2701 San Tomas Expressway
NVIDIA Corporation is the worldwide leader in graphics processors and media and communications devices. The unmatched breadth of NVIDIA's product line enriches 3D, 2D, video, audio, communications, broadband connectivity, and high-definition digital video and television for every audience and platform including desktop PCs, game consoles, workstations, Internet-enabled appliances, Macintosh, and mobile PCs.
Founded in 1993, NVIDIA Corporation is a leading designer of graphics processors. The company's products first became accepted by the computer gaming market, then by other market segments such as the corporate and mobile PC markets. NVIDIA has grown into a dominant force in the graphics market primarily through the introduction of new products that broadened its line, and through the acquisition of some smaller competitors such as 3dfx Interactive Inc. along the way. The company has been selected by Microsoft Corporation as the sole supplier of graphics processors for its X-Box gaming console. Although NVIDIA has been recognized as having the fastest graphics chips available, it remains locked in a battle for a dominant market share of the graphics market with Ontario-based ATI Technologies Inc.
Favorable Reception for Graphics Processors: 1993-98
NVIDIA Corporation was established in 1993 by Jen-Hsun Huang, Chris Malachowsky, and Curtis Priem. Huang, the company's president and CEO, was the former director of coreware at LSI Logic Corp., where he was in charge of LSI's "system-on-a-chip" strategy. He also had experience designing microprocessors for Advanced Micro Devices and held a master's degree in electrical engineering from Stanford University. Malachowsky was NVIDIA's vice-president of engineering, and Priem became its chief technology officer.
NVIDIA's plan was to create microprocessors that would empower full-motion video and stereo sound in personal computers. It was a fabless company, which meant that it did not possess its own manufacturing plant to produce wafers and integrated circuits (ICs) for its graphics processors. For much of 1993 and 1994, leading-edge foundry capacity was scarce due to high demand. NVIDIA required access to half-micron capacity in order to make high-volume, low-cost processors; it could not settle for leftover capacity at older wafer fabrication facilities.
In the second half of 1994 NVIDIA reached an agreement with wafer fabricator SGS-Thomson Microelectronics, a European-based company with plants in France and Italy. It was NVIDIA's first strategic partnership and provided for the manufacture of NVIDIA's single-chip GUI (graphical-user interface) accelerator, or sound/graphics board, at an SGS-Thomson facility near Grenoble, France. In addition, NVIDIA arranged for Diamond Multimedia Systems to install the chips in multimedia accelerator boards.
The agreement with SGS-Thomson enabled NVIDIA to introduce its first multimedia accelerator, the NV1, in May 1995. The NV1 was the first microprocessor to integrate GUI acceleration, wave-table synthesis, full-motion-video acceleration, 3-D rendering, and a precision game port into a single chip. The NV1 made it possible for personal computers to compete with video game systems. As part of NVIDIA's arrangement with SGS-Thomson, NVIDIA marketed the NV1 video VRAM version for high-end PC applications while allowing SGS-Thomson to market its STG2000 version of the NV1 for DRAM to the high-volume consumer market.
The successful launch of NV1 helped NVIDIA achieve its first round of funding from venture capital firms Sequoia Capital and Sierra Ventures. The NV1 also attracted the attention of video game manufacturer Sega of America, Inc. Sega entered into a partnership with NVIDIA to port some of its established video games from its 32-bit Saturn system to personal computers using NVIDIA's NV1 multimedia chip.
By mid-1996 there were about 30 hardware vendors providing 3D chips to the graphics market, and over the next couple of years it was expected that 3-D would become a standard feature in personal computer systems. During the year NVIDIA gave its support to Direct 3D to establish itself with the game developer community. The company also achieved its second round of venture capital financing and focused on delivering leading-edge graphics for the desktop PC market.
In the first half of 1997 NVIDIA introduced the RIVA 128 (Real-time Interactive Video Animation) 3-D multimedia accelerator. The RIVA 128 was launched in part to answer criticism from original equipment manufacturers (OEMs) about the second-generation RV2. The RIVA 128 shipped to board makers Diamond Multimedia Systems and STB Systems, with SGS-Thomson Microelectronics handling the European, Japanese, and Asian-Pacific markets and NVIDIA the U.S. market.
By early 1998 NVIDIA was recognized as one of the dominant companies in the high-performance 3-D graphics market. One analyst estimated that NVIDIA held a 24 percent market share at the end of 1997, ranking it second among high-performance graphics companies. In March 1998 NVIDIA introduced two new products, the Riva 128ZX and the RIVA TNT. The Riva 128ZX was an upgrade of the Riva 128 that enabled better visual effects and was targeted to the corporate market. The RIVA TNT (TwiN Texture) was the industry's first multi-texturing 3-D processor. It could achieve sustained rates of 125 million to 150 million pixels per second and 250 million pixels during peak bursts. According to some analysts, the RIVA TNT represented a sea change in graphics processing, allowing chip designers to calculate pixels and apply color, shading, lighting, filtering, and other effects.
NVIDIA filed for an initial public offering (IPO) in early 1998, but the company delayed its IPO until January 1999. During 1998 it was hit with several patent infringement lawsuits from its competitors, including Silicon Graphics Inc., S3 Inc., and 3dfx Interactive Inc. All of the lawsuits focused on NVIDIA's use of multi-texturing technology. Meanwhile, NVIDIA's RIVA TNT won PC Magazine's Editor's Choice Award for "the most impressive implementation of combined 2-D and 3-D performance we've seen," and NVIDIA was named the "Most Respected Private Fabless Semiconductor Company" for the second consecutive year by the Fabless Semiconductor Association (FSA). At the fall '98 COMDEX trade show, NVIDIA introduced the NVIDIA Vanta, a graphics accelerator designed to bring desktop computer graphics to the corporate market.
Sustained Revenue Growth: 1999-2002
NVIDIA held its IPO in January 1999. For 1998 the company reported tremendous growth in revenue, reflecting both growth in the graphics industry and acceptance of NVIDIA's line of RIVA graphics processors. For the fiscal year ending January 31, 1999, NVIDIA reported revenue of $158.2 million, compared to revenue of only $13.3 million the previous year, and net income of $4.1 million compared to $1.3 million the previous year.
During 1999 NVIDIA continued to expand its product line and enter new market segments. In March it launched a 64-bit version of the Vanta for the low-end PC market. In mid-1999 NVIDIA introduced TNT2, which Computer Gaming World rated as 30 percent faster than TNT when tested on a Diamond Viper 770 display board. NVIDIA also rolled out a new set of software drivers for the TNT and TNT2 that were optimized to work with Advanced Micro Devices' processors.
In July 1999 NVIDIA and Silicon Graphics Inc., which had changed its name to SGI, settled their lawsuits and entered into a strategic partnership. The two companies agreed to cross-license their patent portfolios. In August SGI subsequently transferred nearly 50 graphics engineers to NVIDIA as part of a corporate reorganization. Around this time NVIDIA also formed a strategic alliance with Acer Labs Inc. (ACI) to introduce integrated 3-D and core logic graphics products for low-end PCs.
NVIDIA launched its next-generation graphics accelerator in August 1999. Dubbed the GeForce256, the 256-bit graphics processing unit (GPU) was the industry's first GPU; it was able to offload the entire graphics processing operation from a computer's central processing unit (CPU). According to Electronic Buyers' News, the GeForce256 "cements the company's reputation as a broad-line graphics-chip supplier by pushing 3-D performance into rarefied territory." The GeForce256 not only unburdened the CPU of graphics processing, it also enabled software developers to create more complex attributes such as artificial intelligence and model more complex objects.
Other new products introduced in the final quarter of 1999 included the Aladdin TNT2 for PCs costing less than $1,000. The chipset was the result of NVIDIA's alliance with Taiwan-based Acer Labs and incorporated NVIDIA's TNT2 graphics core with Acer Labs' Northbridge design. In November NVIDIA launched Quadro, the first GPU designed for workstations.
For 1999 NVIDIA reported revenue of $374.5 million, more than double that of the previous year, and net income of $41 million. Following a highly successful year, NVIDIA began building a new headquarters campus in Santa Clara, California. In January the company sued competitor S3 over five patents, but the two companies settled the next month by entering into a seven-year patent cross-licensing agreement. In April NVIDIA announced a secondary stock and bond offering designed to raise about $400 million in capital. The company also received an advance payment of $200 million from Microsoft Corporation, which had selected NVIDIA as the sole supplier of graphics processing units for its forthcoming X-Box video game console.
For the next several years NVIDIA and Ontario-based ATI Technologies Inc. battled for the top spot among graphics chip designers. At the Windows Hardware Engineering Conference in April 2000, ATI introduced a graphics chip it claimed was the best in the industry, only to be bested 24 hours later when NVIDIA introduced the GeForce 2 GTS, which delivered twice the performance of the GeForce256. ATI's Radeon256 could process up to 1.5 billion texels per second, while the GeForce256 could process 1.6 billion texels per second. Although NVIDIA could boast having the fastest graphics chip, it trailed ATI in market share with an estimated 25 to 30 percent of the graphics chipset market.
NVIDIA followed up in June 2000 by introducing the GeForce2 MX for the mainstream desktop graphics market. The GeForce2 MX consumed only four watts of power and was suitable for mobile PCs, PC workstation applications, and Apple's Macintosh computers. The GeForce2 MX featured twin-display architecture that enabled it to serve multiple monitors. Its digital control technology made 2-D and 3-D graphics, images, and video brighter and cleaner. In November 2000 NVIDIA introduced the GeForce2 Go, the industry's first mobile GPU for portable PCs. CEO Jen-Hsun Huang called the GeForce2 Go "one of the most important [product introductions] in the history of our company," when it was introduced at Comdex in the fall of 2000.
Later in 2000 NVIDIA announced a new communications companion chip for the X-Box as part of its partnership with Microsoft. The media communications processor (MCP) was a kind of super-chip that combined traditional I/O (input/output), communications, and audio functions. The MCP went beyond its contract with Microsoft to produce the graphics processor for its X-Box. While NVIDIA was contracted to produce the chips for the X-Box, the MCP was designed to be used in future PC architecture and information-appliance components. Utilizing NVIDIA's signal processing technology, the MCP could process audio signals and link an X-Box to other peripherals through a phone line or a wireless network.
NVIDIA continued to aggressively contest its patents in 2000. After settling with S3 early in the year, NVIDIA filed suit against 3dfx Interactive Inc. in August claiming infringement of its graphics acceleration technology patents. At the time 3dfx also had a lawsuit outstanding against NVIDIA and in October won a favorable ruling. In December, however, it was announced that NVIDIA would acquire the assets of 3dfx Interactive for $70 million in cash and one million shares of common stock.
For its fiscal year ending January 28, 2001, NVIDIA reported another healthy increase in revenue to $735.3 million, nearly double the previous year's revenue, and net income of $98.5 million. The company also earned several awards at Comdex Fall 2000 and was awarded the Editor's Choice Award from Cadence magazine.
The year 2001 began with a healthy market for graphics processors, although the industry had experienced substantial consolidation. According to EBN, the number of firms producing graphics controllers had shrunk from 45 to 12 over the past five years. Several key players, including S3 Inc., NeoMagic Corporation, and Intel Corporation, either had exited the graphics market or cut back on their new product development. During 2001 NVIDIA continued to post record revenue, and it ended the year with revenue of $1.37 billion, again nearly doubling revenue of the previous year. Net income also rose substantially to $176.9 million. The company's stock price rose from $16.50 at the beginning of the year to $66.90 at the end of 2001. In December 2001 NVIDIA replaced Enron Corporation in the S&P 500, and the company was named "The Most Respected Public Fabless Semiconductor Company" by the Fabless Semiconductor Association.
During 2001 NVIDIA completed development of its graphics chip for the Microsoft X-Box, which it called GeForce3. GeForce3 represented the core graphics that would be found in X-Box, which was scheduled for introduction later in the year. The GeForce3 was the graphics industry's first programmable GPU and was chosen by all of the top PC and graphics board OEMs. The chip's programmable features enabled developers to create custom lighting and transform custom pixel effects. NVIDIA also expanded its GeForce2 family of GPUs in the first quarter of 2001. In October NVIDIA launched improved versions of its GeForce3 product line under the name Titanium.
Other new products included the nForce line of chipsets. The nForce chipset represented a move into the core logic market for NVIDIA, where the company would compete with Intel Corporation and other chipset vendors. It included an integrated graphics controller and a separate audio processor and was designed to offer performance up to ten times better than any other PC graphics accelerator. It was compatible with Advanced Micro Devices Inc.'s Athlon processors and Duron-based systems and also was being marketed to PC OEMs to feature in their high-volume systems.
After NVIDIA's stock price doubled in October 2001, some of the company's employees were indicted by a federal grand jury for insider trading. One engineer was fined $250,000. In early 2002 NVIDIA announced that it was conducting an internal review of certain accounting procedures in response to inquiries from the Securities and Exchange Commission.
The year 2002 was more challenging for NVIDIA. Although the company reported record first quarter revenue of $582.9 million, its revenue and net income fell in the second and third quarters, with the company reporting a third quarter loss of $48.6 million on revenue of $430.3 million. The drop in revenue came from a general weakness in the PC market as well as from the write-off of NVIDIA's inventory of first-generation nForce chips. Adding to the company's woes was a contract dispute with Microsoft over the price of its graphics processor designed for the X-Box. Microsoft was pressing NVIDIA to offer it a better price on the processors than was in the original contract.
New products introduced in 2002 included the GeForce4 graphics accelerator, which offered significantly enhanced overall 3-D performance; the Quadro4 family of GPUs for workstations; and NVDVD, a software-based DVD player/decoder designed to power DVD playback for a new generation of NVIDIA-based mobile and desktop PCs. The GeForce4 was based on 0.15 micron process and included 63 million transistors, a 10.4 gigabyte memory bandwidth, a 128-megabyte frame buffer, and a 300MHz core. The Quadro product line was a professional graphics solution for real-time character animation, next-generation game development, and visual effects production.
As NVIDIA faced a challenging business environment at the end of 2002, it was looking forward to the introduction in 2003 of its next-generation NV30 chip, which would run on 500MHz and utilize eight pixel pipelines. The NV30's core memory would be the first to use DDR-II memory, allowing it to run at an effective rate of one GHz. NVIDIA also was pushing to maintain its technological lead in the graphics market with the creation of Cg, a high-level 3-D language. NVIDIA hoped that Cg would be adopted as the language of choice over a competing OpenGL 2.0 language. As part of that effort NVIDIA acquired Exluna Inc., a film-quality graphics rendering company, in July 2002 to facilitate the use of Cg in the film industry.
Principal Competitors: 3Dlabs Inc., Ltd.; ATI Technologies, Inc. (Canada); Creative Technology Ltd. (Singapore); Intel Corporation; VIA Technologies, Inc.