SONET) or Synchronous Digital Hierarchy (SDH) protocols. In contrast, early LANs, SANs and MANs, with their relatively limited performance requirements, short connection distances and low transmission speeds, did not require the performance capabilities of fiber optics. Systems on these networks were generally interconnected using copper cabling or twisted pair wire.

      As the need to access a common database of shared data and network resources became more widespread, it also created the need to connect users over greater distances. As the bandwidth, storage capacity and transmission distance requirements of enterprises and service providers have increased, it has become necessary to replace the limited transmission capabilities of copper cabling and twisted pair wire with the superior transmission capabilities of digital fiber optics to build practical, high-speed LANs, SANs and MANs.

      Interconnecting the various elements of these networks is accomplished with a transceiver, which combines a transmitter for converting an electrical signal into an optical signal for transmission over a fiber optic cable and a receiver for converting an optical signal into an electrical signal so that it can be processed by the network element in which the transceiver resides. Network elements generally include multiple transceivers, or ports, in order to be able to process several signals at the same time.

      Until the mid-1990s, most WAN networks relied on a single wavelength of light to carry the digital information to be transmitted between various points on the network. With the introduction of dense wavelength division multiplexing, or DWDM, multiple wavelengths of light could be combined or multiplexed onto a single fiber, thus enhancing the capacity of these networks without the added cost associated with laying new fiber in the ground. While the use of DWDM has limited applications for MANs, the use of coarse

3

Table of Contents

      Gigabit Ethernet and Local Area Networks. Early LANs were implemented to connect a limited number of users within relatively close proximity. Most of these LANs used the Ethernet transmission protocol which was developed to allow users to access the LAN and share basic common services such as file servers and printers. Because these early LANs had relatively limited performance requirements, short connection distances and low transmission speeds, systems on these LANs were generally connected by copper cabling.

      As deployment of LANs increased, Ethernet became the predominant LAN technology. As bandwidth needs and server processing power increased and larger numbers of users strained the early LAN infrastructure, Ethernet technology evolved from the original 10 megabits per second, or Mbps, version to 100 Mbps Fast Ethernet. In response to continually increasing bandwidth and performance requirements, Gigabit Ethernet technology, which operates at 1,000 Mbps, was introduced in 1998. Most of the Gigabit Ethernet equipment currently being shipped relies on fiber optic technology which enables data to be transmitted accurately, at very high speeds and over long distances. However, we expect that in the near term the demand for copper-based equipment will grow faster than the demand for fiber optic-based equipment in LAN applications due to the deployment of Gigabit Ethernet to the desktop where transmission distances are much shorter, generally less than 100 meters. Although we currently derive most of our revenue for these applications from the sale of optical transceivers, we intend to expand our line of copper-based transceivers in order to capitalize on the expected growth in this segment of the market.

      The growth in Gigabit Ethernet connectivity within the enterprise is fueling increased demand for equipment based on 10-Gigabit Ethernet, or 10GigE. Since the ratification of a 10GigE standard in June 2002, a number of companies have introduced new optical products supporting this higher data rate. We believe that the upgrading of infrastructure based on the 10GigE standard will initially be focused on upgrading data centers and corporate backbones where businesses and other organizations can consolidate their file servers into a smaller number of high-capacity servers, yielding significant cost savings in the process (see “Fibre Channel and Storage Area Networks”).

      Fibre Channel and Storage Area Networks. Like data networking technology, data storage technology has evolved rapidly over the past decade. Traditionally, storage devices were connected to a single server and LAN in close proximity using a standard interface protocol known as the Small Computer Systems Interface, or SCSI. SCSI currently allows storage devices and servers to communicate at a maximum speed of 160 megabytes per second, over a maximum transmission distance of 12 meters and supports a maximum of 16 devices on a single bus. Although these distances and speeds were sufficient for early storage applications, SCSI has become a limiting technology for emerging storage applications, which require networking at high speeds over long distances and need to interconnect large numbers of users.

      With the evolution of the Internet, the amount of data to be stored has increased to the point where the cost of managing and protecting this data has become the dominant cost of a typical information technology department. This in turn has created a demand for faster, more efficient interconnection of data storage systems with servers and LANs. Contributing to this demand are:

	•	the need to connect increasing numbers of storage devices and servers to a growing number of users;
	•	the need to interconnect servers and storage systems supplied by multiple vendors;
	•	the increasingly mission-critical nature of stored data and the need for rapid access to this data; and
	•	the expense and complexity associated with managing increasingly large amounts of data storage.

      Although the development of Gigabit Ethernet increased LAN transmission speeds by more than 1,000 times during the 1990s, storage-to-server data transmission speeds on SCSI-based systems increased by less than ten times during this period. This speed disparity created a bottleneck between storage systems and servers and the LANs connected to those servers. In 1995, the Fibre Channel interconnect protocol was standardized to address the speed, distance and connectivity limitations of SCSI-based storage while

4

Table of Contents

      SANs provide many benefits, including transmission speeds comparable to high-speed LANs and transmission distances that allow broader sharing of resources. SANs also enable enhanced network applications such as storage backup, and better overall storage management achievable through centralized storage resources. According to a report issued by IDC in December 2002, the number of transceivers to be shipped in Fibre Channel systems, including switches, storage arrays and host bus adapters, or HBAs, will increase from 2.7 million in 2002 to over 10 million in 2006, representing a compound annual growth rate of 39%. Most SANs to be deployed will rely on fiber optic subsystems to transmit and receive data at very high speeds with high accuracy, and often over long distances. While SANs were initially deployed with transmission speeds of one Gbps, most SAN systems are now capable of transmitting data at speeds of up to 2.125 Gbps. Like Gigabit Ethernet, the Fibre Channel protocol is scalable, allowing for the potential development of systems with speeds of over 10 Gbps.

      Recently, the Internet Small Computer System Interface, or iSCSI, has emerged as an alternative to the Fibre Channel protocol to facilitate data transfers over intranets and to manage storage over long distances. However, iSCSI is not designed for replication and disaster recovery where Fibre Channel provides the capability to move very large amounts of data quickly over an IP link. With the introduction of equipment capable of supporting 10GigE, the iSCSI alternative is expected to become more competitive with Fibre Channel-based SANs.

      Metropolitan Access Networks. The need for increased bandwidth is also increasing the demand for high-speed connectivity in MANs. The deployment of DWDM-based systems has resulted in a 12,000% increase in capacity for long-haul networks since early 1997. Over the same period, the transmission of data within buildings and corporate campus networks has increased to gigabit speeds. However, connecting these islands of data is a “copper straw” where transmission rates are reduced to megabits per second or slower over a combination of twisted pair wire, T-1 lines, frame relay and wireless links. The opportunities and technical challenges represented by this problem are considerable. Previous technologies used to supply multi-gigabit bandwidth in WANs, such as DWDM, will likely be too costly to deploy in MANs on any large scale. Instead, new technologies that use more cost-effective solutions such as 10GigE and/or a combination of 10GigE and coarse wavelength division multiplexing, or CWDM, are likely to be preferred in most of these networks, with DWDM deployed on a more limited basis where network congestion is particularly severe.

      In addition to saving costs, these new technologies will provide network reliability similar to SONET-based solutions currently used in WANs. While most carriers will be reluctant to replace their installed base of SONET/SDH-based transport equipment in MANs and WANs, a new protocol, Virtual Concatenation, has been developed that allows the transmission of Ethernet traffic over SONET networks. By mapping Ethernet traffic onto SONET, carriers can offer Ethernet services in a cost effective manner and extend the life of their existing infrastructure. A report issued by Infonetics Research predicts that the number of Ethernet transceivers for MAN applications is expected to grow 331%, from approximately 756,000 in 2002 to 3.3 million by 2006. While Infonetics Research expects Ethernet-based systems will ultimately account for the majority of MAN equipment spending by 2013, it expects that 60% of line cards for use in MAN networks in 2006 will still consist of OC-3 (155Mbps) and OC-12 (622Mbps) in order to relieve traffic bottlenecks. While we have historically focused on solutions that require transmission speeds of one Gbps or more, we intend to expand our product line to include these lower speed applications as well, in order to become a broad based supplier of optical subsystems.

      The design and development of data and storage networking systems require extensive testing to ensure system performance and reliability. As new, highly complex transmission protocols such as Gigabit Ethernet, 10-Gigabit Ethernet, iSCSI and Fibre Channel have emerged, system testing has become more difficult, requiring increasingly sophisticated and specialized test systems capable of capturing data at high speeds,

5

Table of Contents

      A report issued by Frost and Sullivan in March 2003 estimates that the market for testing and monitoring equipment decreased by 34% in calendar 2002 to approximately $500 million, of which approximately $400 million was for testing and monitoring equipment deployed in the field to monitor the performance of Ethernet networks within data centers and $100 million was for testing and monitoring equipment sold to network equipment manufacturers for use in developing and manufacturing their products. The demand for testing and monitoring equipment for deployment in data centers will be affected not only by the general economy and its impact on IT spending but also the deployment of new applications such as voice-over-IP, or VOIP, iSCSI and 10GigE.

      The development and manufacture of high quality, cost-effective fiber optic subsystems for LANs, SANs and MANs present a number of significant technical challenges, including the following:

	•	As data rates increase, it becomes significantly more difficult to maintain data integrity because high speed signals can be degraded unless subsystem components such as lasers, detectors and integrated circuits are properly integrated and packaged;
	•	The increasingly mission-critical nature of data transmission and storage has magnified the impact of system failures, increasing the need for system reliability and the importance of real-time performance monitoring;
	•	Manufacturers of high speed networking equipment require optical subsystems that support a wide range of transmission distances, protocols, applications and form factors; and
	•	Compliance with standards set by the Federal Communications Commission, or FCC, for electromagnetic interference emissions, or EMI, is significantly more difficult to achieve at higher data rates.

The Finisar Solution

      We are a leading provider of fiber optic subsystems and network performance test and monitoring systems which enable high-speed data communications over LANs, SANs and MANs. We are focused on providing high-performance, reliable, value-added optical subsystems for data networking and storage equipment manufacturers that develop and market systems based on Gigabit Ethernet, 10-Gigabit Ethernet, Fibre Channel and SONET protocols. Our line of optical subsystems supports a wide range of network applications, transmission speeds, distances, and physical configurations. We also provide unique network performance test and monitoring systems to original equipment manufacturers for testing and validating their equipment designs and to networking and storage operators for testing, monitoring and troubleshooting the performance of their systems. Our products provide the following key benefits to manufacturers of high-speed data networking and storage systems:

      Value-Added Functions and Intelligence. Our high-speed fiber optic subsystems are engineered to provide our customers with value-added functionality beyond the basic capability of enabling high-speed transmission. Many of our optical subsystems include a microprocessor containing specially-developed software that allows customers to monitor the optical performance of each port on their systems in real time. Real-time monitoring and interoperability are particularly important in the Gigabit Ethernet LAN and Fibre

6

Table of Contents

      High Level of Data Integrity. Through the use of advanced packaging and circuit design, our optical subsystems deliver data at very high speeds over varying distances with very low error rates. We engineer our subsystems to exceed the industry standard error rate of 1 bit per trillion bits transmitted. This degree of data integrity allows our subsystems to operate reliably over a wide range of temperatures and other field conditions which we believe enables our customers to design and deliver more robust systems.

      High Reliability. We design all of our optical subsystems to provide the high reliability required for data networking and storage applications that are critical to an enterprise. Using standard statistical methodology and testing, we have been able to predict that some of our products can be expected to operate reliably for up to 40 million hours. Our subsystems are engineered to operate with minimal power requirements thereby increasing product life, and to function across a wide range of temperatures and voltages. This reliability and flexibility have allowed our subsystems to be designed into the products of manufacturers who provide systems for a variety of mission-critical applications. In addition, because our subsystems emit lower levels of electromagnetic interference, or EMI, than the standards set by the FCC, we offer manufacturers greater flexibility in the design of their systems and integration of other components and subsystems.

      Broad Product Line of Optical Subsystems. We offer a broad line of optical subsystems which operate at varying protocols, speeds, fiber types, voltages, wavelengths and distances and are available in a variety of industry standard packaging configurations, or form factors. Our optical subsystems are designed to comply with key networking protocols such as Fibre Channel, Gigabit Ethernet and 10-Gigabit Ethernet and to plug directly into standard port configurations used in our customers’ products. The breadth of our optical subsystems product line is important to many of our customers who manufacture a wide range of networking products for diverse applications.

      Broad Product Line of Test and Monitoring Systems. We offer a broad line of test and monitoring systems to assist our customers in efficiently designing reliable, high-speed networking systems and testing and monitoring the performance of Fiber Channel and Ethernet-based networks. We believe our test systems enable original equipment manufacturers to focus their attention on the development of new products, reduce overall development costs and accelerate time to market while our monitoring systems provide real time feedback to data center operators enabling them to detect network bottlenecks and other performance related hardware issues.

Strategy

      Our objective is to be the leading provider of fiber optic subsystems and components and test systems to manufacturers and users of high-speed data networking and storage systems. Key elements of our strategy include the following:

      Maintain Technology Leadership in High-Speed Fiber Optic Transmission. We have been focused on the development of fiber optic subsystems since 1988. Finisar was actively involved in the original development of the Fibre Channel standard and, more recently, in the development and implementation of Gigabit Ethernet, 10GigE and iSCSI. Our years of engineering experience, our multi-disciplinary technical expertise and our participation in the development of industry standards have enabled us to become a leader in the design and development of fiber optic subsystems and test systems. We intend to maintain our technological leadership through continual enhancement of our existing products and the development of new products as evolving technology permits higher speed transmission of data, with greater capacity, over longer distances. We have been at the forefront of a number of important breakthroughs in the development of innovative products for fiber optic applications including the first 1Gbps 850nm transceiver for use in multimode fiber applications (1992), the first transceiver incorporating digital diagnostics (1995), the first 1 and 2.125Gbps 850nm transceivers based on the small form factor (2000 and 2001), the first CWDM GBIC transceiver (2001) and the first DWDM GBIC transceiver (2002). We have also been a pioneer of using the small form factor for 10GigE applications, or XFP, and shipped the first product under this new protocol in 2002. We are also focused on increased product integration to enhance the price/performance capabilities of

7

Table of Contents

      Leverage Core Competencies Across Multiple, High-Growth Markets. We believe that fiber optic technology will remain the transmission technology of choice for multiple data communication markets, including Gigabit and 10-Gigabit Ethernet-based LANs and MANs, Fibre Channel-based SANs and SONET-based MANs and WANs. These markets are characterized by differentiated applications with unique design criteria such as product function, performance, cost, in-system monitoring, size limitations, physical medium and software. We intend to target opportunities where our core competencies in high-speed data transmission protocols such as Gigabit Ethernet and Fibre Channel can be leveraged into leadership positions as these technologies are extended across multiple markets and applications.

      Strengthen and Expand Customer Relationships. Over the past 15 years, we have established valuable relationships and a loyal base of customers by providing high-quality products and superior service. Our service-oriented approach has allowed us to work closely with leading data and storage network system manufacturers, understand and address their current needs and anticipate their future requirements. We intend to leverage our relationships with our existing customers as they enter new, high-speed data communications markets and to establish relationships with new customers primarily for SONET-based communications applications.

      Acquire Critical Technologies. The ability to develop innovative products frequently requires that we control the critical underlying technologies and core competencies to be used in the development process. This enhances our ability to speed the development process as well as to protect any intellectual property that might be created in the process. This has been the primary motivation for the acquisitions that we have completed to date. We have acquired six companies and certain assets of two other companies since October 2000. We believe these acquisitions will enable us to respond more quickly to new market opportunities. These acquisitions have enabled us to:

	•	develop an internal capability for the automated assembly and testing of optical subsystems,
	•	create an internal capability for manufacturing certain active optical components such as Fabry-Perot, or FP, and distributed feedback, or DFB, lasers, and positive-intrinsic-negative photodiodes, or PINs, and avalanche photodiodes, or APDs; and
	•	create an internal capability for manufacturing certain passive optical products such as isolators, filters, splitters, quarter wave plates, interleavers and polarization beam combiners.

      Both active and passive component technologies have been incorporated into our product offerings for optical subsystems. We intend to expand the use of internal sourcing for many of these components during fiscal 2004. We believe that these acquisitions have also enhanced our position in testing and monitoring equipment for Fibre Channel, Gigabit Ethernet, 10-Gigabit Ethernet, iSCSI and FICON network protocols particularly for field deployed applications. We expect to continue to acquire new technologies that may enable us to expand our product offerings, introduce new innovative products or reduce our product costs.

      Develop Low Cost Manufacturing Capabilities. We believe that new markets can be created by the introduction of new low cost, high value-added products. Lower product costs can be achieved through the introduction of new technologies, product design or market presence. In each case, access to low-cost manufacturing resources are a key factor in the ability to offer a low-cost product solution. We have developed unique product designs and automated test processes that reduce the time to manufacture many of our products. In fiscal 2002, we purchased a manufacturing facility in Ipoh, Malaysia, in order to take advantage of low-cost off-shore labor while protecting access to our intellectual property and know-how. By the fourth quarter of fiscal 2002, most of our volume manufacturing was being conducted at this new facility. We anticipate that we will continue to manufacture low volume products at our facilities in the U.S. and China while continuing to rely on third-party manufacturers for a portion of our overall manufacturing requirements.

8

Table of Contents

Products

      In accordance with the guidelines established by the Statement of Financial Accounting Standards No. 131, “Disclosures about Segments of an Enterprise and Related Information” (“SFAS 131”), we have determined that, beginning in fiscal 2001, we operate in two segments: optical subsystems and components; and network test and monitoring systems .

      We provide a broad line of complementary products within each of these segments for high-speed data communications over Gigabit Ethernet LANs and MANs and Fibre Channel SANs.

      Optical data networks require optical subsystems that convert electrical signals into optical signals and back into electrical signals at high speeds. Our optical subsystems are integrated into our customers’ systems and used for both short- and intermediate-distance fiber optic communications.

      Our family of optical subsystem products consists of transmitters, receivers and transceivers principally based on the Gigabit Ethernet and Fibre Channel protocols and, to a lesser extent, on the SONET protocol. A transmitter converts electrical signals into optical signals for transmission over fiber optics. Receivers incorporating photodetectors convert incoming optical signals into electric signals. A transceiver combines both transmitter and receiver functions in a single device. Our optical subsystem products perform these functions with high reliability and data integrity and support a wide range of protocols, transmission speeds, fiber types, wavelengths, transmission distances, physical configurations and software enhancements.

      Our high-speed fiber optic subsystems are engineered to deliver value-added functionality and intelligence. Most of our optical subsystem products include a microprocessor with proprietary embedded software that allows customers to monitor transmitted and received optical power, temperature, drive current and other link parameters of each port on their systems in real time. In addition, our intelligent optical subsystems are used by many enterprise networking and storage system manufacturers to enhance the ability of their systems to diagnose and correct abnormalities in fiber optic networks.

      For storage applications which rely on the Fibre Channel standard, we provide optical subsystems for transmission applications at 1 and 2.125 Gbps. Data networking applications based on the Gigabit Ethernet standard continue to rely on devices which transmit signals at 1Gbps. The capability to transmit signals at 10Gbps is currently being developed. However, we believe that the adoption of such technologies will not occur on any significant scale until 2004. For SONET-based MANs, we supply optical subsystems which are capable of transmitting at 2.5 Gbps and intend to expand that product line to include products that operate at less than 1Gbps that we believe will continue to comprise a significant portion of spending for equipment to be deployed by carriers in this portion of their network.

      We have introduced a full line of optical subsystems for MANs using CWDM technologies designed to deliver dramatic cost savings to optical networking manufacturers, compared to solutions based on the use of DWDM technologies. DWDM systems are typically designed to add capacity in long haul telecommunications networks by using 32 or more wavelengths, spaced 1.6 nanometers, or nm, apart to transport data in a point-to-point or ring configuration. CWDM systems typically use only eight wavelengths, spaced 20 nm apart. While offering additional capacity, DWDM systems are far more complex than CWDM subsystems and must be cooled, further adding to the cost of such systems. Our CWDM subsystems include every major optical transport component needed to support a MAN, including transceivers, optical add/drop multiplexers, or OADMs, for adding and dropping wavelengths in a network without the need to convert to an electrical signal and multiplexers/demultiplexers for SONET, Gigabit Ethernet and Fibre Channel protocols. Where the need for additional bandwidth exists, we have introduced optical subsystems which incorporate DWDM technologies that allow these CWDM subsystem products to scale incrementally in terms of the amount of bandwidth handled, thus providing additional cost savings to network operators, whose customers are in the early stages of deploying new IP-based systems.

      With the acquisitions of Sensors Unlimited, Demeter Technologies, Transwave Fiber and Genoa Corporation, we gained access to leading-edge technology for the manufacture of a number of active and

9

Table of Contents

      Our test and monitoring systems allow engineers, service technicians and network managers to generate and capture data at high speeds, filter the data and identify various types of intermittent errors and other network problems for SANs, LANs, wireless networks, voice-over-internet protocol applications and newly emerging technologies including 10GigE and iSCSI.

      The test and monitoring systems sold for Fibre Channel applications consist principally of analyzers, generators, bit-error-rate testers, or BERTs, and “Jammer” systems sold to Fibre Channel development groups to quickly debug and test switches and disk array products. An analyzer is used to capture data traffic into a large memory buffer so that the data can be analyzed by developers to detect problems on a Fibre Channel network. A generator is used to generate Fibre Channel traffic to stress a Fibre Channel network and is typically used in combination with an analyzer. Our BERT product sends, receives, and compares bit patterns on a Fibre Channel network while the Jammer product injects errors into a Fibre Channel network in order to simulate how the network responds and recovers from such problems. In addition, our SAN Metrics product is the first product to deliver expert analysis for Fibre Channel networks in a field environment. SAN Metrics speeds an engineer through the troubleshooting process by automatically analyzing captured traces to identify problems. In response to the newly emerging technologies of Infiniband and iSCSI, we have developed an iSCSI analyzer and InfiniBand analyzer which enable our customers to develop new SAN products for multi-protocol environments. We provide testing, training and software development services primarily for Fibre Channel applications through our Medusa Labs facility.

      We also build LAN analyzers and monitoring systems for Gigabit Ethernet networks which are used by network administrators to monitor and troubleshoot their networks. The analyzer captures Ethernet traffic while our Surveyor Expert examines the captured traffic to identify network problems. For Voice over Internet Protocol, or VoIP, applications, our Multi QoS product examines voice traffic on an Ethernet network and builds a table of quality metrics for each call. Our recently announced product for wireless networks, Surveyor Wireless, captures Ethernet wireless traffic for analysis, troubleshooting, and monitoring.

      We have recently introduced the first of a series of products designed to consolidate many of our test and monitoring products for both Fibre Channel and Ethernet applications onto a single hardware platform which we believe will reduce product costs and enhance our competitive position. We expect that our entire product line for these applications will be transferred to this new platform by the end of fiscal 2004.

Customers

      To date, our revenues have been principally derived from sales to equipment manufacturers who sell products for building LAN and SAN networks. Sales to these customers accounted for 68% of our total revenues in fiscal 2001, 66% in fiscal 2002 and 74% in fiscal 2003. Sales to our top three customers represented approximately 48% of our total revenues in fiscal 2001, 31% in fiscal 2002 and 28% in fiscal 2003. Sales to our top three customers, Brocade, EMC Corporation and Emulex accounted for 20%, 17% and 11% of our total revenues, respectively, in fiscal 2001. Sales to our top two customers, EMC Corporation and Emulex, accounted for 12% and 11%, respectively, in fiscal 2002. Sales to Cisco accounted for 10% of our total revenues in fiscal 2003. No other customer accounted for 10% of revenues in any of these years.

10

Table of Contents

Technology

      The development of high quality fiber optic subsystems and components and network performance test systems for high-speed data communications requires multidisciplinary expertise in the following technology areas:

	High Frequency Semiconductor Design. Our fiber optic subsystems development efforts are supported by an engineering team that specializes in analog/digital integrated circuit design. This group works in both silicon and gallium arsenide, or GaAs, semiconductor technologies where circuit element frequencies are very fast and can be as high as 60 gigahertz, or GHz. We have designed proprietary circuits including laser drivers and receiver pre- and post-amplifiers. Our designs have made us early entrants in the 1.0 Gbps data communications market and more recently in the 2.5 Gbps data communications market. These advanced semiconductor devices provide significant cost advantages and will be critical in the development of future products capable of even faster data rates.
	Optical Subsystem Design. We have established ourselves as a low-cost design leader beginning with our initial Gbps optical subsystems in 1992. From that base we have developed new single-mode laser alignment approaches and low-cost, all-metal packaging techniques for improved EMI performance and environmental tolerance. We develop our own component and packaging and designs and integrate these designs with proprietary manufacturing processes that allow our products to be manufactured in high volume.
	We enhanced our capability to develop and integrate new optical components into our optical subsystems through the acquisition of Transwave Fiber in fiscal 2002 and the acquisition of certain assets and intellectual property from AIFOtec GmbH in fiscal 2002 and New Focus in fiscal 2003.
	The acquisition of Transwave Fiber provided us with the capability for making passive optical components used in products that multiplex and demultiplex signals in CWDM subsystems to eliminate bandwidth bottlenecks and expand the performance of MANs. These products include wavelength division multiplexer couplers and isolator products. WDM couplers are used to split and combine signals in optical networks and rely on the use of thin-film filters, fused fiber couplers, microlenses and/or special optical materials. Isolator products are used to cause light signals in a network to propagate in one direction within a network, but prevent that signal from returning in the opposite direction.
	The acquisition of certain assets and intellectual property from AIFOtec provided access to a new approach to producing optical subassemblies using micro-module technology which promises to lower the cost to produce and test components at the wafer level. We also gained the capability for manufacturing automated laser welding equipment currently used in the manufacture of optical subassemblies for many of our existing transceiver products.
	With the purchase of certain assets from New Focus, we expanded our product offering of optical components to include circulators and interleavers. Circulators are similar to isolators in that they cause light in a system to flow in only one direction, but are different in that circulators incorporate multiple ports and use these multiple ports to perform routing functions within the network. Interleavers provide a means of segregating wavelengths of light in DWDM systems such that they can be more easily controlled. We also produce tunable narrow-bandpass filters that are wavelength-tunable by voltage control. These components were sold in limited quantities into the merchant market in fiscal 2003.
	Complex Logic Design. Our network performance test equipment designs are based on field programmable gate arrays, or FPGAs. In recent customer trials, our newest products are being used to operate with clock frequencies of up to 125 megahertz, or MHz, and logic densities up to 1 million gates per chip. Our test systems use FPGAs that are programmed by the host PC and therefore can be configured differently for different tests. All of our logic design is done in the very high density logic, or VHDL, hardware description language which will enable migration to application specific integrated circuits, or ASICs, as volumes warrant. We develop VHDL code in a modular fashion for reuse in logic design which comprises a critical portion of our intellectual property. This re-usable technology base of

11

Table of Contents

	logic design is available for use in both our test system and optical subsystem product lines and allows us to reduce the time to market for our new and enhanced products.
	Software Technology. We devote substantial engineering resources to the development of software technology for use in all of our product lines. We have developed software to control our test systems, analyze data collected by our test systems, and monitor, maintain, test and calibrate our optical subsystems. A majority of our software technology and expertise is focused on the use of object-oriented development techniques to develop software subsystems that can be reused across multiple product lines. We have created substantial intellectual property in the area of data analysis software for our Fibre Channel test equipment. This technology allows us to rapidly sort, filter and analyze large amounts of data using a proprietary database format. This database format is both hardware platform-independent and protocol-independent. This independence allows all of the software tools developed for our existing test products to be utilized in all of our new test products that collect data traces. Because the database format is also protocol-independent, new protocols can be added quickly and easily. Another important component of our intellectual property is our graphical user interface, or GUI, design. Many years of customer experience with our test products have enabled us to define a simple yet effective method to display complex protocols in clear and concise GUIs for intuitive use by engineers.
	The acquisition of Shomiti Systems in fiscal 2001 strengthened our base of software technology by adding expertise in the development of test and monitoring products based on the Gigabit Ethernet protocol. The acquisition of Medusa Technologies in fiscal 2001 provided additional software engineering expertise related to testing system products manufactured by customers of our transceiver products as well a capability for training software engineers both within and outside Finisar.
	System Design. The design of all of our products requires a combination of sophisticated technical competencies — optical engineering, high-speed digital and analog design, ASIC design and software engineering. We have built an organization of people with skills in all of these areas. It is the integration of these technical competencies that enables us to produce products that meet the needs of our customers. Our combination of these technical competencies has enabled us to design and manufacture optical subsystems with built-in optical test multiplexing and network monitoring, as well as test systems that integrate optical and protocol testing with user interface software.
	Manufacturing System Design. The design skills gained in our test systems group are also used in the manufacturing of our optical subsystems. We utilize our high-speed FPGA design blocks and concepts and GUI software elements to provide specialized manufacturing test systems for our internal use. These test systems are optimized for test capacity and broad test coverage. We use automated, software-controlled testing to enhance the field reliability of all Finisar products. All of our products are subjected to temperature testing of powered systems as well as full functional tests.
	Wafer Fabrication. The acquisitions of Sensors Unlimited and Demeter Technologies in fiscal 2001 provided new capabilities with respect to developing and producing active devices requiring the use of indium phosphide semiconductor materials. The acquisition of Sensors Unlimited provided expertise in the production of PIN receivers at 2.5 and 10 Gbps and APDs that are used primarily in our CWDM and DWDM transceiver products to enhance their sensitivity and performance. The acquisition of Demeter Technologies added the capability for making FP lasers at 1.25 and 2.5Gbps and DFB lasers at 1.25, 2.5 and 10 Gbps that are incorporated into many of our transceiver designs. To date, use of these components in our own transceiver products and their sales in the merchant market has been limited. As a result our acquisition of Genoa Corporation in the fourth quarter of fiscal 2003, we will be consolidating our wafer fabrication operations formerly conducted at the facilities of our subsidiaries, Sensors Unlimited and Demeter Technologies, into the former Genoa facility in Fremont, California. The use of components previously manufactured in the Sensors Unlimited and the Demeter Technologies facilities in our transceiver products and optical subsystems will be greatly expanded once these capabilities have been transferred to the Fremont facility during fiscal 2004.
	The acquisition of Genoa Corporation not only provided a state-of-the-art foundry for manufacturing the active and passive components formerly manufactured at our subsidiaries, Sensors Unlimited and Demeter Corporation, but also provided us access to two additional technologies. Genoa developed the

12

Table of Contents

world’s first linear semiconductor optical amplifier that is currently being sold in limited quantities primarily for evaluation by a number of customers. Use of this technology to amplify light at the chip level, in conjunction with other optical components we recently acquired from New Focus, should enable us to offer a distinct cost and performance advantage compared to erbium-doped fiber optic amplifiers, or EDFAs. Efficient light amplification is an important requirement of DWDM in MANs. In addition, Genoa developed a novel approach for making long wavelength VCSELs at 1310 nm and 1550 nm that should result in improved reliability and cost. Revenues from the sale of products related to those technologies are not expected to be significant until fiscal 2005.

Competition

      The market for optical subsystems and components and network test and monitoring systems for use in LANs, SANs and MANs is highly competitive. We believe the principal competitive factors in the optical subsystem and test system markets are:

	•	product performance, features, functionality and reliability;
	•	price/performance characteristics;
	•	timeliness of new product introductions;
	•	adoption of emerging industry standards;
	•	service and support;
	•	size and scope of distribution network;
	•	brand name;
	•	access to customers; and
	•	size of installed customer base.

      We believe we compete favorably with our competitors with respect to most of the foregoing factors. However, we cannot assure you that we will be able to compete successfully against either current or future competitors.

Sales, Marketing and Technical Support

      We sell our products in North America through our direct sales force and a network of independent manufacturers’ representatives. For sales of our optical subsystems and components, we utilize a direct sales force augmented by 13 domestic manufacturers’ representatives and 11 international resellers. For sales of our network test and monitoring systems, we utilize a direct sales force augmented by nine domestic manufacturers’ representatives and 30 international resellers. Our direct sales force maintains close contact with our customers and provides technical support to our manufacturers’ representatives. In our international markets, our direct sales force works with local resellers who assist us in providing support and maintenance to the territories they cover.

      Our marketing efforts are focused on increasing awareness of our product offerings for optical subsystems and network test and monitoring systems and our brand name. Key components of our marketing efforts include:

	•	continuing our active participation in industry associations and standards committees to promote and further enhance Gigabit Ethernet and Fibre Channel technologies, promote standardization in the LAN, SAN and MAN markets, and increase our visibility as industry experts; and
	•	leveraging major trade show events and LAN, SAN, and MAN conferences to promote our broad product lines.

      In addition, our marketing group provides marketing support services for our executive staff, our direct sales force and our manufacturers’ representatives and resellers. Through our marketing activities, we provide

13

Table of Contents

      A high level of continuing service and support is critical to our objective of developing long-term customer relationships. We emphasize customer service and technical support in order to provide our customers and their end users with the knowledge and resources necessary to successfully utilize our product line. Our customer service utilizes a technical team of field and factory applications engineers, technical marketing personnel and, when required, product design engineers. We provide extensive customer support throughout the qualification and sale process. In addition, we also provide many resources through our World Wide Web site, including product documentation and technical information. We intend to continue to provide our customers with comprehensive product support and believe it is critical to remaining competitive.

Manufacturing

      During fiscal 2002, we transitioned most of our manufacturing, assembly and test operations from a number of Asia-based contract manufacturers to our own manufacturing facility in Malaysia which we purchased in May 2001. This facility consists of 640,000 square feet, of which 240,000 square feet is suitable for cleanroom operations. The acquisition of this facility has allowed us to transfer most of our manufacturing processes to a lower-cost manufacturing facility and to maintain greater control over our intellectual property. We expect to continue to use contract manufacturers for a portion of our manufacturing needs. During fiscal 2003, we conducted manufacturing engineering, supply chain management, quality assurance and documentation control operations primarily at our facility in Sunnyvale, California, and wafer fabrication operations at our subsidiaries’ facilities located in El Monte, California, Fremont, California, and Shanghai, China. During fiscal 2004, our facilities in El Monte will be closed and certain assets, personnel and intellectual property transferred to our new Fremont facility.

      We design and develop a number of the key components of our products, including photodetectors, lasers, ASICs, printed circuit boards and software. In addition, our manufacturing team works closely with our engineers to manage the supply chain. To assure the quality and reliability of our products, we conduct product testing and burn-in at our facilities in conjunction with inspection and the use of testing and statistical process controls. In addition, most of our optical subsystems have an intelligent interface that allows us to monitor product quality during the manufacturing process. Our facilities in Sunnyvale and Malaysia are qualified under ISO 9001-9002.

      Although we use standard parts and components for our products where possible, we currently purchase a few key components used in the manufacture of our products from single or limited sources. Our principal single source components include ASICs, and DFB lasers. Generally, purchase commitments with our single or limited source suppliers are on a purchase order basis. Any interruption or delay in the supply of any of these components, or the inability to procure these components from alternate sources at acceptable prices and within a reasonable time, would substantially harm our business. In addition, qualifying additional suppliers can be time-consuming and expensive and may increase the likelihood of errors.

      We use a rolling 12-month forecast based on anticipated product orders to determine our material requirements. Lead times for materials and components we order vary significantly, and depend on factors such as the specific supplier, contract terms and demand for a component at a given time.

Research and Development

      In fiscal 2001, fiscal 2002 and fiscal 2003, our research and development expenses were $33.7 million, $54.4 million, and $60.3 million, respectively. We believe that our future success depends on our ability to continue to enhance our existing products and to develop new products that maintain technological competitiveness. We focus our product development activities on addressing the evolving needs of our customers within the LAN, SAN and MAN markets. We work closely with our original equipment manufacturers and system integrators to monitor changes in the marketplace. We design our products around current industry standards and will continue to support emerging standards that are consistent with our

14

Table of Contents

      We are currently undertaking development efforts for our product lines with emphasis on increasing reliability, integrity and performance, as well as value-added functions. Some examples of products that we are working on include 10 Gbps Ethernet and CWDM and inexpensive DWDM optical subsystems. We also intend to focus on increased product integration to enhance the price/ performance capabilities of our products. We believe that our research and development efforts are key to our ability to maintain technical competitiveness and to deliver innovative products that address the needs of the market. However, there can be no assurance that our product development efforts will result in commercially successful products, or that our products will not be rendered obsolete by changing technology or new product announcements by other companies.

Intellectual Property

      Our success and ability to compete is dependent in part on our proprietary technology. We rely on a combination of patent, copyright, trademark and trade secret laws, as well as confidentiality agreements and licensing arrangements, to establish and protect our proprietary rights. To date, we are the owner of 54 issued U.S. patents and, as of July 1, 2003, we have 340 patent applications pending in the United States Patent and Trademark Office. We cannot assure you that any patents will issue as a result of pending patent applications or that our issued patents will be upheld. Any infringement of our proprietary rights could result in significant litigation costs, and any failure to adequately protect our proprietary rights could result in our competitors offering similar products, potentially resulting in loss of a competitive advantage and decreased revenues. Despite our efforts to protect our proprietary rights, existing patent, copyright, trademark and trade secret laws afford only limited protection. In addition, the laws of some foreign countries do not protect our proprietary rights to the same extent as do the laws of the United States. Attempts may be made to copy or reverse engineer aspects of our products or to obtain and use information that we regard as proprietary. Accordingly, we may not be able to prevent misappropriation of our technology or deter others from developing similar technology. Furthermore, policing the unauthorized use of our products is difficult. Litigation may be necessary in the future to enforce our intellectual property rights or to determine the validity and scope of the proprietary rights of others. This litigation could result in substantial costs and diversion of resources and could significantly harm our business.

      The networking industry is characterized by the existence of a large number of patents and frequent litigation based on allegations of patent infringement. We have previously been involved in a series of patent infringement lawsuits. From time to time, other parties may assert patent, copyright, trademark and other intellectual property rights to technologies and in various jurisdictions that are important to our business. Any claims asserting that our products infringe or may infringe proprietary rights of third parties, if determined adversely to us, could significantly harm our business. Any claims, with or without merit, could be time-consuming, result in costly litigation, divert the efforts of our technical and management personnel, cause product shipment delays or require us to enter into royalty or licensing agreements, any of which could significantly harm our business. Royalty or licensing agreements, if required, may not be available on terms acceptable to us, if at all. In addition, our agreements with our customers typically require us to indemnify our customers from any expense or liability resulting from claimed infringement of third party intellectual property rights. In the event a claim against us was successful and we could not obtain a license to the relevant technology on acceptable terms or license a substitute technology or redesign our products to avoid infringement, our business would be significantly harmed.

Employees

      As of April 30, 2003, we employed approximately 2,040 full-time employees. We also from time to time employ part-time employees and hire contractors. Our employees are not represented by any collective

15

Table of Contents

Factors That Could Affect Our Future Performance

      OUR FUTURE PERFORMANCE IS SUBJECT TO A VARIETY OF RISKS. IF ANY OF THE FOLLOWING RISKS ACTUALLY OCCUR, OUR BUSINESS COULD BE HARMED AND THE TRADING PRICE OF OUR COMMON STOCK COULD DECLINE. YOU SHOULD ALSO REFER TO THE OTHER INFORMATION CONTAINED IN THIS REPORT, INCLUDING OUR CONSOLIDATED FINANCIAL STATEMENTS AND THE RELATED NOTES.

      Our quarterly and annual operating results have fluctuated substantially in the past and are likely to fluctuate significantly in the future due to a variety of factors, some of which are outside of our control. Accordingly, we believe that period-to-period comparisons of our results of operations are not meaningful and should not be relied upon as indications of future performance. Some of the factors that could cause our quarterly or annual operating results to fluctuate include market acceptance of our products and the Gigabit Ethernet and Fibre Channel standards, market demand for the products manufactured by our customers, the introduction of new products and manufacturing processes, manufacturing yields, competitive pressures and customer retention.

      We may experience a delay in generating or recognizing revenues for a number of reasons. Orders at the beginning of each quarter typically represent a small percentage of expected revenues for that quarter and are generally cancelable at any time. Accordingly, we depend on obtaining orders during each quarter for shipment in that quarter to achieve our revenue objectives. Failure to ship these products by the end of a quarter may adversely affect our operating results. Furthermore, our customer agreements typically provide that the customer may delay scheduled delivery dates and cancel orders within specified time frames without significant penalty. Because we base our operating expenses on anticipated revenue trends and a high percentage of our expenses are fixed in the short term, any delay in generating or recognizing forecasted revenues could significantly harm our business. It is likely that in some future quarters our operating results will again decrease from the previous quarter or fall below the expectations of securities analysts and investors. In this event, the trading price of our common stock would significantly decline.

      As a result of the sale of our 5 1/4% convertible subordinated notes in October 2001, we incurred $125 million of indebtedness, substantially increasing our ratio of debt to total capitalization. We may incur additional indebtedness in the future. The level of our indebtedness, among other things, could:

	•	make it difficult for us to make payments on the notes;
	•	make it difficult for us to obtain any necessary future financing for working capital, capital expenditures, debt service requirements or other purposes;
	•	limit our flexibility in planning for, or reacting to changes in, our business; and
	•	make us more vulnerable in the event of a downturn in our business.

      We will be required to generate cash sufficient to pay our indebtedness and other liabilities, including interest payable on the notes, and to conduct our business operations. We may not be able to cover our anticipated debt service obligations from our cash flow. This may materially hinder our ability to make payments on the notes. We must also generate sufficient cash to repay the principal of the notes which is payable in October 2008. Our ability to meet our future debt service obligations will depend upon our future

16

Table of Contents

      We believe that our existing balances of cash, cash equivalents and short-term investments, together with the cash expected to be generated from our future operations, will be sufficient to meet our cash needs for working capital and capital expenditures for at least the next 12 months. We may however require additional financing to fund our operations in the future or to repay the principal of our 5 1/4% convertible subordinated notes, due in October 2008. The significant contraction in the capital markets, particularly in the technology sector, may make it difficult for us to raise additional capital if and when it is required, especially if we continue to experience disappointing operating results. If adequate capital is not available to us as required, or is not available on favorable terms, we could be required to significantly reduce or restructure our business operations.

      We base many of our operating decisions, and enter into purchase commitments, on the basis of anticipated revenue trends which are highly unpredictable. Some of our purchase commitments are not cancelable, and in some cases we are required to recognize a charge representing the amount of material or capital equipment purchased or ordered which exceed our actual requirements. In the past, we have experienced significant growth followed by a significant decrease in customer demand such as occurred between the quarters ended July 31, 2000 and July 31, 2001, when quarterly revenues fluctuated from $27.2 million to $64.8 million followed by a decrease to $34.2 million. Based on projected revenue trends during these periods, we acquired inventories and entered into purchase commitments in order to meet anticipated increases in demand for our products which did not materialize. As a result, we recorded significant charges for obsolete and excess inventories and non-cancelable purchase commitments which contributed to substantial operating losses in fiscal 2001 and 2002. We recorded additional charges for obsolete and excess inventories during fiscal 2003, due to unanticipated changes in demand and the mix for our products. Should revenue in future quarters again fall substantially below our expectations, or should we fail again to accurately forecast changes in demand mix, we could be required to record additional charges for obsolete or excess inventories or non-cancelable purchase commitments.

      We experienced a significant decline in revenues and operating results during fiscal 2002. While revenues recovered to some extent in fiscal 2003, they have not yet reached levels required to operate on a profitable basis due primarily to higher fixed costs related to a number of acquisitions, low gross margins and continued high levels of spending for research and development in anticipation of future revenue growth. While we continue to expect future revenue growth, we have taken steps to reduce our operating costs in order to conserve our cash and accelerate our return to profitability, and we may be required to take further action to reduce costs. These cost reduction measures may adversely affect our ability to market our products, introduce new and improved products and increase our revenues, which could adversely affect our business and cause the price of our stock to decline. In order to be successful in the future, we must continue to reduce our operating expenses and product costs, while at the same time completing our key product development programs and penetrating new customers.

17

Table of Contents

      We have based our product offerings principally on Fibre Channel and Gigabit Ethernet standards and, to a lesser extent, the SONET standard and our future success is substantially dependent on the continued market acceptance of these standards. If an alternative technology is adopted as an industry standard within our target markets, we would have to dedicate significant time and resources to redesign our products to meet this new industry standard. Our products comprise only a part of an entire networking system, and we depend on the companies that provide other components to support industry standards as they evolve. The failure of these companies, many of which are significantly larger than we are, to support these industry standards could negatively impact market acceptance of our products. Because we may develop some products prior to the adoption of industry standards, we may develop products that do not comply with the eventual industry standard. Our failure to develop products that comply with industry standards would limit our ability to sell our products. Moreover, if we introduce a product before an industry standard has become widely accepted, we may incur significant expenses and losses due to lack of customer demand, unusable purchased components for these products and the diversion of our engineers from future product development efforts. Finally, if new standards evolve, we may not be able to successfully design and manufacture new products in a timely fashion, if at all, that meet these new standards.

      In the United States, our products must comply with various regulations and standards defined by the Federal Communications Commission and Underwriters Laboratories. Internationally, products that we develop also will be required to comply with standards established by local authorities in various countries. Failure to comply with existing or evolving standards established by regulatory authorities or to obtain timely domestic or foreign regulatory approvals or certificates could significantly harm our business.

              We are dependent on widespread market acceptance of two product families, and our revenues will decline if the market does not continue to accept either of these product families.

      We currently derive substantially all of our revenue from sales of our optical subsystems and network test and monitoring systems. We expect that revenue from these products will continue to account for substantially all of our revenue for the foreseeable future. Accordingly, widespread acceptance of these products is critical to our future success. If the market does not continue to accept either our optical subsystems or our network test and monitoring systems, our revenues will decline significantly. Factors that may affect the market acceptance of our products include the continued growth of the markets for LANs, SANs, and MANs and, in particular, Gigabit Ethernet and Fibre Channel-based technologies as well as the performance, price and total cost of ownership of our products and the availability, functionality and price of competing products and technologies.

      Many of these factors are beyond our control. In addition, in order to achieve widespread market acceptance, we must differentiate ourselves from the competition through product offerings and brand name recognition. We cannot assure you that we will be successful in making this differentiation or achieving widespread acceptance of our products. Failure of our existing or future products to maintain and achieve widespread levels of market acceptance will significantly impair our revenue growth.

      A small number of customers have accounted for a significant portion of our revenues. Our success will depend on our continued ability to develop and manage relationships with significant customers. Although we are attempting to expand our customer base, we expect that significant customer concentration will continue for the foreseeable future.

      The markets in which we sell our products are dominated by a relatively small number of systems manufacturers, thereby limiting the number of our potential customers. Our dependence on large orders from a relatively small number of customers makes our relationship with each customer critically important to our

18

Table of Contents

      Typically, we do not have long-term contracts with our customers. As a result, our agreements with our customers do not provide any assurance of future sales. Accordingly:

	•	our customers can stop purchasing our products at any time without penalty;
	•	our customers are free to purchase products from our competitors; and
	•	our customers are not required to make minimum purchases.

      Sales are typically made pursuant to individual purchase orders, often with extremely short lead times. If we are unable to fulfill these orders in a timely manner, we will lose sales and customers.

      The markets for our products are characterized by rapid technological change, frequent new product introductions, changes in customer requirements and evolving industry standards. We expect that new technologies will emerge as competition and the need for higher and more cost effective bandwidth increases. Our future performance will depend on the successful development, introduction and market acceptance of new and enhanced products that address these changes as well as current and potential customer requirements. The introduction of new and enhanced products may cause our customers to defer or cancel orders for existing products. In addition, a slowdown in demand for existing products ahead of a new product introduction could result in a writedown in the value of inventory on hand related to existing products. We have in the past experienced a slowdown in demand for existing products and delays in new product development and such delays may occur in the future. To the extent customers defer or cancel orders for existing products due to a slowdown in demand or in the expectation of a new product release or if there is any delay in development or introduction of our new products or enhancements of our products,