Many people are shocked to hear the story about NVIDIA’s stock surging more than 1200% over the past five years, and how other new AI companies are surging to the top of big business in short amounts of time, but not many know what truly makes these businesses so successful. It is common to attribute AI companies’ success solely to the growing usage of AI hardware, but that explanation does not fully capture the sophistication and strategy of the business models that these companies utilize.
On April 5, 1993, Jensen Huang and two other experienced microchip designers founded NVIDIA. Originally, this company focused on manufacturing chips to accelerate 3D graphics for video games through their first product, the NV1. This product integrated 2D and 3D graphics and audio. Unfortunately, Microsoft developed software technology standards that were not compatible with the NV1, which led to economic disaster for NVIDIA. Soon after, NVIDIA lost major funding from Sega, a prominent videogame and entertainment company, and was forced to lay off employees, investing its dwindling funds in many projects in hopes that one would strike it big. After funneling their remaining resources in the development of a new consumer graphics card which made 3D video games possible, the RIVA 128, they finally got the lifeline they were hoping for. This chip was built to render within Direct3D and OpenGL API specifications, meaning they aligned with Microsoft’s industry standards, which gave consumers a product they could actually use.
With this excess capital from RIVA 128, they developed GeForce 256 in 1999. This was branded as the world’s first GPU, graphics processing unit, which delivered an incredible leap in 3D graphics performance and became the dominant product for PC gaming graphics. GPUs can also process thousands of small tasks simultaneously. To pile onto these business-saving developments, in 2006, they would deliver their most pivotal development that had nothing to do with gaming. NVIDIA introduced CUDA, a software platform that allowed GPUs to perform computing far beyond game graphics. CUDA allowed engineers and scientists to harness the advantages of GPUs to perform computational problems. In effect, NVIDIA had developed the perfect technological infrastructure that would later serve as the building blocks for future AI infrastructure.
NVIDIA’s dominance is the product of a business model that makes it extremely difficult for them to be removed from the market. They focus on maintaining two connected product lines, hardware and software, which work together to create a market dependence on their products.
For hardware, NVIDIA sells data center GPUs, most notably the H100 and Blackwell B200, which are essential infrastructure in creating successful AI data centers. As major companies such as Microsoft, Google, and Meta spend hundreds of billions of dollars to build AI infrastructure, they have to purchase NVIDIA GPUs in order to produce the most successful centers possible. As a result of the major demands for these GPUs, they can price these GPU units at around 25 to 40 thousand a piece, despite it costing NVIDIA a mere $3500 to produce, demonstrating their incredible profit margin.
Even with their incredible profit from hardware, they also have an incredible market influence through their software. In 2006, NVIDIA released CUDA, which would go on to be the software that millions of programmers harnessed to perform their own research work for the next two decades. As a result, the most widely used AI frameworks run on CUDA, creating a dependence on NVIDIA software to create competitive AI. Through these processes, NVIDIA has solidified itself as a near-insurmountable force in the AI industry by forcing the market to depend on its products.
The risks and rewards that have come to shape NVIDIA’s presence in the market is incredibly relevant information for investors as well as people aspiring to be business leaders. For aspiring business leaders, NVIDIA demonstrates that even when it seems a company has its back to the wall, it only takes one breakthrough to get back on top. Likewise, for investors, this can serve as a cautionary tale for making large bets against tech companies in bad positions, as they could easily surge in value with the release of a single product. At the same time, this also shows the relative un-sustainability of investing in tech companies because in the AI market, products are constantly being upgraded and revamped to further compete with market competition. As a result of constant innovation, companies that are unable to create winning products will eventually be left behind. That being said, there is no way to gauge whether a company will produce a winning product which perfectly highlights the danger with investing in this sector.
More specifically, NVIDIA seems to be an unstoppable business model through its dominance in the AI industry, there are still major uncertainties and risks that coincide with investing in this company. First, there is a major geopolitical threat to this company. Most notably, the United States government has repeatedly limited NVIDIA’s chip selling to China, which cuts off a significant portion of its revenue. Also, hyper-scalers such as Microsoft and Google make up the vast majority of data center revenue for NVIDIA. These corporations will likely want to reduce their dependence on NVIDIA in the future while developing their own products, such as Google’s TPU, which will likely offer 30% better price than NVIDIA’s GPUs. The bottom line is that NVIDIA’s GPUs will likely remain dominant in the near future, but as technology continues to advance and large companies continue to pour money into software and hardware development, NVIDIA’s dominance will likely diminish over time unless it is able to keep up with AI innovation.
