There is no hotter stock in the world than Nvidia (NASDAQ:NVDA) right away. The company’s H100, A100 and new Blackwell semiconductor chips are widely considered superior to the competition. Given Nvidia’s impressive roster of graphics processing units (GPUs), the company is undoubtedly benefiting from the moment when generative AI applications take off.
While this is encouraging, investors might be surprised to learn that Nvidia offers many other services beyond GPU chips and data center services. Let’s dive into one of Nvidia’s secret and little-known businesses and assess why this specific operation may be the most important of all.
Beyond chips and data centers
Nvidia reports its revenue in two categories: compute and networking, and graphics. When Nvidia was founded in the early 1990s, the company’s initial mission was to improve the graphics processing capabilities of video games.
Although gaming remains a core part of Nvidia, the company’s biggest businesses now fall into computing and networking. Granted, computing and networking – and gaming – are pretty general terminologies. What do they really mean?
According to Nvidia’s filings, the Compute and Networking segment primarily includes the company’s data center, robotics, and artificial intelligence (AI) businesses, while graphics operations represent Nvidia’s GPU business. .
During the company’s first quarter of fiscal 2025 (ended April 30), compute and networking generated $22.7 billion in revenue, a 408% year-over-year increase. other. One of the subtle enablers of the compute and networking industry is Nvidia’s unified compute device architecture (CUDA) platform. CUDA is a program that allows developers to write software applications that run in parallel with Nvidia GPUs.
Although it may not seem like much at first glance, the CUDA infrastructure is extremely important to Nvidia. Let’s see why.
Why is CUDA important to Nvidia?
Currently, industry analysts estimate that Nvidia has a staggering 80% market share in AI-based chips. One of the primary use cases for AI chips is training and inference of AI models. Specifically, these applications are needed to create large language models (LLMs) such as ChatGPT, Anthropic, Claude, Gemini, etc.
CUDA is very lucrative for Nvidia because it provides the company with a complementary service that can be layered on top of its main GPU hardware products. This is a major differentiator for Nvidia because it allows the company to sell both hardware and software services related to AI development.
Additionally, as competition in the chip space continues to intensify, CUDA provides Nvidia with some flexibility beyond semiconductors. Additionally, given that software tends to generate much higher margins than hardware, I’m optimistic that Nvidia’s gross profit and cash flow profiles won’t take much (if any) hit if the The company is starting to lose part of its market. .
Discover more from The Times Of Update
Subscribe to get the latest posts sent to your email.