You might not have ever purchased an Nvidia product (perhaps you can't afford it), or even heard of Nvidia before, but in today's world, your life is inevitably intertwined with this company.

Not only does every gamer want a "4090" graphics card, but Nvidia has also infiltrated various aspects of your daily life with its GPUs. Whether it's the games you play with friends, the electric vehicles you travel in, the 3D blockbusters you watch in theaters, or every conversation you have with a chatbot, Nvidia is likely working silently behind the scenes. While not everyone understands them, no one can do without them.

Against this backdrop, Nvidia's market value once surpassed $2.3 trillion this year, second only to Microsoft and Apple. So, what exactly is Nvidia?

Silicon Valley Giant Born from a Fast Food Restaurant

Every Silicon Valley giant seems to have an unusual origin story—like a garage or a student dormitory. Nvidia is no different.

In 1992 to 1993, Nvidia's three founders, Jensen Huang, Chris Malachowsky, and Curtis Priem, conceived the idea of creating Nvidia while discussing it in a Denny's restaurant located next to a highway in San Jose, California.

Interestingly, the choice of Denny's as their meeting place was partly due to Jensen Huang's sentimental reasons. During high school, he had worked as a dishwasher at that very restaurant and did quite well.

The three of them chatted enthusiastically over coffee for several hours, to the point where an impatient manager moved them to a back table. However, after noticing bullet holes in the window, they decided it was best not to linger and left the restaurant.

The name "Nvidia" was coined when the trio returned to Priem's un-air-conditioned house and used a dictionary to come up with it. The name is derived from the Latin word "invidia," signifying a desire for their competitors to be "green with envy."

Initially, Nvidia was a company focused on 3D graphics processing for video games. They aimed to help games achieve finer and smoother graphics with more powerful chips. At a time when graphics technology on personal computers was still quite "primitive," the three saw immense potential for improvement.

However, not everyone could understand their entrepreneurial vision. When Jensen Huang's mother heard about the nature of his company's business, she even asked the soul-searching question: "Can't you just get a job?"

But undeniably, Nvidia has done "pretty well" on this path. In its early years, Nvidia developed a series of 3D accelerator cards, including the Riva 128, which helped Nvidia stand out among its competitors. In 1999, Nvidia developed the GeForce 256, the world's first truly meaningful GPU, a processor specifically designed for computing graphics and images, effectively becoming the "brain" of the graphics card.

In the niche field of gaming, Nvidia's continuous technological iterations have led to increasingly astonishing rendered graphics. Today, if your device is equipped with a newer Nvidia graphics card, its powerful computing performance and real-time ray tracing technology can simulate realistic lighting, allowing you to see lifelike graphics in AAA titles such as "Cyberpunk 2077" and "Red Dead Redemption."

Nvidia's revenue from gaming has risen in tandem with the performance and reputation of its graphics cards.

Until fiscal year 2022, gaming remained a core segment of Nvidia's four main businesses (gaming, data centers, professional visualization, and automotive). That year, Nvidia's gaming revenue reached $12.5 billion, accounting for 46% of its total revenue. In the discrete graphics card market driven by gamers, Nvidia has long held over 80% market share, firmly securing its leading position with a significant margin.

The goal of making competitors "green with envy" has been achieved.

The Big Winner in the AI Wave

Nvidia's performance in the gaming sector is undeniable, but if they had only dominated this "mountain," perhaps only "gamers" would rave about their graphics cards, let alone becoming the world's third most valuable company. What truly propelled Nvidia to new heights was the AI (artificial intelligence) wave that surged in 2023.

What do graphics cards have to do with AI? If you didn't think of the answer right away, don't worry—Nvidia's CEO Jensen Huang, who has led the company since its founding, didn't either.

He had long recognized that the potential of graphics cards wasn't limited to gaming. However, what he hadn't anticipated was that their graphics card technology would create a remarkable "chemical reaction" with deep learning, the hottest field in AI.

This all starts with how GPUs work. To simulate the infinite shapes in the real world, GPUs first generate simple basic triangles and then "approximate" complex shapes through the division and combination of these triangles, rendering them on the screen.

This task requires GPUs to perform simple, repetitive parallel computations on large amounts of data in a short period of time. This is where the distinction between GPUs and CPUs lies—the latter excels at sequential and complex calculations.

To put it in perspective, if a CPU is better at transporting a large load of goods to a destination in one go, a GPU is like a fleet of delivery drivers that can simultaneously transport many small items.

In 2006, Nvidia introduced the CUDA architecture. You can think of CUDA as a GPU programming toolkit that allows users to "command" the graphics card to perform tasks beyond graphical calculations using more familiar "high-level" programming languages, without having to write extensive low-level code. This "liberated" the supercomputing power of graphics cards.

But before deep learning, which utilizes large datasets for training, sparked the AI revolution, not everyone could make good use of this "dragon-slaying sword," and CUDA initially met with a cold reception.

In fact, deep learning was once considered an impractical and outdated technology. Fortunately, some "prophets" believed in both deep learning and Nvidia's graphics cards. Among them were Geoffrey Hinton, a professor at the University of Toronto later known as the "Godfather of AI," and his students Alex Krizhevsky and Ilya Sutskever.

In the 2012 ImageNet visual recognition challenge, they used two GeForce graphics cards to train their convolutional neural network, AlexNet. The result was a dominant victory with a correct rate that far surpassed the second place, demonstrating the astonishing potential of deep learning. Earlier that year, Google researchers achieved similar results using 16,000 CPUs.

Subsequently, deep learning attracted increasing attention and investment. Nvidia's graphics cards, perfectly suited for deep learning and powered by CUDA, became highly sought-after treasures in the AI field.

]

As mentioned earlier, Ilya Sutskever later became a co-founder and chief scientist at OpenAI. Thanks to his critical contributions, the globally renowned ChatGPT was officially introduced to the public at the end of 2022. The "T" in "GPT" stands for "Transformer." Generative AI, represented by ChatGPT, which can produce text, images, and even videos based on human instructions, has become the "Holy Grail" that every high-tech giant strives to obtain.

Possibly the largest "AI war" in history is unfolding, and Nvidia is almost the sole "arms supplier" in this battle.

In the first half of 2023, Nvidia's H100, touted as the "most powerful graphics card," saw its market price soar to nearly 300,000 RMB, yet it remained in short supply. The demand for these graphics cards was so high that Elon Musk remarked, "At this moment, getting a GPU is much harder than getting drugs."

Regardless of the outcome of this "AI war," Nvidia will be the ultimate winner. The AI arms race requires computing power, and Nvidia's graphics cards are synonymous with computing power. The more fiercely tech companies compete, the more Nvidia stands to gain.

In fiscal year 2024, Nvidia's revenue skyrocketed by 126% compared to the previous year, reaching $60.9 billion. Its operating profit surged by 681% year-on-year, hitting $32.972 billion. Revenue from AI-related data center business was $47.5 billion, accounting for 78% of the company's total revenue.

Nvidia does not primarily market AI applications, yet it has become a giant that influences the entire AI development landscape.

Once "30 Days from Bankruptcy"

Although under the leadership of "leather jacketed cowboy" Jensen Huang, Nvidia has become one of the world's most valuable companies, the road to its success was far from smooth. For Huang, Nvidia faced three heart-wrenching failures, with the company even coming close to bankruptcy.

The first major failure occurred in the early days of Nvidia.

At that time, Nvidia had won a contract to co-develop a gaming console with Sega. However, during development, Nvidia realized they had chosen the wrong technical direction, which was incompatible with the upcoming Windows 95 Direct3D. This put Nvidia in a dilemma: either complete the contract and produce a product that the market would reject or fail to complete it and face bankruptcy.

At this critical juncture, Huang chose to personally contact Sega's top executives, admit the mistake, and persuade them to find another partner. At the same time, he "embarrassingly" asked Sega to pay the full contract fee so that Nvidia could continue operating.

To Huang's surprise, Sega's CEO generously agreed. During this hard-won respite, Nvidia launched the "hit" RIVA 128 graphics card, finally overcoming the crisis.

The second "failure" was the development of CUDA. As a technology capable of delivering supercomputing power, CUDA's development cost Nvidia billions of dollars. However, the profits Nvidia could generate from CUDA were initially very low due to the small market size, leading to significant criticism from shareholders. By the end of 2008, Nvidia's stock price had even plummeted by 70%.

Despite the difficulties, Nvidia persevered. They tirelessly reached out to universities, businesses, and research institutions globally to build a customer base. With deep learning ushering in a new era for AI, Nvidia's steadfast vision finally became a reality.

The third failure was Nvidia's dismal outcome in the mobile computing chip sector.

Around 2010, the booming smartphone industry presented a vast market opportunity that Jensen Huang was eager to seize. Nvidia entered the fray with its Tegra mobile processors, competing against formidable rivals like Qualcomm and MediaTek. When Xiaomi launched its Mi 3, the first phone to feature the Tegra 4 processor, Huang even attended the launch event, enthusiastically declaring in Chinese, "I am also a Mi fan."

However, faced with fierce market competition and the challenges of developing new technologies, Nvidia eventually decided to abandon the mobile market and refocus on its core mission—providing solutions for problems that ordinary computers could not solve.

Nvidia's investment in Tegra was not in vain, though. The company later developed the DRIVE PX and DRIVE PX2 chips for autonomous driving based on Tegra technology, successfully entering the automotive industry.

Although Nvidia has faced these significant setbacks, the company's crisis management and innovative mindset in the face of challenges ultimately led to its current success. Jensen Huang's statement during the company's first major crisis, "Our company is 30 days from going out of business," has become an unofficial motto for Nvidia.

Will Nvidia Become the Next Microsoft or Apple?

At this point in time, Nvidia holds more than 70% of the market share for AI training chips and an overwhelming 92% share of the data center GPU market.

Considering that AI is increasingly becoming the "engine" of modern development, Nvidia, which has firmly hit the trend, may not yet be as widely recognized as Microsoft or Apple. However, given its current trajectory, it seems only a matter of time before it achieves that level of recognition.

But will everything go smoothly? The "cake" in the AI field is enormous, and Nvidia's graphics cards are indeed expensive. Not only are competitors like AMD and various startup chip companies trying to challenge Nvidia, but also major players such as Microsoft, Intel, Amazon, Google, Meta, and Apple are developing their own AI chips to break Nvidia's dominance.

However, Nvidia has dug a deep "moat" by building a software ecosystem around CUDA. Creating hardware with superior performance is not easy, and competitors will find it challenging to match Nvidia's long-established CUDA software services. Even if competitors can produce products that rival Nvidia's, customers are likely to stick with Nvidia due to consumer inertia and trust in the company.

Moreover, Nvidia has not been standing still. In addition to continuing to develop faster graphics cards, Nvidia acquired Israeli network technology company Mellanox in 2019. Mellanox's high-performance networking technology helps Nvidia better connect chips to create more powerful data centers, potentially building "AI factories" that continuously provide AI computing power to various industries, much like today's power plants.

For now, at least in the foreseeable future, Nvidia remains at the forefront, holding a position that is difficult to challenge. And that man who loves to wear a black leather jacket will continue to stand at the pinnacle of the world.