On April 5, 1993, which was also Jen-Hsun Huang's 30th birthday, he and two friends decided to start their own company in a chain restaurant in San Jose, California, to fulfill a promise to his wife that he would have his own company by the age of 30. Thus, Nvidia was born.
Since its establishment, Jen-Hsun Huang has been the CEO of Nvidia for nearly thirty-one years, making him one of the most dedicated figures in Silicon Valley.
In the legendary narrative of Silicon Valley startups, a tech company is typically led by a very young founder who dropped out of an Ivy League school and quickly grows into a global giant. Conversely, a company that was once a global giant but failed to keep up with the times may rapidly decline. After its decline, it might once again rise to greatness with a disruptive product. This cycle, characterized by rapid rises and falls, emphasizes speed.
However, Nvidia's story is somewhat different. Except for nearly going bankrupt at the beginning, Nvidia enjoyed smooth sailing and quiet progress for the first twenty years. In recent years, Nvidia has experienced an explosive surge, as if aided by divine forces. From artificial intelligence to cryptocurrencies, from the metaverse to ChatGPT, it doesn't seem like Nvidia is chasing trends; rather, it appears as if the trends are gravitating towards Nvidia, propelling it from the leading position in graphics cards to the throne of AI dominance.
Looking back on Nvidia's rise, it can be divided into five key phases.
Phase One: Difficult startup period, on the brink of bankruptcy (1993-1997)
Nvidia's early goal was to turn the personal computer into a consumer-grade device capable of gaming and multimedia. It released the NV1 and Riva series of graphics chips, but they were not well received in the market.
This led to the company's first difficult period. Its financial situation took a severe hit, and it faced the risk of bankruptcy. To survive, the company laid off staff, reducing from over 100 employees to just over 30.
Fortunately, during this time of crisis, Sega extended a helping hand by investing $7 million in Nvidia, supporting the development of a new generation of products. However, this cooperation ultimately failed, and the NV2 project under development was aborted. Consequently, Nvidia's founders made a significant decision: to abandon some existing patents and fully support Microsoft's DirectX interface. This move marked a turning point in Nvidia's history. In 1996, it launched a DirectX driver that supported Direct3D, making Nvidia one of the few graphics chip manufacturers on the market to support the DirectX interface.
In 1997, Nvidia introduced the key chip product Riva 128 (NV3), the world's first 128-bit 3D processor, which also supported Microsoft's DirectX interface. Hearing the news, many Microsoft downstream manufacturers started cooperating with Nvidia, and the Riva 128 chip became the first choice for many ODM manufacturers. In just the first four months, shipments of the Riva 128 chip exceeded one million units. These orders not only saved Nvidia from bankruptcy but also confirmed its leading position in technology, marking a significant victory in the graphics chip market.
Phase Two: Collaboration with TSMC, entering rapid development phase (1998-2005)
In 1998, Nvidia formed a strategic alliance with TSMC, which took over chip manufacturing and wafer processing for Nvidia, kicking off more than two decades of cooperation. Nvidia’s founder, Jen-Hsun Huang, has always considered Morris Chang, founder of TSMC, a close friend, claiming "without TSMC, there would be no Nvidia." The partnership with TSMC provided Nvidia with a stable chip supply, which was crucial for subsequent product innovations.
In 1999, Nvidia went public on NASDAQ, with an initial market value of around $230 million. That same year, Nvidia introduced the GeForce 256—the world's first fully functional GPU capable of taking over rendering tasks from the CPU, thus marking its birth.
In 2000, the company acquired 3dfx, the then-leader in graphics chip manufacturing, setting the stage for a duopoly in the graphics card market between Nvidia and the Canadian company ATI.
In 2001, Nvidia introduced the first programmable GPU, the GeForce3, allowing developers to create custom visual effects. Beyond that, Nvidia made scientific computing possible with GPUs. Consequently, GPUs started to be favored by AI scientists, laying the groundwork for the upcoming surge in GPU usage driven by the explosion of AI models.
In 2003, Nvidia announced the acquisition of MediaTek, a leader in wireless graphics and multimedia technology. In 2004, Nvidia introduced SLI technology, allowing multiple GPUs to be linked together, significantly enhancing the graphics processing capability of a single PC. That same year, the company also helped NASA reconstruct Martian terrain, showcasing its technology's great potential for scientific research applications. In 2005, Nvidia developed a processor for Sony's PlayStation 3 gaming console and announced its acquisition of core logic technology developer ULi Electronics.
From 1998 to 2005, Nvidia firmly established its unshakable position in the fields of high-performance computing and graphics processing through continuous technological innovation and product iterations.
Phase Three (1): The pivotal year of 2006
2006 is considered a landmark year for Nvidia, not only because of its technological innovations but also because it marked a turning point in its business strategy and market position.
To address the complexity of GPU programming, Nvidia introduced CUDA in 2006, a revolutionary general-purpose GPU computing architecture that significantly enhanced the capability of scientists and researchers to perform complex calculations using GPUs. This expanded the application domain of GPUs from gaming (graphics rendering) to later high-performance computing, autonomous driving, and more.
In the following AI era, CUDA gradually became the preferred GPU architecture for deep learning and AI training, securing Nvidia's leading position in the AI training and inference market.
Phase Three (2): Architecture Upgrade Period (2007 - 2014)
Between 2007 and 2014, Nvidia, having undergone a series of major changes and achievements, consolidated its leading position in the graphics processor market. It introduced a series of innovative products and technologies and continuously upgraded its graphics card architecture, altering the landscape of gaming graphics cards and high-performance computing.
The introduction of the Tesla GPU in 2007 marked a major breakthrough for Nvidia in the field of high-performance computing, bringing computational capabilities once only seen in supercomputers to fields such as drug discovery and medical imaging.
In 2010, the GPU based on the 40nm Fermi architecture was introduced, increasing the CUDA core count to 512.
In 2012, the Kepler architecture, based on the 28nm process, was introduced along with the NVIDIA GTX 680 graphics card based on this architecture. The emergence of the Kepler structure began the distinction between gaming cards and computing cards for Nvidia, creating the GK104 core focusing on graphics gaming performance and the GK110 core with stronger computing performance.
In 2013, Nvidia introduced the NVIDIA GTX Titan graphics card for gamers, thereby creating a new product line—the Titan series of graphics cards. This series featured a different public version design from previous products, significantly influencing the design of subsequent GTX XX80/80 Ti series graphics cards.
In 2014, Nvidia fully realized the Maxwell design architecture, simultaneously launching the GeForce GTX 980 and 970 GPUs, the most advanced gaming and graphics GPUs of the time, fully supporting CUDA 6.5 and all the latest features of the CUDA platform.
From 2007 to 2014, through continuous technological innovation and product iteration, Nvidia not only dominated the gaming graphics card market but also made significant progress in the high-performance computing field.
Phase Four: AI Data Center Business Launches Second Growth Curve (2015-2021)
In 2015, Nvidia launched the Pascal architecture for GPUs, CPUs, and interconnections between multiple GPUs. With the Pascal architecture, Nvidia's GPUs began to be used for deep learning in data centers and supercomputing. In July 2016, Nvidia released the Titan X Pascal graphics card, which not only supported the Tesla P100 compute GPU but also the GeForce GTX 1080 consumer-grade graphics card.
The Volta architecture, released in 2017, placed deep learning at its core, especially with the introduction of the NVIDIA Tesla® V100 GPU accelerator, which provided a much more powerful computational capability for AI operations, greatly enhancing the efficiency of AI computation.
In 2018, Nvidia once again led the industry with the Turing architecture, achieving real-time ray tracing rendering available in gaming. The Tesla T4 GPU computing accelerator, produced with the Turing structure, became the classic AI compute card supporting the GeForce RTX 2080 Ti.
The release of the Ampere architecture in 2020 marked another major breakthrough for Nvidia in GPU technology. The Ampere architecture GPUs demonstrated outstanding performance in gaming, data centers, AI applications, and more, becoming a trusted compute resource card for numerous AI companies worldwide.
From 2015 to 2021, Nvidia achieved remarkable success in GPU technology, deep learning, AI applications, and more, making it not only an outstanding hardware supplier but also a pioneer and promoter in the era of deep learning and artificial intelligence.
Phase Five: The Reign of the AI Superpower (2022-Present)
From 2022 to the present, Nvidia has not only released a series of industry-leading benchmark new products but also expanded its business globally, especially in the data center and artificial intelligence fields. During this phase, Nvidia has made breakthrough progress in AI, robotics, research, gaming, creative design, automotive, and other areas. Nvidia is driving progress in many of today's most influential fields, including artificial intelligence, digital biology, climate science, gaming, creative design, autonomous vehicles, and robotics.
In September 2022, Nvidia released the RTX 40 series graphics cards, especially the RTX 4090, which is 52% more powerful than its predecessor, the RTX 3090 Ti, becoming the dream card for many gamers.
However, the journey has not been without challenges. On October 17, 2022, the US government upgraded its export control on advanced artificial intelligence chips, causing Nvidia's stock price to drop by 5% that day. Facing this adverse situation, Nvidia still showed strong market resilience, with the company's stock price remaining strong in the subsequent quarters.
In 2022, in addition to GPU-related releases, Nvidia also launched its first CPU product, Grace, designed for high-performance computing and cloud computing. The Grace Hopper superchip, combining Grace and Hopper architectures, provides powerful CPU+GPU combinations for AI and high-performance computing.
On May 25, 2023, the Voyager project, a lifelong learning agent project driven by the Minecraft large model, was announced by Nvidia in collaboration with many universities in the United States. Through automatic tutorials, a growing skill library, and prompt iteration mechanisms, the project achieved strong situational lifelong learning capabilities and demonstrated excellent proficiency when playing Minecraft, capable of solving new tasks in new Minecraft worlds using the learned skill library.
Financially, the data center business has been a core driver of growth, with a five-year Compound Annual Growth Rate (CAGR) of 51%. In contrast, although the growth rate of the gaming business is lower, it gradually warmed up entering the fourth quarter of the 2023 fiscal year. According to data from the 2023 fiscal year, the data center business's revenue was about $15 billion, accounting for 56% of total revenue, while the gaming business accounted for 33%. In early June 2023, Nvidia's market value officially surpassed $1 trillion, becoming the world’s eighth company to achieve a trillion-dollar market value.
Now: In 2024, surpassing a $2 trillion market value and overtaking Apple is just a matter of time
By February 2024, just eight months after its market value exceeded $1 trillion, Nvidia's market value has already surpassed $2 trillion, becoming the world’s third largest company by market value after Microsoft and Apple.
In the upcoming wave of AI PCs, Nvidia is determined to be the leading pioneer. At the beginning of the year, it has prepared several major releases.
From the GeForce RTX™ SUPER desktop-end GPU with high-performance generative AI features to new NVIDIA RTX™ accelerated AI applications and tools for developers and consumers. In addition, Nvidia has also announced on its official website the launch dates for the GeForce RTX 4070 T SUPER and GeForce RTX 4080 SUPER chips.
In the future, computers that support AI will gradually become popular and change the existing landscape of the electronic gaming industry. Generative AI represents the most important platform shift in computing history, and it will transform all industries, including gaming. China's game development giants, such as miHoYo, Tencent, Ubisoft, and NetEase, have already adopted Nvidia's Nvidia Avatar Cloud Engine (ACE) microservice platform as their next-generation game engine.
In the new year of 2024, we look forward to Nvidia bringing surprises and legendary creations for developers and consumers!
