Back to GetFilings.com

Table of Contents

UNITED STATES

SECURITIES AND EXCHANGE COMMISSION

Washington, D.C. 20549

FORM 10-K

(Mark One)

For the fiscal year ended March 31, 2004

or

For the transition period from              to             

COMMISSION FILE NO.: 0-33213

MAGMA DESIGN AUTOMATION, INC.

(Exact name of Registrant as specified in its charter)

5460 Bayfront Plaza

Santa Clara, California 95054

(408) 565-7500

(Address, including zip code, and telephone number, including area code, of the registrant’s principal executive offices)

SECURITIES REGISTERED PURSUANT TO SECTION 12(b) OF THE ACT:

None

SECURITIES REGISTERED PURSUANT TO SECTION 12(g) OF THE ACT:

COMMON STOCK, par value $.0001 per share

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  x    No  ¨

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 the 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.  ¨

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

The aggregate market value of the registrant’s common stock held by non-affiliates of the registrant, based upon the closing sale price of the Common Stock on September 30, 2003 as reported on the Nasdaq National Market, was $636,120,000. This calculation does not reflect a determination that certain persons are affiliates of the Registrant for any other purpose.

As of May 28, 2004 Registrant had outstanding 34,165,152 shares of Common Stock, $.0001 par value.

DOCUMENTS INCORPORATED BY REFERENCE

Portions of the Registrant’s proxy statement to be delivered to the stockholders in connection with Registrant’s 2004 Annual Meeting of Stockholders to be held on August 31, 2004, are incorporated by reference into Part III of this Form 10-K. The Registrant’s proxy statement is required to be filed within 120 days after the Registrant’s fiscal year end.

Table of Contents

MAGMA DESIGN AUTOMATION, INC.

FISCAL 2004 FORM 10-K

TABLE OF CONTENTS

Magma, Blast Fusion, Blast Noise and FixedTiming are registered trademarks, and ArchEvaluator, Blast RTL, Blast Fusion APX, Blast Plan, Blast Rail, “The Fastest Path from RTL to Silicon,” and PALACE are trademarks, of Magma Design Automation. All other product and company names are trademarks and registered trademarks of their respective companies.

1

Table of Contents

PART I

ITEM 1.    BUSINESS.

Overview

Magma Design Automation, Inc. provides electronic design automation, or EDA, software products and related services. Our software enables chip designers to reduce the time it takes to design and produce complex integrated circuits used in the communications, computing, consumer electronics, networking and semiconductor industries. Our products are used in all major phases of the chip development cycle, from initial design through physical implementation.

Our software products allow chip designers to meet critical time-to-market objectives, improve chip performance and handle chip designs involving millions of components. Blast Create^™ enables logic designers to visualize, evaluate and improve code quality, design constraints, testability and analysis. Blast Create, Blast Fusion^® and Blast Fusion APX^™ combine into one integrated chip design flow what traditionally had been separate logic design and physical design processes. This integrated flow significantly reduces timing closure iterations, allowing our customers to accelerate the time it takes to design and produce deep submicron integrated circuits. Blast Plan^™ enables hierarchical planning and partitioning of a design into blocks that can be designed separately and later combined into a complex chip or system on a chip. Blast Noise^® detects and corrects signal interference, or crosstalk, in physical designs. Blast Rail^™ is a correct-by-construction rail design solution that is integrated with our design implementation flow.

We provide consulting, training and services to help our customers more rapidly adopt our technology. We also provide post-contract support, or maintenance, for our products.

Evolution of the Electronic Design Automation Market

The trend toward deep submicron and system-on-chip designs has driven demand for improved electronic design automation software that enables the efficient design and implementation of these complex chips. Limitations in traditional electronic design automation technology could slow the adoption of deep submicron processes due to the difficulty in implementing designs at these small feature sizes. Historically, electronic design automation companies developed software for use by separate engineering groups to address either the front-end chip design or back-end chip implementation processes.

In the front-end design process, the chip design is conceptualized and written as a register transfer level computer program, or RTL file, that describes the required functionality of the chip. For large chips, the design is often divided into a number of individual blocks, each with its own associated RTL file. This is often done because of capacity limitations in existing electronic design automation tools. The designer also develops constraints for the design that are used to describe the desired timing performance of the chip. Finally, a target library is specified that contains detailed information about the basic functional building blocks, or logic gates, that will be used in the design. This library is typically provided by the semiconductor vendor or a third party library vendor. The next step is to run the RTL files through synthesis software that generates a netlist. The netlist describes the circuit in terms of logic gates selected from the target library and connected such that the functionality specified in the RTL files is realized. The synthesis software also performs optimizations to attempt to meet the timing constraints specified by the designer.

A critical objective of chip design is to minimize total circuit delay, which is comprised of gate delay and wire delay. Front-end software was initially developed when the gate delay, or the time it takes for an electrical signal to travel through a logic gate, was the most significant component of total circuit delay. Wire delay, or the time it takes for a signal to travel through a wire connecting two or more gates, was negligible and designers could use simple estimates and still meet targeted circuit speeds.

2

Table of Contents

In the back-end implementation process, physical design software is used to transform the netlist generated by the front-end process into a physical layout of the chip. The resulting physical layout is usually output in a binary file format, commonly referred to as GDSII, that is used to generate the photomasks used to manufacture the integrated circuit. The two primary functions provided by traditional physical design software are placement and routing. Placement determines the optimal physical location for the logic gates on the integrated circuit. After placement is completed, routing connects the logic gates with wires to achieve the desired circuit functionality. After the layout is completed, the final step in the back-end process is to run timing analysis to verify that the chip will run at the desired circuit speed. If circuit speeds are slower than the speeds reported by the synthesis software, the design must often be iterated back through the synthesis step in an attempt to improve the timing. Since each timing closure iteration cycle can take one or more weeks, successive iterations of the design process can significantly lengthen the time it takes to design and produce new chips.

Deep Submicron Challenges

The trend toward deep submicron technology has rendered traditionally separate front-end and back-end electronic design automation processes less effective for rapid, cost-effective and reliable chip designs. As integrated circuits have increased in complexity and feature sizes have dropped, the problems faced by chip designers have changed. Wire delay now accounts for the majority of total circuit delay and has become the most significant factor in circuit performance for deep submicron technologies. Front-end estimates of wire delay may vary considerably from actual wire delays measured in the final layout. As a result, the front-end timing might meet the design requirements, but the final layout timing at the completion of the back-end process may be unacceptable, requiring time-consuming iterations back through the front-end process.

Deep submicron process technologies bring additional complexities to the design and implementation process that can cause chip failures. These include signal integrity problems such as electrical interference from wires in close proximity, commonly referred to as crosstalk or noise, that can affect both circuit performance and functionality. Using existing design flows and software, designers must contend with analyzing and fixing these problems manually after the layout is completed. These adjustments often change the chip timing and further contribute to the timing closure problem.

These deep submicron challenges make it difficult to efficiently design chips using separate front-end and back-end processes. Semiconductor manufacturers and electronic products companies are currently seeking alternatives to older generation electronic design automation software in order to shorten design time, improve circuit speed, and handle larger chip designs. As a result, a significant opportunity exists for a new electronic design automation approach to chip design that can enable the design of more complex deep submicron integrated circuits, improve performance, and significantly reduce the time it takes to design and produce next generation electronic products.

Our Solution

An important technical foundation of our software products is our patented FixedTiming^® methodology, which allows our customers to reduce the number of iterations that are often required in conventional integrated circuit (“IC”) design processes. Our unified data model architecture is a key enabler for this methodology and for our ability to deliver automated signal integrity detection and correction. It contains logical and physical information about the design and is resident in core memory during execution, which makes it possible to analyze the design and make rapid tradeoff decisions during the physical design process.

Technology

FixedTiming Methodology

Our patented FixedTiming methodology allows us to reduce the timing closure iterations that are often required between the front-end and back-end processes in conventional integrated circuit design flows. These timing closure iterations are caused by the fact that the final circuit timing cannot be accurately calculated until

3

Table of Contents

the physical layout is completed. In deep submicron integrated circuits, timing performance is primarily determined by the physical layout of the wiring that connects the logic gates to achieve the desired circuit functionality. Timing that is estimated during the front-end process is often not realized in the final layout, and the design team must iterate between the front-end and back-end processes, modifying the design in an attempt to reach the desired timing performance. Since each timing closure iteration can add one or more weeks to the design cycle, the time it takes to design and produce an integrated circuit can be severely affected.

Our FixedTiming methodology is designed to predict circuit speeds prior to detailed physical design. We then use a series of design refinements during physical design to achieve a final timing that is very close to the predicted circuit speed. This approach reduces the need for timing closure iterations that exist in conventional flows and can significantly reduce the time it takes to design and produce deep submicron integrated circuits.

There are several differences between the conventional approach to integrated circuit design and our FixedTiming methodology. In the conventional flow, synthesis is used to transform a computer program description of the desired circuit functionality into a circuit-level description, or netlist, that is comprised of gates from a semiconductor manufacturer’s library. A gate is a basic building block that performs a specific logic function. Gates are typically available in different sizes, or drive strengths, in the library. Larger gates are required to drive large loads, which are caused by long wires or wires that are connected to the inputs of many other gates. Smaller gates are used to drive smaller loads. For a given wire, the larger the size of the gate, the shorter the signal delay through the gate and the wire that it is driving. The job of the synthesis tool is to produce a netlist that delivers the desired circuit functionality and meets the required circuit timing. The synthesis tool produces this netlist without knowing what the final layout will look like. Since the synthesis tool must determine which size gates to choose from the library, it must either rely on statistical estimates of the wire loads or perform a coarse placement of the gates to build estimates of what the wiring might look like. In both of these cases, the estimates often do not correlate well with the actual loads presented by the wires in the final layout.

Following synthesis, the gates specified in the netlist are placed in the layout. If the actual load on a given gate is larger than the load that was estimated during synthesis, the delay will be longer than was predicted by synthesis. If the particular gate and load are critical to the performance of the integrated circuit, this will limit the operating speed of the integrated circuit and force the design team into timing closure iterations. Typically, there are many of these critical paths on a complex integrated circuit that must be addressed.

Our FixedTiming methodology recognizes that wire loads cannot be accurately estimated prior to layout. Because of this, we do not choose gate sizes during the synthesis process. Instead we rely on the use of placeholder gates, called SuperCells, that we create automatically by analyzing the vendor’s library. Each SuperCell is just like a gate from the library, but we assume that its size is completely flexible. Therefore only one SuperCell is required for each logic function in the library, rather than the collection of gates of different sizes that are required in the conventional approach.

Before beginning physical layout, we apply our optimization technology to determine and set the delays that each gate and its load must have to meet the desired circuit speed. During placement, we use the SuperCells instead of the actual gates in the library. As the design progresses and we gain more information about the location and length of the wires, we continuously adjust the size of each SuperCell to keep the circuit delay as constant as possible. We increase the size of a SuperCell as the load on it increases and decrease it in size as the load decreases. As a result, we develop an overall circuit that is well balanced electrically, since each gate is sized optimally for the wire load that it is driving. This often results in layouts that are more compact and use less power than layouts derived using the conventional approach. Once we have determined the final placement for each gate, we replace each SuperCell in the layout with the closest matching size gate in the semiconductor vendor’s library. Using this approach, we are able to reduce the timing closure iterations that often occur in conventional integrated circuit design approaches.

4

Table of Contents

In addition to helping reduce timing closure iterations, we believe SuperCells enable faster and higher capacity synthesis. In conventional synthesis, the tool optimizes the circuit using library cells. Because a given logical function may be represented in the library by a collection of different gate sizes, the synthesis tool must try every permutation of gate size during optimization. If the circuit is large, the number of permutations becomes very large, which negatively affects run times and memory usage and puts a practical limit on capacity. Since the SuperCell concept has only one gate per logical function, the optimization search space can be much smaller. As a result, run times are significantly improved and the capacity of the system is much larger. Running on a standard engineering workstation, our system has a capacity of up to five million gates, an order of magnitude improvement over existing systems.

Unified Data Model Architecture

Our unified data model architecture is a key enabler for our FixedTiming methodology as well as our ability to deliver automated signal integrity detection and correction. We believe we are the only electronic design automation vendor that offers a complete integrated circuit design implementation flow based on a unified data model. The unified data model contains all the logical and physical information about the design and is resident in core memory during execution. The various functional elements of our software such as the implementation engines for synthesis, placement and routing, and our analysis software for timing, delay extraction and signal integrity, all operate directly on this data model. Because the data model is concurrently available to all the engines and analysis software, it makes it possible to analyze the design and make rapid tradeoff decisions during the physical design process. During optimization and placement, for example, our system continuously adjusts the sizes of SuperCells in the design as more accurate information about the layout is obtained from the data model. Additionally, our implementation software can instantly access our analysis software and continuously check for signal integrity problems during layout and take steps to avoid them. Existing approaches force the designer to perform signal integrity analysis after the layout is completed. Problems that are found then must be manually corrected, which may also affect timing closure and cause further iterations.

5

Table of Contents

Conventional electronic design automation flows are typically based on a collection of software programs that have their own associated data models. Data sharing and communications between software tools are accomplished either through file interfaces or through the establishment of a common database. If a common database is used, then each tool communicates with the database through a programming interface. For example, a software tool that is requested to send information to the common database must extract the data from its own data model, translate it into a form usable by the common database and then call on the programming interface to write the information to the database. Similarly, the software tool that requested the data must obtain the information from the common database through the programming interface, translate it into the format of its local data model and re-build the data model before the data can be used. The multiple data model approach has several limitations. It results in inefficient use of memory because the design data is replicated in various forms in memory. There are also capacity limitations due to the inefficient use of memory. In addition, there are performance limitations because the process of sharing data among software tools requires the use of a programming interface and the rebuilding of the data models each time that data is exchanged.

Our unified data model is designed to overcome these limitations. Memory is used more efficiently, capacity is higher, and performance is faster than in conventional systems because there is only one copy of the design data in memory. This eliminates the need for cumbersome data translations or reading and writing of data through a programming interface.

Products

Blast Fusion is our flagship product that provides significant advantages over traditional back-end design software. Our Blast Create product broadens the capabilities of Blast Fusion by adding front-end synthesis capability. In the front-end process, the chip design is conceptualized and written as a register transfer level computer program, or Register Transfer Level (“RTL”) file, that describes the required functionality of the chip. We also offer Blast Noise, our product that detects and corrects noise and other electrical problems in deep submicron chips, as a separate product to be used with Blast Create and Blast Fusion.

6

Table of Contents

Similar to the conventional design flow, our design flow starts by reading in technology libraries and constraint files. The following diagram illustrates our integrated design flow and where our products fit within this design flow.

Blast Create, first shipped in June 2001, is a key component of Magma’s RTL-to-GDSII IC design solution. It enables logic designers to synthesize, visualize, evaluate and improve the quality of their RTL code, design constraints, testability requirements and floorplan by building and analyzing a flat silicon virtual prototype that portrays the design in silicon. The physical netlist generated by Blast Create provides a clean handoff between RTL designer and layout engineer, eliminating back-to-front iterations necessary for timing closure in conventional flows.

Blast Fusion, first shipped in April 1999, is our physical design software that shortens the time it takes to design and produce deep submicron integrated circuits. Blast Fusion APX, first shipped in May 2002, is the high-end version of Blast Fusion targeted to address emerging nanometer design issues such as low power design and manufacturability of high performance, high complexity integrated circuits (ICs). The Blast Fusion flow starts by reading in the netlist, target library and design constraints. The netlist is optimized for circuit performance taking into account placement information that specifies the location of the gates in the chip layout. At the conclusion of this step, Blast Fusion generates a report that predicts the final timing performance that is achievable in the completed chip layout. In the final step, detailed physical design, Blast Fusion generates the final chip layout by performing the routing of wires that are needed to connect the gates into the desired circuit configuration and meet the timing performance requirements.

Blast Fusion is intended for use by chip design teams and other groups whose responsibility it is to take a design from netlist to completed chip layout. In the conventional Application Specific Integrated Circuit (“ASIC”) design flow, front-end designers use synthesis software to translate and optimize their RTL files into a

7

Table of Contents

netlist that is then handed off to the ASIC or semiconductor vendor or separate layout design group for physical design using Blast Fusion. Sales of Blast Fusion account for the majority of our revenue.

Blast Noise, first shipped in September 2000, is our noise detection and correction product. Interference, or noise from wires in close proximity to each other, can decrease chip performance or cause chip failure, particularly at 0.18 micron and below. Blast Noise works with Blast Fusion and Blast Chip to actively detect potential noise problems and correct them during the physical design process.

Blast Plan, first shipped in September 2001, delivers hierarchical design planning capabilities for use in implementing complex integrated circuit and system-on-chip designs. In a hierarchical design methodology, a chip design is partitioned into blocks that are designed and implemented individually and then later assembled to create the entire chip. Blast Plan works with Blast Fusion and Blast Chip to streamline the hierarchical planning and design of large chips and system-on-chips within a single environment.

Blast Prototype^™, Magma’s virtual prototyping system, which was first shipped in April 2002, provides design exploration and early problem detection. Blast Prototype uses the same analysis engines as Magma’s implementation system, thus providing a direct path to IC implementation using Blast Plan and Blast Fusion.

Blast Rail, first shipped in May 2003, provides IC designers with integrated power analysis and planning, voltage-drop analysis, voltage-drop-induced delay analysis, and electromigration analysis on rail wires and vias. These features enable designers to maintain power integrity in their designs. Blast Rail is fully integrated with Magma’s RTL-to-GDSII implementation flow to enable a correct-by-construction rail design solution.

Blast Power^™, launched in May 2004 and now starting to ship, is the industry’s first and only integrated power management and power minimization solution from RTL to GDSII. Blast Power is available as an option to Magma’s Blast Create and Blast Fusion APX IC implementation system, enabling Magma to offer a low-power design methodology that includes embedded power, timing, and rail analysis and power minimization techniques. With Blast Power, Magma users will be able to make power-vs-timing and power-vs-area tradeoffs throughout the RTL-to-GDSII flow—without having to export design data out of the Magma system. This tight integration of power optimization and management into the implementation process will enable users to deliver lower power and more cost- effective development cycles than point tool flows.

In June 2003 Magma acquired Aplus Design Technologies, Inc. (“Aplus”), a leader in physical synthesis and architecture analysis. Aplus products include PALACE^™, a physical synthesis tool for programmable devices (FPGAs), and ArchEvaluator^™, an architectural analysis tool. With the addition of these products to our product portfolio, we now offer implementation and physical design for cell-based, programmable and structured ASIC designs. Our customers are increasingly using structured ASIC designs, which enable a combination of cell-based and programmable logic, to reduce manufacturing costs.

PALACE, which first shipped in July 2001, is a fully automated physical synthesis tool for programmable logic devices. PALACE combines FPGA architecture-specific synthesis and mapping technologies with FPGA physical layout using a unified single data model throughout the synthesis process. PALACE offers an average of 15% better timing compared to best available FPGA synthesis solutions. PALACE supports all the popular FPGA architectures from Xilinx, Altera, Actel, and QuickLogic and it closely interfaces with FPGA vendor physical design tools.

ArchEvaluator, which first shipped June 2000, is the only commercial EDA tool that enables the programmable or Structured ASIC architecture designers to discover new synthesis-friendly architectures with the best performance and density advantages. ArchEvaluator is able to evaluate a wide scope of architecture parameters.

Blast FPGA^™, which recently started to ship, is a unified RTL to FPGA tool that combines RTL synthesis technology from Blast Create and physical synthesis technology from PALACE within a single data model.

8

Table of Contents

BLAST FPGA includes features such as an intuitive graphical user interface designed specific for FPGA designers, RTL and schematic views and cross probes, and embedded timing analysis. Blast FPGA offers an average of 20% better timing and 10% better area utilization compared to best available FPGA synthesis solutions. Blast FPGA also enables an easy FPGA migration to Structured ASIC or cell based ASIC within the same unified synthesis environment.

Blast Create SA, which recently started to ship, is a comprehensive front end design tool that enables synthesis, and partitioning of RTL description of the design into cell-based blocks and programmable blocks.

Similarly, Blast Fusion SA, which recently started to ship, is a complete physical design solution for programmable, cell-based or structured ASIC designs.

With the acquisition of Random Logic Corporation in October 2003, we acquired a capacitance extractor called QuickCap^®, long considered the industry’s leading parasitic extraction technology, and QuickInd^™, an inductance extractor based on the same core architecture as QuickCap. QuickCap is a highly accurate 3D-field solver used in parameter extraction and rules generation, library cell extraction, critical cell analysis, and critical net analysis.

Our acquisition of Silicon Metrics Corporation in October 2003, forming our Silicon Correlation Division, has allowed Magma to provide highly accurate models and characterization of various intellectual property (IP) blocks in nanometer designs. IP vendors, library developers, and COT design teams rely on software models to accurately represent the electrical behavior of circuits implemented with advanced process technologies. To meet the needs of these customers, Silicon Correlation Division’s SiliconSmart^™ products provide robust timing, power, and signal integrity models in a variety of industry standard formats. When used with popular construction and verification tools, these models offer silicon predictability and designer productivity. As a result, SiliconSmart models help customers shorten design cycles and improve chip performance.

We are in the process of integrating into our design flow certain verification and design for manufacturability (or “DFM”) technologies that we acquired by way of an April 2004 merger with Mojave, Inc. This development effort is expected to result in an ability to design ICs that are more manufacturable, and with inherently better yield, than those designed by flows that do not incorporate DFM capability. Magma believes that by incorporating DFM into IC implementation, Magma will be well positioned to address the next generation of designs at 65 nanometers and below.

Services

We provide consulting, training and chip design services to help our customers more rapidly adopt our technology. Design services include assisting our customers on complex chip design challenges and providing services ranging from the design and implementation of specific blocks to complete chip designs, including the delivery of the final chip layout, ready for release to manufacturing. We also provide post-contract support, or maintenance, for our products.

Customers

We license our software products to semiconductor manufacturers and electronic products companies around the world. Our customers include Broadcom, Infineon, NEC, Nokia, Texas Instruments, Renesas Technology, Toshiba and Vitesse.

In fiscal 2004, Texas Instruments and Broadcom each accounted for at least 10% of our total revenue and together accounted for 24% of our total revenue.

9

Table of Contents

Product Backlog

As of March 31, 2004, we had approximately $270 million in backlog, which we define as non-cancelable contractual commitments by our customers through purchase orders or contracts. Approximately 7% of the backlog is variable based on volume of usage of our products by the customers, approximately 4% includes specific future deliverables, and approximately 11% is recognized in revenue on a cash receipts basis. We have estimated variable usage, for the purposes of determining our backlog, based on information from customers’ forecasts available at the contract execution date. It is possible that customers from whom we expect to derive revenue from backlog will default and as a result we may not be able to recognize expected revenue from backlog.

Revenue and Orders Mix

Our license revenue in any given quarter depends on the volume of short term licenses shipped during the quarter and the amount of long term, ratable and cash receipts revenue from deferred revenue that is recognized out of backlog and recognized on orders received during the quarter. We set our revenue targets for any given period based in part, upon an assumption that we will achieve a certain level of orders and a certain license mix of short term licenses. The precise mix of orders is subject to substantial fluctuation in any given quarter or multiple quarter periods, and the actual mix of licenses sold affects the revenue we recognize in the period. If we achieve the target level of total orders but are unable to achieve our target license mix, we may not meet our revenue targets (if we deliver more-than-expected long term or ratable licenses) or may exceed them (if we deliver more-than-expected short term licenses).

Unbilled Accounts Receivable

Unbilled accounts receivable represent revenue that has been recognized in advance of contractual invoicing to the customer. We typically generate invoices 45 days in advance of contractual due dates, and we invoice the entire amount of the unbilled accounts receivable within one year from the contract inception. As of March 31, 2004 and March 31, 2003, unbilled accounts receivable were approximately $14.9 million and $6.8 million, respectively. These amounts were included in accounts receivable on our consolidated balance sheets for these periods.

Revenue by Geographic Areas

We generated 48% of our total revenue from sales outside the United States for fiscal 2004, compared to 39% in fiscal 2003. Additional disclosure regarding financial information on geographic areas is included in Note 11 of our Consolidated Financial Statements in Item 8 of this Annual Report.

Sales and Marketing

We license our products primarily through a direct sales force focused primarily on the industry leaders in the communications, computing, consumer electronics, networking and semiconductor industries. We have North American sales offices in California, Massachusetts, North Carolina, Pennsylvania, Texas, Washington and Canada. Internationally, we have European offices in Germany and the United Kingdom, an office in Israel and Asian offices in China, India, Japan, Korea and Taiwan. Our direct sales force is supported by a larger group of field application engineers that work closely with the customers’ technical chip design professionals.

As of March 31, 2004, we had 242 employees in our marketing, sales and technical sales support organizations. We intend to continue to expand our sales and field application engineering personnel on a worldwide basis.

10

Table of Contents

Competition

The electronic design automation industry is highly competitive and characterized by technological change, evolving standards, and price erosion. Major competitive factors in the market we address include technical innovation, product features and performance, level of integration, reliability, price, total system cost, reduction in design cycle time, customer support and reputation.

We currently compete with companies that hold dominant shares in the electronic design automation market. In particular, Cadence Design Systems, Inc. and Synopsys, Inc. are continuing to broaden their product lines to provide an integrated design flow. Each of these companies has a longer operating history and significantly greater financial, technical and marketing resources, as well as greater name recognition and larger installed customer bases than we do. These companies also have established relationships with our current and potential customers and can devote substantial resources aimed at preventing us from establishing or enhancing our customer relationships. Our competitors are better able to offer aggressive discounts on their products, a practice that they often employ. Our competitors offer a more comprehensive range of products than we do; for example, we do not offer logic simulation, full-feature custom layout editing, analog, or mixed signal products, which can sometimes be an impediment to our winning a particular customer order. In addition, our industry has traditionally viewed acquisitions as an effective strategy for growth in products and market share and our competitors’ greater cash resources and higher market capitalization may give them a relative advantage over us in buying companies with promising new chip design products or companies that may be too large for us to acquire without a strain on our resources. Further consolidation in the electronic design automation market could result in an increasingly competitive environment. Competitive pressures may prevent us from increasing market share or require us to reduce the price of products and services, which could harm our business. To execute our business strategy successfully, we must continue to increase our sales worldwide. If we fail to do so in a timely manner or at all, we may not be able to gain market share and our business and operating results could suffer.

Also, a variety of small companies continue to emerge, developing and introducing new products. Any of these companies could become a significant competitor in the future. We also compete with the internal chip design automation development groups of our existing and potential customers. Therefore, these customers may not require, or may be reluctant to purchase, products offered by independent vendors.

Our competitors may develop or acquire new products or technologies that have the potential to replace our existing or new product offerings. The introduction of these new or additional products by competitors may cause potential customers to defer purchases of our products. If we fail to compete successfully, we will not gain market share and our business will fail.

Research and Development

We devote a substantial portion of our resources to developing new products and enhancing our existing products, conducting product testing and quality assurance testing, improving our core technology and strengthening our technological expertise in the electronic design automation market. Our research and development expenditures for fiscal 2004, 2003 and 2002 were $26.1 million, $18.7 million and $18.2 million, respectively. There have not been any customer-sponsored research activities since the inception of the Company.

As of March 31, 2004, our research and development group consisted of 199 employees. We have engineering centers in California and Texas and in China, India, the Netherlands and Korea. Our engineers are focused in the areas of product development, advanced research, product engineering and design services. Our product development group develops our common core technology and is responsible for ensuring that each product fits into this common architecture. Our advanced research group works independently from our product development group to assess and develop new technologies to meet the evolving needs of integrated circuit design automation. Our product engineering group is primarily focused on product releases and customization. Our design services group is specifically focused on, and assists in completing, customer designs for commercial applications.

11

Table of Contents

Intellectual Property

Currently, we hold, directly or indirectly, nineteen issued patents in the U.S and two issued patents outside the U.S. Patent protection affords only limited protection for our technology. Our patents will expire on various dates between April 2018 and July 2021. We do not know if our patent applications or any future patent application will result in a patent being issued with the scope of the claims we seek, if at all, or whether any patents we may receive will be challenged or invalidated. Rights that may be granted under our patent applications that may issue in the future may not provide us competitive advantages. Further, patent protection in foreign jurisdictions where we may need this protection may be limited or unavailable.

It is difficult to monitor unauthorized use of technology, particularly in foreign countries where the laws may not protect our proprietary rights as fully as in the United States. In addition, our competitors may independently develop technology similar to ours. We will continue to assess appropriate occasions for seeking patent and other intellectual property protections for those aspects of our technology that we believe constitute innovations providing significant competitive advantages.

Our success depends in part upon our rights in proprietary software technology. We have patent applications pending for some of our proprietary software technology. We rely on a combination of copyright, trade secret, trademark and contractual protection to establish and protect our proprietary rights that are not protected by patents, and we enter into confidentiality agreements with those of our employees and consultants involved in product development. We routinely require our employees, customers and potential business partners to enter into confidentiality and nondisclosure agreements before we will disclose any sensitive aspects of our products, technology or business plans. We require employees to agree to surrender to us any proprietary information, inventions or other intellectual property they generate or come to possess while employed by us. Despite our efforts to protect our proprietary rights through confidentiality and license agreements, unauthorized parties may attempt to copy or otherwise obtain and use our products or technology. These precautions may not prevent misappropriation or infringement of our intellectual property.

Third parties may infringe or misappropriate our copyrights, trademarks and similar proprietary rights. Many of our contracts contain provisions indemnifying our customers from third-party intellectual property infringement claims. In addition, other parties may assert infringement claims against us. Although we have not received notice of any alleged infringement, our products may infringe issued patents that may relate to our products. In addition, because patent applications in the United States are not publicly disclosed until the patent is issued, applications may have been filed that relate to our software products. We may be subject to legal proceedings and claims from time to time in the ordinary course of our business, including claims of alleged infringement of the trademarks and other intellectual property rights of third parties. Intellectual property litigation is expensive and time consuming and could divert management’s attention away from running our business. This litigation could also require us to develop non-infringing technology or enter into royalty or license agreements. These royalty or license agreements, if required, may not be available on acceptable terms, if at all, in the event of a successful claim of infringement. Our failure to develop non-infringing technology or license the proprietary rights on a timely basis would harm our business.

Employees

As of March 31, 2004, we had 501 full-time employees, including 199 in research and development, 242 in sales and marketing and 60 in general and administrative. None of our employees are covered by collective bargaining agreements. We believe our relations with our employees are good.

Corporate Information

We were incorporated in Delaware in 1997. Our principal executive offices are located at 5460 Bayfront Plaza, Santa Clara, California 95054 and our telephone number is (408) 565-7500. Our common stock is traded on the Nasdaq National Market under the ticker symbol LAVA. Our Web site address is www.magma-da.com.

12

Table of Contents

The information in our Web site is not incorporated by reference into this annual report. Through a link on the Investor Relations section of our web site, we make available our annual report on Form 10-K, quarterly reports on Form 10-Q, current reports on Form 8-K, and any amendments to those reports filed or furnished pursuant to Section 13(a) or 15(d) of the Securities Exchange Act of 1934 as soon as reasonably practicable after they are filed with, or furnished to, the Securities and Exchange Commission. Our 2004 annual meeting will be held on August 31, 2004 at the law offices of Fenwick & West in Mountain View, California.

ITEM 2.    PROPERTIES.

Our corporate headquarters are located in Santa Clara, California, where we occupy approximately 130,000 square feet under a lease expiring on July 31, 2010. We have North American sales offices in California, Massachusetts, North Carolina, Pennsylvania, Texas, Washington and Canada. Internationally, we have European offices in Germany, the Netherlands and the United Kingdom; offices in Israel and Asian offices in China, India, Japan, Korea and Taiwan. We believe our current facilities are adequate to support our current and near-term operations. However, if we need additional space, adequate space may not be available on commercially reasonable terms or at all.

ITEM 3.    LEGAL PROCEEDINGS.

From time to time, the Company is involved in other disputes that arise in the ordinary course of business. The number and significance of these disputes is increasing as the Company’s business expands and the Company grows larger. Any claims against the Company, whether meritorious or not, could be time consuming, result in costly litigation, require significant amounts of management time and result in the diversion of significant operational resources. As a result, these disputes could harm the Company’s business, financial condition, results of operations or cash flows.

ITEM 4.    SUBMISSION OF MATTERS TO A VOTE OF SECURITY HOLDERS.

Not applicable.

EXECUTIVE OFFICERS OF THE REGISTRANT

Pursuant to General Instruction G(3) of Form 10-K, the information regarding our executive officers required by Item 401(b) of Regulation S-K is listed below.

The following table provides the names, offices, and ages of each of our executive officers as of May 31, 2004:

Rajeev Madhavan has served as our Chief Executive Officer and Chairman of the Board of Directors since our inception in April 1997. Mr. Madhavan served as our President from our inception until May 2001. Prior to co-founding Magma, from July 1994 until February 1997, Mr. Madhavan founded and served as the President and Chief Executive Officer of Ambit Design Systems, Inc., an electronic design automation software company, later acquired by Cadence Design Systems, Inc., an electronic design automation software company.

13

Table of Contents

Roy E. Jewell has served as our President since May 2001 and as one of our directors since July 2001. Mr. Jewell has served as our Chief Operating Officer since March 2001. From March 1999 to March 2001, Mr. Jewell served initially as the Chief Executive Officer and later as a consultant at a company he co-founded, Clarisay, Inc., a supplier of surface acoustic wave filters. From January 1998 to March 1999, Mr. Jewell was a member of the CEO Staff at Avant! Corporation, a provider of software products for integrated circuit designs. From July 1992 to January 1998, Mr. Jewell was the President and Chief Executive Officer of Technology Modeling Associates, Inc. or TMA, subsequently acquired by Avant! Corporation. Prior to that time, Mr. Jewell served in various marketing positions at TMA.

Gregory C. Walker has served as our Chief Financial Officer and Vice President—Finance since August 2002, and as our Senior Vice President—Finance since September 2002. From April 1999 to April 2002 he served as Chief Financial Officer, and most recently as interim Chief Executive Officer, for Accrue Software, Inc., a leading provider of customer relationship management products. From October 1997 to March 1999, Mr. Walker was Chief Financial Officer at Duet Technologies, Inc., a provider of semiconductor design services and software. From January 1997 through September 1997, Mr. Walker served as Chief Financial Officer of NeTpower, Inc., a manufacturer of work stations and servers. From December 1990 to January 1997, Mr. Walker served as Treasurer, Vice President of Finance and acting Chief Financial Officer, successively, at Synopsys, Inc., a supplier of electronic design automation solutions for the global electronic market. Prior to working at Synopsys, Mr. Walker held various positions in financial operations at Xerox Corporation and IBM Corporation.

Saeid Ghafouri has served as our Senior Vice President, Worldwide Field Operations since September 2002. From September 1999 to September 2002 Mr. Ghafouri was President and Chief Executive Officer of Empact Software, Inc., an enterprise software company. He served as President and Chief Executive Officer of an electronic design automation company, interHDL, which was acquired by Avant! Corporation, from April 1998 to September 1999. Prior to that Mr. Ghafouri served in various management positions between June 1996 and April 1998 at Synopsys, Inc., most recently as Vice President—Business Development for library products. He spent eight years with Cadence Design Systems Inc., between March 1986 and May 1994, where he served in various positions in Sales, Marketing and Applications Engineering.

Hamid Savoj co-founded our company and has served as our Senior Vice President, Product Development since September 2002. Before that he served as our Vice President, Product Development since July 2000. Between April 1997 and July 2000 he served as Magma’s principal engineer. From April 1994 to April 1997 Mr. Savoj was a senior member of the consulting staff at Cadence Design Systems.

Venktesh Shukla has served as our Senior Vice President, Marketing and Business Development since September 2002. Before that Mr. Shukla was Chief Executive Officer of Everypath, Inc., a leader in enterprise mobile computing, from April 1999 to January 2002. Prior to Everypath, he served from June 1996 to April 1999 as Vice President of Marketing at Ambit Design Systems where he was the key architect of Ambit’s successful entry into the logic synthesis market. Prior to Ambit, from January 1995 to January 1996, Mr. Shukla served as Vice President of Marketing at Systems & Networks, Inc., an enterprise network planning software provider. He was at Cadence Design Systems Inc. between June 1990 and December 1994 where he served most recently as Vice President of Marketing, Director of Product Marketing, and Strategic Marketing Manager.

14

Table of Contents

PART II

ITEM 5.    MARKET FOR REGISTRANT’S COMMON EQUITY, RELATED STOCKHOLDER MATTERS AND ISSUER PURCHASES OF EQUITY SECURITIES

Our common stock is traded on the Nasdaq National Market under the symbol “LAVA”. Public trading commenced on November 20, 2001. Prior to that, there was no public market for our common stock. The following table sets forth, for the periods indicated, the high and low per share sale prices of our common stock, as reported by the Nasdaq National Market on its consolidated transaction reporting system.

As of May 28, 2004, there were 345 holders of record (not including beneficial holders of stock held in street names) of our common stock.

Dividend Policy

We have not declared or paid cash dividends on our common stock and do not anticipate paying any cash dividends in the foreseeable future. We expect to retain future earnings, if any, to fund the development and growth of our business. Our Board of Directors will determine future dividends, if any.

Recent Sales of Unregistered Securities

During the year ended March 31, 2004, we issued a total of 623,372 shares of our common stock in connection with the acquisition of Aplus, pursuant to an agreement dated June 10, 2003 (the “Aplus Agreement”). We may issue up to 456,048 additional shares of our common stock upon the achievement of the earn-out milestones set forth in the Aplus Agreement. The securities were issued in reliance upon the exemption from the registration requirements of the Securities Act of 1933 provided by Section 4(2) and Regulation D thereof.

On April 29, 2004, during our first quarter of fiscal 2005, we issued a total of 607,554 shares of our common stock in connection with our acquisition of Mojave, Inc. pursuant to a definitive agreement signed on February 23, 2004. In addition to the initial merger consideration we may issue contingent consideration of up to $115 million, half in stock and half in cash, based on product orders over a period ending March 31, 2009, but such payments are contengent on the achievement of certain technology milestones. These securities were issued in reliance upon the exemption from the registration requirements of the Securities Act of 1933 provided by Section 3(a)(10) thereof.

Issuer Purchases of Equity Securities

We repurchased no shares of our common stock during the fourth quarter of fiscal 2004.

15

Table of Contents

ITEM 6.    SELECTED FINANCIAL DATA

The following selected consolidated financial data are qualified by reference to, and should be read in conjunction with, “Management’s Discussion and Analysis of Financial Condition and Results of Operations” and the Consolidated Financial Statements and related Notes included in Item 8 of this Report. The selected consolidated balance sheet data as of March 31, 2004 and 2003 and selected consolidated statements of operations data for the years ended March 31, 2004, 2003 and 2002, are derived from our audited consolidated financial statements included elsewhere in this Report. The selected consolidated balance sheet data as of March 31, 2002, 2001 and 2000 and the selected consolidated statements of operations data for the years ended March 31, 2001 and 2000 were derived from audited consolidated financial statements not included in this Report. Our historical results are not necessarily indicative of our future results.

16

Table of Contents

ITEM 7.    MANAGEMENT’S DISCUSSION AND ANALYSIS OF FINANCIAL CONDITION AND RESULTS OF OPERATIONS.

This Management’s Discussion and Analysis of Financial Condition and Results of Operation section should be read in conjunction with “Selected Consolidated Financial Data” and our condensed consolidated financial statements and results appearing elsewhere in this report. Throughout this section, we make forward-looking statements within the meaning of Section 27A of the Securities Act of 1933 and Section 21E of the Securities Exchange Act of 1934. You can often identify these and other forward looking statements by terms such as “becoming,” “may,” “will,” “should,” “predicts,” “potential,” “continue,” “anticipates,” “believes,” “estimates,” “seeks,” “expects,” “plans,” “intends,” or comparable terminology. These forward-looking statements include, but are not limited to, our expectations about revenue and various operating expenses. Although we believe that the expectations reflected in these forward-looking statements are reasonable, and we have based these expectations on our beliefs and assumptions, such expectations may prove to be incorrect. Our actual results of operations and financial performance could differ significantly from those expressed in or implied by our forward-looking statements. Factors that could cause or contribute to such differences include, but are not limited to: increasing competition in the electronic design automation market; the continuing impact of the economic recession; any delay of customer orders or failure of customers to renew licenses; weaker-than-anticipated sales of Magma’s products and services; weakness in the semiconductor or electronic systems industries; Magma’s ability to manage Magma’s expanding operations; the ability to attract and retain the key management and technical personnel needed to operate Magma successfully; the ability to continue to deliver competitive products to customers to help them get their products to market; and changes in accounting rules.

Executive Summary

Magma Design Automation provides electronic design automation, or EDA, software products and related services. Our software enables chip designers to reduce the time it takes to design and produce complex integrated circuits used in the communications, computing, consumer electronics, networking and semiconductor industries. Our products are used in all major phases of the chip development cycle, from initial design through physical implementation. Our focus is on software used to design the most technologically advanced integrated circuits, specifically those with a measurement of 0.13 micron and smaller. See “Item 1, Business” for a more complete description of our business.

As an EDA software provider, we generate substantially all of our revenues from the semiconductor and electronics industries. Our customers typically fund purchases of our software and services out of their research and development budgets. As a result, our revenues are heavily influenced by our customers’ long-term business outlook and willingness to invest in new chip designs.

Beginning in late calendar 2000, the semiconductor industry entered its steepest and longest downturn of the past 20 years, with industry sales dropping significantly from late 2000 to early 2002. As a result, over the past three years our customers have focused on controlling costs and reducing risk, reducing R&D expenditures, cutting back on design starts, purchasing from fewer suppliers, requiring more favorable pricing and payment

17

Table of Contents

terms from suppliers, and pursuing consolidation within their own industry. Further, during this downturn, many start-up semiconductor design companies failed or were acquired, and the pace of investment in new companies declined.

In response to these conditions, we have focused on providing the most technologically advanced products to address each step in the integrated circuit, or IC, design process, on integrating these products into broad platforms, and on expanding our product offerings. Our goal is to be the EDA technology supplier of choice for our customers as they pursue longer-term, broader and more flexible relationships with fewer suppliers.

While the semiconductor industry experienced a moderate recovery in 2003, our customers have remained cautious. It is therefore not yet clear when improved demand in our own customers’ electronics end markets will cause them to significantly increase their R&D spending or their design starts, and hence their spending on EDA.

Despite the condition of the semiconductor industry as described above, we were able to achieve the following during fiscal 2004:

Recent Business and Asset Acquisitions

We have recently acquired companies and purchased technologies to enable us to expand into markets for sign-off-quality tool sets for chip timing and parasitic extraction and in-library model generation and development. We believe that these acquisitions are a significant factor in Magma being able to compete successfully in the EDA industry and we expect to make similar acquisitions in the future. These acquisitions increased our headcount by more than 80 people and increased our research and development and sales and marketing expenses. Acquisitions may decrease our liquidity in the short term if earnout milestones are achieved and we must pay contingent cash consideration under the terms of some of these acquisitions.

18

Table of Contents

Business acquisitions

On July 1, 2003, we acquired Aplus, a developer of physical synthesis and physical prototyping solutions for programmable structured logic devices. We acquired all the outstanding shares of Aplus in exchange for initial consideration of $0.9 million cash and 0.3 million shares of our common stock. We also agreed to pay a total of $3.2 million of cash and 0.8 million shares of our common stock (collectively, the “Contingent Consideration”) to the Aplus shareholders pursuant to an earnout arrangement. The shares of common stock included in the Contingent Consideration were issued and placed in escrow and considered to be issued and outstanding as of the consummation date. Under the terms of the earnout provision, the Contingent Consideration was to be distributed to Aplus shareholders upon achieving or exceeding certain revenue, technology or financial targets. The earnout provisions were amended in April 2004 to revise the technical milestones and eliminate the financial targets, but the total Contingent Consideration remains the same. As of March 31, 2004, we had paid Contingent Consideration of approximately $1.4 million in cash and released 0.3 million shares of our common stock from the escrow, based on the achievement of the targets as of March 31, 2004. The Aplus acquisition was accounted for as a purchase business combination. The Contingent Consideration, when earned, is considered an additional acquisition cost and recorded as an increase to the developed technology intangible asset. That amount is being amortized to cost of sales over the remaining economic life of the developed technology intangible asset.

On October 17, 2003, we acquired Silicon Metrics, a developer of chip design characterization and modeling software for initial consideration of $18.0 million in cash. We also agreed to pay up to $14.0 million of cash in contingent consideration to the Silicon Metrics shareholders upon achieving or exceeding certain financial milestones. As of March 31, 2004, no contingent consideration had been earned. The contingent consideration, when earned, will be considered an additional acquisition cost and will be recorded as an increase to goodwill. Pursuant to the terms of the merger agreement, $1.8 million of the initial consideration continues to be retained by us in a segregated bank account as of March 31, 2004 to secure certain indemnification obligations of the Silicon Metrics shareholders and bonus plan participants. An additional amount of $0.8 million was retained by us to secure indemnification obligations with respect to certain litigation, but this amount was released to the Silicon Metrics stockholders in February of 2004 in connection with the settlement of the litigation. The Silicon Metrics acquisition was accounted for as a purchase business combination.

On October 20, 2003, we acquired Random Logic, a developer of the parasitic extraction software product QuickCap^® and QuickInd, for cash consideration of $20.0 million. Pursuant to the terms of the merger agreement, $5.0 million of that consideration was withheld and placed in an escrow account to secure the indemnification obligations of the Random Logic shareholders. The Random Logic acquisition was accounted for as a purchase business combination.

On February 23, 2004, we acquired SiliconCraft, a developer of advanced timing and power solutions for the high-end IC design industry. Prior to the acquisition, we had a 20% equity interest in Silicon Craft as a result of an earlier equity investment in July 2003. In this transaction, we acquired all remaining outstanding shares of SiliconCraft in exchange for the initial cash consideration of $1.2 million. In addition to the initial consideration, we may pay up to $1.5 million of cash in contingent consideration to the SiliconCraft shareholders upon achieving certain technology milestones. As of March 31, 2004, no contingent consideration had been earned. The contingent consideration, when earned, will be considered an additional acquisition cost and will be recorded as an increase to goodwill. The SiliconCraft acquisition was accounted for as a purchase business combination.

On April 16, 2004, we acquired Lemmatis, Inc. (“Lemmatis”), a developer of formal verification technology, for cash consideration of approximately $600,000, less $60,000 withheld to secure indemnification obligations. In addition, we may pay up to an additional $1.4 million upon the achievement of technology milestones set forth in the acquisition agreement. No contingent consideration yet has been paid under the agreement because the milestone dates have not occurred.

On April 29, 2004, we acquired Mojave, Inc. (“Mojave”), a developer of advanced technology for integrated circuit manufacturability and verification. The acquisition was effected by means of a two-step merger in which

19

Table of Contents

Mojave stockholders received initial consideration of $25.0 million, half in stock and half in cash. In addition to the initial merger consideration, we agreed to pay contingent consideration of up to $115 million, half in stock and half in cash, based on product orders over a period ending March 31, 2009, but such payments are contingent on the achievement of certain technology milestones. The contingent consideration, when earned, will be considered an additional acquisition cost. The Mojave acquisition will be accounted for as a purchase business combination in the first quarter of fiscal 2005.

Asset purchases

On March 26, 2004, we acquired a technology license and certain other information for a total fee of $22.8 million. The licensed technology will be integrated into our current product offerings as a formal verification equivalency checking tool that will be used to verify whether two different representations of a circuit are logically equivalent. Under the license agreement, we obtained a perpetual, fully-paid, royalty-free, non-exclusive, assignable, worldwide license. Further, we have a three-year period of exclusivity before the licensor can offer the licensed technology to our competitors.

During the year ended March 31, 2004, we completed three other asset purchases for an aggregate consideration of $17.7 million in upfront payments and related acquisition expenses of $0.5 million. Two of these purchase transactions included earnout provisions under which we would pay contingent consideration up to $2.8 million in cash based on the achievement of certain technology milestones as outlined in the respective asset purchase agreements. As of March 31, 2004, we had not paid any contingent consideration under these arrangements because no performance milestones had been met.

Critical Accounting Policies and Estimates

In preparing our financial statements, we make estimates, assumptions and judgments that can have a significant impact on our net revenue, operating income or loss and net income or loss, as well as on the value of certain assets and liabilities on our balance sheet. We believe that the estimates, assumptions and judgments involved in the accounting policies described below have the greatest potential impact on our financial statements, so we consider these to be our critical accounting policies. We consider the following accounting policies related to revenue recognition, allowance for doubtful accounts, investments, asset purchases and business combinations, deferred taxes and valuation of long-lived assets to be our most critical policies due to the estimation processes involved in each.

Revenue recognition

We recognize revenue in accordance with Statement of Position (“SOP”) 97-2, as modified by SOP 98-9, which generally requires revenue earned on software arrangements involving multiple elements (such as software products, upgrades, enhancements, maintenance, installation and training) to be allocated to each element based on the relative fair values of the elements. The fair value of an element must be based on evidence that is specific to us. If evidence of fair value does not exist for each element of a license arrangement and maintenance is the only undelivered element, then all revenue for the license arrangement is recognized over the term of the agreement. If evidence of fair value does exist for the elements that have not been delivered, but does not exist for one or more delivered elements, then revenue is recognized using the residual method, under which recognition of revenue for the undelivered elements is deferred and the residual license fee is recognized as revenue immediately.

Our revenue recognition policy is detailed in Note 1 of the Notes to Consolidated Financial Statements. Management has made significant judgments related to revenue recognition; specifically, in connection with each transaction involving our products (referred to as an “arrangement” in the accounting literature) we must evaluate whether our fee is “fixed or determinable” and we must assess whether “collectibility is probable”. These judgments are discussed below.

20

Table of Contents

The fee is fixed or determinable.    With respect to each arrangement, we must make a judgment as to whether the arrangement fee is fixed or determinable. If the fee is fixed or determinable, then revenue is recognized upon delivery of software (assuming other revenue recognition criteria are met). If the fee is not fixed or determinable, then the revenue recognized in each period (subject to application of other revenue recognition criteria) will be the lesser of the aggregate of amounts due and payable or the amount of the arrangement fee that would have been recognized if the fees had been fixed or determinable.

Except in cases where we grant extended payment terms to a specific customer, we have determined that our fees are fixed or determinable at the inception of our arrangements based on the following:

In order for an arrangement to be considered fixed or determinable, 100% of the arrangement fee must be due within one year or less from the order date. We have a history of collecting such arrangements fee according to contractual terms. Arrangements with payment terms extending beyond 12 months are considered not to be fixed or determinable.

Collectibility is probable.    In order to recognize revenue, we must make a judgment of the collectibility of the arrangement fee. Our judgment of the collectibility is applied on a customer-by-customer basis pursuant to our credit review policy. We typically sell to customers for which there is a history of successful collection. New customers are subjected to a credit review process, which evaluates the customers’ financial positions and ability to pay. If it is determined from the outset of an arrangement that collectibility is not probable based upon our credit review process, revenue is recognized on a cash receipts basis (as each payment is collected).

License revenue

We derive license revenue primarily from licenses of our design and implementation software and, to a much lesser extent, from licenses of our analysis and verification products. We license our products under time-based and perpetual licenses.

We recognize license revenue after the execution of a license agreement and the delivery of the product to the customer, provided that there are no uncertainties surrounding the product acceptance, fees are fixed or determinable, collection is probable and there are no remaining obligations other than maintenance. For licenses where we have vendor-specific objective evidence of fair value “VSOE”, for maintenance, we recognize license revenue using the residual method. For these licenses, license revenue is recognized in the period in which the license agreement is executed assuming all other revenue recognition criteria are met. For licenses where we have no VSOE for maintenance, we recognize license revenue ratably over the maintenance period, or if extended payment terms exist, based on the amounts due and payable.

For transactions in which we bundle maintenance for the entire license term into a time-based license agreement, no VSOE of fair value exists for each element of the arrangement. For these agreements, where the only undelivered element is maintenance, we recognize revenue ratably over the contract term. If an arrangement involves extended payment terms—that is, where payment for less than 100% of the license, services and initial post contract support is due within one year of the contract date—we recognize revenue to the extent of the lesser of the portion of the amount due and payable or the ratable portion of the entire fee.

For our perpetual licenses and some time-based license arrangements, we unbundle maintenance by including maintenance for up to first year of the license term, with maintenance renewable by the customer at the rates stated in their agreements with us. In these unbundled licenses, the aggregate renewal period is greater than

21

Table of Contents

or equal to the initial maintenance period. The stated rate for maintenance renewal in these contracts is VSOE of the fair value of maintenance in both our unbundled time-based and perpetual licenses. Where the only undelivered element is maintenance, we recognize license revenue using the residual method. If an arrangement involves extended payment terms, revenue recognized using the residual method is limited to amounts due and payable.

If we were to change any of these assumptions or judgments, it could cause a material increase or decrease in the amount of revenue that we report in a particular period. Amounts invoiced relating to arrangements where revenue cannot be recognized are reflected on our balance sheet as deferred revenue and recognized over time as the applicable revenue recognition criteria are satisfied.

Services revenue

We derive services revenue primarily from consulting and training for our software products and from maintenance fees for our products. Most of our license agreements include maintenance, generally for a one-year period, renewable annually. Services revenue from maintenance arrangements is recognized on a straight-line basis over the maintenance term. Because we have VSOE of fair value for consulting and training services, revenue is recognized as these services are performed or completed. Our consulting and training services are generally not essential to the functionality of the software. Our products are fully functional upon delivery of the product. Additional factors considered in determining whether the revenue should be accounted for separately include, but are not limited to: degree of risk, availability of services from other vendors, timing of payments and impact of milestones or acceptance criteria on our ability to recognize the software license fee.

Unbilled Accounts Receivable

Unbilled accounts receivable represent revenue that has been recognized in advance of being invoiced to the customer. We typically generate invoices 45 days in advance of contractual due dates, and we invoice the entire amount of the unbilled accounts receivable within one year from the contract inception. As of March 31, 2004 and March 31, 2003, unbilled accounts receivable were approximately $14.9 million and $6.8 million, respectively. These amounts were included in accounts receivable on our consolidated balance sheets for these periods.

Allowance for doubtful accounts

We maintain allowances for doubtful accounts for estimated losses resulting from the inability of our customers to make required payments. We regularly review the adequacy of our accounts receivable allowance after considering the size of the accounts receivable balance, each customer’s expected ability to pay and our collection history with each customer. We review significant invoices that are past due to determine if an allowance is appropriate using the factors described above. We also monitor our accounts receivable for concentration in any one customer, industry or geographic region.

To date, our receivables have not had any particular concentrations that, if not collected, would have a significant impact on our operating income. The allowance for doubtful accounts represents our best estimate, but changes in circumstances relating to accounts receivable may result in a requirement for additional allowances in the future. If actual losses are significantly greater than the reserve we have established, that would increase our general and administrative expenses and reduce our reported net income. Conversely, if actual credit losses are significantly less than our reserve, this would decrease our general and administrative expenses and our reported net income would increase.

Accounting for asset purchases and business combinations

We are required to allocate the purchase price of acquired assets and business combinations to the tangible and intangible assets acquired, liabilities assumed, as well as in-process research and development based on their estimated fair values. Such a valuation requires management to make significant estimates and assumptions, especially with respect to intangible assets.

22

Table of Contents

Critical estimates in valuing certain of the intangible assets include but are not limited to: future expected cash flows from license sales, maintenance agreements, consulting contracts, customer contracts, acquired workforce and acquired developed technologies and patents; expected costs to develop the in-process research and development into commercially viable products and estimating cash flows from the projects when completed; the acquired company’s brand awareness and market position, as well as assumptions about the period of time the acquired brand will continue to be used in the combined company’s product portfolio; and discount rates. Management’s estimates of fair value are based upon assumptions believed to be reasonable, but which are inherently uncertain and unpredictable. Assumptions may be incomplete or inaccurate, and unanticipated events and circumstances may occur.

Other estimates associated with the accounting for these acquisitions may change as additional information becomes available regarding the assets acquired and liabilities assumed resulting in changes in the purchase price allocation.

Goodwill impairment

Our long-lived assets include goodwill and other intangible assets. Statement of Financial Accounting Standards No. 142, “Goodwill and Other Intangible Assets,” requires that goodwill be tested for impairment at the reporting unit level (operating segment or one level below an operating segment) on an annual basis and between annual tests in certain circumstances. Application of the goodwill impairment test requires judgment, including the identification of reporting units, assigning assets and liabilities to reporting units, assigning goodwill to reporting unit, and determining the fair value of the reporting unit. We have determined that we have one reporting segment (see Note 11 of the Notes to the Consolidated Financial Statements under Item 8 of this report) Significant judgments required to estimate the fair value of a reporting unit include estimating future cash flows, determining appropriate discount rates and other assumptions. Changes in these estimates and assumptions could materially affect the determination of fair value for the reporting unit. Any impairment losses recorded in the future could have a material adverse impact on our financial condition and results of operations.

Valuation of long-lived intangible assets

Statement of Financial Accounting Standards No. 144, “Accounting for the Impairment or Disposal of Long-Lived Assets,” requires that we record an impairment charge on finite-lived intangibles or long-lived assets to be held and used when we determine that the carrying value of intangible assets and long-lived assets may not be recoverable. Based on the existence of one or more indicators of impairment, we measure any impairment of intangibles or long-lived assets based on a projected discounted cash flow method using a discount rate determined by our management to be commensurate with the risk inherent in our business model. Our estimates of cash flows require significant judgment based on our historical results and anticipated results and are subject to many factors.

Deferred tax asset valuation allowance

We account for income taxes in accordance with SFAS 109, “Accounting for Income Taxes.” We assess the likelihood that our net deferred tax assets will be recovered from future taxable income and to the extent we believe that recovery is not likely, we establish a valuation allowance. We consider future taxable income and ongoing prudent and feasible tax planning strategies in assessing the amount of the valuation allowance. However, adjustments could be required in the future if we determine that the amount to be realized is greater or less than the amount we have recorded.

Strategic investments in privately held companies

At March 31, 2004, the carrying value of our portfolio of strategic equity investments in non-marketable equity securities (privately held companies) totaled $1.8 million. Our ability to recover our investments in private, non-marketable equity securities and to earn a return on these investments is primarily dependent on how

23

Table of Contents

successfully these companies are able to execute to their business plans and how well their products are accepted, as well as their ability to obtain additional capital funding to continue operations and to grow. In the current equity market environment, their ability to obtain additional funding as well as to take advantage of liquidity events, such as initial public offerings, mergers and private sales, may be significantly constrained.

Under our accounting policy, the carrying value of a non-marketable investment is the amount paid for the investment unless it has been determined to be other than temporarily impaired, in which case we write the investment down to its impaired value. We review all of our investments periodically for impairment; however, for non-marketable equity securities, the impairment analysis requires significant judgment. This analysis includes assessment of each investee’s financial condition, the business outlook for its products and technology, its projected results and cash flows, the likelihood of obtaining subsequent rounds of financing and the impact of any relevant contractual equity preferences held by us or others. If an investee obtains additional funding at a valuation lower than our carrying amount, we presume that the investment is other than temporarily impaired, unless specific facts and circumstances indicate otherwise, such as when we hold contractual rights that give us a preference over the rights of other investors. As the equity markets have declined significantly over the past few years, we have experienced substantial impairments in our portfolio of non-marketable equity securities. If equity market conditions do not improve, as companies within our portfolio attempt to raise additional funds, the funds may not be available to them, or they may receive lower valuations, with more onerous investment terms than in previous financings, and the investments will likely become impaired. However, we are not able to determine at the present time which specific investments are likely to be impaired in the future, or the extent or timing of individual impairments. We recorded impairment charges related to these non-marketable equity investments of $1.2 million in fiscal 2004 and $0.6 million in fiscal 2003.

Results of Operations

Revenue

Revenue consists of license revenue and services revenue. License revenue consists of fees for time-based or perpetual licenses of our products. Services revenue consists of fees for services, such as post-contract customer support (“PCS”), customer training and consulting. As described in more detail under “Critical Accounting Policies and Estimates” above, revenue recognition depends upon the nature of the contract under which our products and services are sold.

Our licenses revenue in any given quarter is dependent upon the mix and volume of short-term licenses on orders received during the quarter and the amount of long term, ratable and cash receipts revenue from deferred revenue recognized out of backlog. We set our revenue targets for any given period based, in part, upon an assumption that we will achieve a certain level of orders and a certain license mix of short term licenses. The precise mix of orders fluctuates substantially from period to period and affects the revenue we recognize in the period. If we achieve our target level of total orders but are unable to achieve our target license mix, we may not meet our revenue targets (if we have more-than-expected long term or ratable licenses) or may exceed them (if we have more-than-expected short term licenses). If we achieve the target license mix but the overall level of orders is below the target level, then we will not meet our revenue targets as described in the risk factors below.

24

Table of Contents

Fiscal Years Ended March 31, 2004 and 2003

Revenue, cost of revenue and gross profit

Table below sets forth the fluctuations in revenue, cost of revenue and gross profit from fiscal 2003 to fiscal 2004 (in thousands, except percentage data):

Table sets forth the fluctuations in geographic distribution of revenue from fiscal 2003 to fiscal 2004 below (in thousands, except percentage data):

Revenue

25

Table of Contents

Gross profit

Cost of revenue includes personnel and related costs to provide product support, consulting services and training. Cost of revenue also includes software production costs, product packaging, documentation, amortization of acquired developed technology and other intangible assets, and amortization of deferred stock-based compensation. Management allocates these expenses to cost of upfront licenses, cost of time-based licenses and cost of services, based on orders booked within a given quarter. Accordingly, the costs allocated to upfront licenses, time-based licenses and services are heavily dependent on the mix of software orders received during any given period.

Operating expenses

Table below (in thousands, except percentage data) sets forth the fluctuations in operating expenses from fiscal 2003 to fiscal 2004:

26

Table of Contents

Other items

The table below (in thousands, except percentage data) sets forth the fluctuations in other items from fiscal 2003 to fiscal 2004:

27

Table of Contents

Fiscal Years Ended March 31, 2003 and 2002

Revenue

Revenue increased from $46.4 million in fiscal 2002 to $75.1 million in fiscal 2003 due to the increases in both license and service revenue. Of the $75.1 million of revenue recognized during fiscal 2003, approximately $63.6 million was license revenue. The increase in revenue was due to sales to new customers, which increased by more than 100% from the prior year, proliferation to additional design group designers in existing customers and the additions of new products.

License revenue

License revenue increased 66.5%, from $38.2 million in fiscal 2002 to $63.6 million in fiscal 2003. For fiscal 2003, the increase was primarily due to increased license revenue from our design and implementation software products, specifically from the introduction of Blast Fusion APX as well as from existing products, such as Blast Fusion, Blast Noise and to a lesser extent from Blast Chip.

Services revenue

Services revenue increased 40.2%, from $8.2 million in fiscal 2002 to $11.5 million in fiscal 2003. This increase was primarily due to consulting services provided to licensees of our design and implementation software products.

Cost of Revenue

Cost of revenue increased in absolute dollars from $8.4 million or 18% of revenue in fiscal 2002 to $11.6 million or 15% of revenue in fiscal 2003. The increase was primarily due to costs associated with additional support required by our expanding customer base combined with costs related to increased design services. The cost of sales relating to design services increased as a result of increased services revenue. In addition, we transferred the costs associated with all research and development employees who work in design services to cost of sales at the beginning of the second quarter of fiscal 2003 because such employees’ activities were revised and they now only perform work on customer related activities.

Operating Expenses

Research and development

During the second quarter of fiscal 2003, we transferred the costs associated with all research and development employees who work in design services to cost of sales because such employees would only perform work on customer related activities. Research and development expense consists primarily of salaries,

28

Table of Contents

bonuses and benefits of engineering personnel, depreciation of engineering equipment, and outside engineering services from contractors and consultants as we continue to invest in our technology. Research and development expense increased in absolute dollars from $18.2 million or 39% of revenue in fiscal 2002 to $18.7 million or 25% of revenue in fiscal 2003. The fiscal 2003 increase of $0.5 million was attributed to an increase in travel expense of $0.3 million, payroll related expenses of $3.5 million due to an increase in headcount, and outside services of consultants of $0.5 million. These increases were offset by $3.8 million of costs transferred to cost of sales. These costs were associated with a number of research and development personnel who previously performed a combination of research and development and services activities for our services organization. Their activities were changed limiting them to providing services only.

Sales and marketing

Sales and marketing expense consists primarily of salaries, sales commissions, bonuses, benefits and related costs of sales and marketing personnel, tradeshows and other marketing activities. Sales and marketing expense increased from $22.9 million or 49% of revenue in fiscal 2002 to $25.7 million or 34% of revenue in fiscal 2003. The increase was primarily due to increases in payroll related expenses of $2.9 million resulting from headcount increases to support the sales plan, advertising expense of $0.3 million, third party sales representative commissions of $1.0 million resulting from growth outside the United States, principally in Asia, travel expense of $0.6 million and office related expense of $0.5 million. In addition, an increase of $0.9 million related to the change in business focus of the field application engineers toward sales support, from customer design services charged to cost of sales in the prior year. These increases were partially offset by a reduction of $3.3 million in commission expense resulting from a change in the compensation plan, and $0.1 million in industry tradeshow related expenses.

General and administrative

General and administrative expense consists of salaries, bonuses, benefits and related costs of finance and administrative personnel and outside service expenses including legal, accounting and recruiting services. General and administrative expense increased 78% from $6.0 million or 13% of revenue in fiscal 2002 to $10.7 million or 14% of revenue in fiscal 2003. Of the net increase of $4.7 million, $1.9 million was related to the settlement with Prolific, Inc. Other increases include outside services of $1.2 million primarily from legal and accounting services, insurance expense of $0.4 million, bad debt expense of $0.6 million, travel expenses of $0.1 million, office related expenses of $0.2 million and payroll related expenses of $0.3 million.

Stock-based compensation

Stock-based compensation expense consists of the amortization of deferred stock-based compensation resulting from the grant of stock options at exercise prices less than the fair value of the underlying common stock on the grant date for officers and employees and the fair value of the stock options granted to consultants and other non-employees. The options granted to officers and employees generally vest over four years, with 25% vesting after one year and the balance vesting ratably over the remaining 36 months. Stock-based compensation declined 29%, to $4.8 million or 6% of revenue in fiscal 2003 from $6.7 million or 15% of revenue in fiscal 2002. For the year ended March 31, 2003, stock-based compensation included $4.3 million related to employees, of which $1.2 million was associated with our asset purchase of VeraTest, which occurred in fiscal 2003, and $0.5 million of expense related to consultant options.

Other Income (Expense), Net

Other income (expense), net was $1.3 million in fiscal 2003 and $(13.8) million in fiscal 2002. Interest income increased from $1.0 million in fiscal 2002 to $1.8 million in fiscal 2003 due to a change in the investment policy to include long-term investments beginning in the fourth quarter and more funds being available for investment on average for the year ended March 31, 2003. Interest expense decreased from $14.6 million in fiscal 2003. This decrease was primarily attributable to a non-cash charge of $14.6 million related to the $25

29

Table of Contents

million subordinated convertible promissory notes issued in July and August 2001. Such charges would not be incurred in future periods since the promissory notes were automatically converted into common stock on the completion of our initial public offering in November 2001. Other expense, net for the year ended March 31, 2003 represents an other-than-temporary decline in fair value of $0.6 million related to two of our strategic investments.

Income Taxes

The increase in income taxes of $0.9 million between the years ended March 31, 2003 and 2002 was primarily due to foreign tax expenses and U.S. federal alternative minimum tax.

Liquidity and Capital Resources

At March 31, 2004, our principal source of liquidity was our cash and cash equivalents and long-term investments. We had $72.7 million in cash and cash equivalents, and $78.2 million in long-term investments. Our investment portfolio consists primarily of fixed-income securities, with maturities of two years or less, diversified among industries and individual issuers. Our investments are generally liquid, investment grade securities.

On May 22, 2003, we issued $150.0 million principal amount of our Zero Coupon Convertible Subordinated Notes due May 15, 2008 (the “Notes”) resulting in net proceeds to us of approximately $145.1 million. The Notes do not bear coupon interest and are convertible into shares of our common stock at an initial conversion price of $22.86 per share, for an aggregate of approximately 6.56 million shares. The Notes are subordinated to our existing and future senior indebtedness and effectively subordinated to all indebtedness and other liabilities of our subsidiaries. We may not redeem the Notes prior to their maturity date. In order to minimize the dilutive effect from the issuance of the Notes, we undertook the following additional transactions concurrent with the issuance of the Notes:

The table below (in thousands) sets forth the key components of cash flow for fiscal 2004, 2003 and 2002:

30

Table of Contents

Fiscal 2004

Our operating activities provided net cash of $24.8 million. Cash was provided by net income adjusted for non-cash related items and changes in working capital including increases in accounts payable, accrued liabilities and deferred revenue. The increases in accounts payable and accrued liabilities are primarily due to the growth in our operations including headcount. Our headcount increased by 86% in fiscal 2004. Cash was also provided by an increase in other-long term liabilities, which primarily represented deferred rent related to the new lease agreement signed in fiscal 2004 for our headquarters facility. These increases were offset by cash used for changes in working capital including increases in accounts receivable, prepaid expenses and other current assets. Accounts receivable increased due to the timing of installment billings to customers. A detail of our accounts receivable balance as of March 31, 2004 is provided below (in millions, except for percentages):

Gross accounts receivable balance and their due dates

Gross billed accounts receivable balance and their aging