A company founded in 1993 with the goal of better graphics in computer games became the world’s most valuable enterprise in 2025. NVIDIA not only survived but redefined the technology of the 21st century. Its processors became the brain of the AI revolution, and the story of this transformation is one of vision, risk and relentless determination. From humble beginnings in a booth at Denny’s to the peak of global market capitalisation – this is a journey that changed the face of modern civilisation.
In 1993, when the digital world was still in its infancy and artificial intelligence was the domain of science fiction, three engineers – Jensen Huang, Curtis Priem and Chris Malachowsky – met at a Denny’s restaurant in San Jose, California. Huang was 30 years old at the time, married and the father of two children. It was there, over a breakfast table, that the idea of founding a company that would change the face of computer imaging was born. „We came to this Denny’s, sat down there and the three of us decided to start a company,” Huang recalled, adding: „Honestly, I had no idea how to do it, and neither did they. None of us knew how to do anything.”
Huang believed that the future of computer graphics required specialised processors that would take over rendering tasks from central processing units. Few people at the time suspected that this decision would lay the foundations for one of the greatest technological revolutions in human history. The founders had only $40,000 in capital, and Huang, before founding NVIDIA, had worked at LSI Logic Corp. and Advanced Micro Devices. He earned a degree in electrical engineering from Oregon State University and a master’s degree from Stanford University – an education that would prove to be the foundation of his later achievements.
How three engineers changed the world?
The company’s first product was the NV1 processor, developed in 1995. This chip, combining the functionality of 2D and 3D graphics, was intended to help Sega bring its games to PC. Unfortunately, the chip proved to be a failure – it was technically too weak to render graphics at any significant level. This mistake almost drove the company to bankruptcy. Huang, however, negotiated a contract buyout with Sega, which provided NVIDIA with a financial lifeline and the resources to develop the NV1’s successor. The new chip, named RIVA 128, debuted in 1997 and quickly achieved commercial success. This was the first moment NVIDIA could breathe a sigh of relief and think about further development.
From the very beginning of its operations, NVIDIA faced formidable competition, especially from 3dfx Interactive, whose legendary Voodoo card dominated the market for a time. Huang recalled that in the early years, the company experienced several near-death experiences. The key moment was the strategic decision to work closely with Microsoft and focus on the DirectX standard. When 3dfx made a series of strategic errors, NVIDIA was ready to take the lead. Ultimately, in 2002, the company made a spectacular move by acquiring its former rival. This not only eliminated the competition but also allowed NVIDIA to acquire valuable patents and technologies.
The birth of the GPU that changed everything
The breakthrough moment in NVIDIA’s history came with the 1999 release of the GeForce 256 processor. The company called it the world’s first GPU (Graphics Processing Unit). This chip was the first to move the entire process of transformation, shading and lighting of 3D objects in real time onto the graphics card, relieving the central processor. As Tom’s Hardware noted, the GeForce 256 can take the load off the CPU, prevent the 3D pipeline from stalling and allow game developers to use significantly more polygons, which automatically translates into a huge increase in detail.
The GeForce 256 was not an ordinary graphics card – it opened up new possibilities for game developers and laid the groundwork for future achievements in both gaming and computing. The chip contained 17 million transistors, an integrated geometry transformation engine and a dynamic lighting engine, as well as support for DirectX 7.0 and OpenGL. For gamers, firing up Quake III Arena on a GeForce 256 was a revelation. As AnandTech enthusiasts put it, the moment you launch your favourite game, you feel as if you have never seen that title before.
Over the next quarter-century, collaboration between game developers and NVIDIA continuously pushed the boundaries of possibility, driving the development of increasingly realistic textures, dynamic lighting and smoother animations. These innovations not only provided more immersive experiences for gamers but also transformed players into artists and entertainers, fuelling the global growth of esports. It was this relentless pursuit of better graphics that kept NVIDIA constantly improving its chips, preparing the ground for the revolution that was to come.
The risky bet that changed the company’s fate
The following years saw NVIDIA solidify its position in the graphics card market, but the real foundation for future success was laid in 2006, when the company introduced the CUDA (Compute Unified Device Architecture) programming platform. The decision to create and develop CUDA was one of the largest and most risky bets in NVIDIA’s history. The company staked everything on a technology that would not generate any profits for years. As Jensen Huang admitted, NVIDIA increased costs by as much as 50 percent, and its market value fell to around $1.5 billion. At that time, it was a move that balanced on the edge of survival.
CUDA gave programmers direct access to the computing power of the GPU, transforming it from a specialised graphics accelerator into a universal tool for massively parallel computations. The aim was to turn graphics cards into tools for general-purpose computing. Thanks to technologies such as programmable shaders and support for FP32 computations, GPUs began to be used in entirely new areas. GeForce cards played a key role, reaching millions of users – they were the ones that made CUDA spread and build an ecosystem. As Jensen Huang emphasises, it was GeForce that built the house on which NVIDIA now stands.
For many years, the technology had no key applications, and its maintenance generated costs. CUDA only began to deliver real benefits about 10 years after its introduction. This shows how long-term the company’s strategy was. It was only after 2012, when the scientific community discovered that the CUDA architecture was ideal for training neural networks, and deep learning began to revolutionise the field of artificial intelligence, that the risky decision made years earlier began to bear fruit.
How graphics processors became the brain of artificial intelligence
The breakthrough came when the world realised that training the largest artificial intelligence models required enormous computing power that traditional CPUs could not provide. It turned out that the GPU architecture, designed to process millions of pixels in real time, was ideal for the parallel data processing needed in machine learning. As Huang himself admitted, it was luck based on vision. They invented this capability, and one day deep learning researchers discovered this architecture. Because this architecture turned out to be perfect for AI.
By 2011, AI researchers had discovered that NVIDIA processors could handle the immense computational demands of deep learning. Scientists from Google, Stanford University and New York University began using them for their research. This was the moment when NVIDIA, a company mainly associated with gaming, began to transform into a leader in artificial intelligence.
Today, processors such as the H100 and the newer Blackwell generation have become key components of modern data centres and supercomputers worldwide. The Blackwell architecture, which debuted in 2024, represents a true technological leap. The B200 chip contains 208 billion transistors and uses advanced chiplet technology. It is equipped with 192 GB of HBM3e memory with 8 TB/s bandwidth, and its computing power in FP16/BF16 precision reaches 2250 TFLOPS. The GB200 version, the flagship product, integrates the Grace CPU with the B200 GPU, creating an integrated solution for the most demanding AI tasks. The new Blackwell Ultra B300 architecture offers impressive performance of 30 PFLOPS in FP4 precision.
The numbers that change the world
The scale of NVIDIA’s success is almost unimaginable. In July 2025, the company became the first in history to exceed a market capitalisation of $4 trillion. Just three months later, on 29 October 2025, NVIDIA became the first publicly traded company in the world to surpass the $5 trillion mark. Its shares reached a record high of $212.19. For comparison, this is a value greater than the annual GDP of all countries in the world except the United States and China. NVIDIA’s market value exceeds the combined market capitalisation of its biggest competitors: AMD, Intel, Broadcom, Micron, TSMC, ASML, Qualcomm and Arm Holdings.
The company needed only 41 trading days to increase its value from $3 to $4 trillion, and another 79 days to reach $5 trillion. In 2025, NVIDIA shares had already risen by more than 50 percent. Over the past five years, NVIDIA shares have risen by 1,500 percent. Today, the data centre segment is the company’s most important source of revenue. Revenues from this segment in fiscal year 2025 amounted to $51.2 billion, representing 66 percent year-on-year growth. According to forecasts, NVIDIA’s data centre revenues will grow at an average annual rate of 30.3 percent through fiscal year 2028.
In October 2025, Jensen Huang announced that NVIDIA expects to sell $500 billion worth of AI chips and is building seven supercomputers for the US, which will support scientific research, energy and security, among other areas. The company also announced a $1 billion investment in Nokia to develop AI-powered 5G-Advanced and 6G networks. Jensen Huang himself, thanks to this growth, found himself among the richest people in the world.
GPU as currency
NVIDIA has built a position as the undisputed leader, controlling about 94 percent of the GPU processor market used in generative AI software. However, this power brings new challenges. NVIDIA chips have become a subject of global political tensions. The US administration imposed export restrictions on advanced chips, and in response, Chinese authorities ordered companies such as ByteDance and Alibaba to stop purchasing and testing NVIDIA RTX Pro 6000D chips. Nevertheless, Alibaba announced cooperation with NVIDIA, and Huang himself admitted that China is only nanosecods behind the US in the AI race.
Competition from other technology giants is also growing. Amazon and Google are creating their own AI chips, such as Trainium 3 and TPU, trying to become independent from NVIDIA. However, as analysts note, CUDA makes switching to competing solutions extremely difficult – it would mean rewriting huge amounts of software, which involves enormous costs and delays. This software advantage is as important as the hardware advantage. As media reports, Google and Amazon reportedly call Jensen Huang to inform him of their chip plans before announcing them publicly – which shows the enormous influence NVIDIA still has over the entire industry.
The story of NVIDIA is the best proof that success does not come overnight. It is a story of a visionary who, against prevailing trends and the scepticism of his surroundings, consistently pursued his vision for over two decades. It is a lesson that true technological revolution is often born from a passion for games and ends up changing the face of the entire world. NVIDIA has proven that the greatest successes are born from the greatest risks, and the future belongs to those who can see potential where others see only emptiness.
Fot. Unsplash
