In 1999, Nvidia (NVDA +1.00%) revolutionized computer graphics with the invention of the graphics processing unit (GPU), a chip capable of rendering ultra-realistic visual effects. Nvidia GPUs have since become the gold standard in gaming and professional design. The company holds more than 95% market share in workstation graphics processors.
In 2006, Nvidia moved beyond graphics when it introduced CUDA, a programming model that turned its GPUs into general purpose processors fit to accelerate data center applications like artificial intelligence (AI). Nvidia GPUs are now the gold standard in data center accelerators, and the company holds an 80% to 95% market share in AI chips, according to analysts.
Not surprisingly, given its dominance in graphics and AI processors, Nvidia is frequently described as a chipmaker. But that definition is somewhat misleading, because it fails to capture the scope and direction of the business. Nvidia currently derives about 15% of total revenue from adjacent multi-hundred-billion markets.
NASDAQ: NVDA
Key Data Points
Nvidia has a $1 billion software and services business
Nvidia reported revenue of $18 billion in the third quarter. That figure becomes $72 billion when annualized (multiplied by four). GPUs for gaming, professional design, and data centers accounted for most of that total, but management said software and services revenue would exceed an annual run rate of $1 billion by year end. That means Nvidia derives about 1.3% of total revenue from software and services.
That percentage may sound insignificant, but the figure is growing rapidly. The three products detailed below form the core of the software and services business. The first became available in late 2021, but the other two became available less than one year ago. As such, Nvidia has rapidly turned software and services into a billion-dollar business.
- AI Enterprise is a suite of software and frameworks that supports the development of AI applications across a range of disciplines, such as recommender systems in retail, intelligent digital assistants in customer service, and route optimization in logistics.
- DGX Cloud is a cloud service that combines supercomputing infrastructure, pretrained models, and Nvidia AI Enterprise software to create an AI-as-a-service offering. Businesses can use DGX cloud to build AI applications without actually purchasing Nvidia GPUs.
- Omniverse Cloud is a cloud service that combines GPU-accelerated infrastructure with design and simulation software. It supports 3D application development and synthetic data generation, which can be used to train autonomous robots and self-driving cars.
For context, management says AI Enterprise, DGX Cloud, and Omniverse Cloud collectively represent a $300 billion addressable market. That means they account for almost one-third of the company's $1 trillion total addressable market.
Here's the upshot: Nvidia is more than a chipmaker. It is a full-stack accelerated computing company, because its portfolio spans infrastructure, software, and services. That makes Nvidia a one-stop shop for graphics and AI, which cements its leadership status in both spaces.
To quote CFRA analyst Angelo Zino, "We believe ... [Nvidia's] full-stack AI and software capabilities provide an incredible competitive moat."
Nvidia has a $10 billion networking business
In 2019, Nvidia extended its data center presence by acquiring networking company Mellanox. Nvidia has since melded its own innovations with Mellanox technology to great effect. Networking revenue nearly tripled over the past year, becoming a $10 billion business. That means Nvidia derives about 13.8% of revenue from networking infrastructure, a market expected to approach $200 billion in 2024.
Moreover, Nvidia networking products are purpose-built for AI. Its InfiniBand networking platform is the gold standard in AI infrastructure for large-scale consumer internet companies and cloud providers, and it has already been adopted by Microsoft, according to CFO Colette Kress. Additionally, Nvidia recently announced Spectrum-X Ethernet networking for enterprise AI. That platform will be available early this year.
Here's the upshot: Nvidia has improved its ability to monetize AI with InfiniBand and Spectrum. But those platforms also make its entire product portfolio more attractive by extending its full-stack strategy. Larger cloud companies and smaller enterprises can purchase all the requisite technology for developing, deploying, and running AI workloads from a single vendor. That drives cost savings by simplifying implementation and reducing complexity.
To quote Nvidia CEO Jensen Huang, "Nvidia's value proposition at the core is we are the lowest cost solution."
Nvidia stock is a worthwhile long-term investment
Nvidia was the best-performing S&P 500 stock last year, as enthusiasm surrounding generative AI drove its share price 239% higher. But the company remains well positioned to create value for patient shareholders, especially as its ability to monetize AI improves through continued success with software, services, and networking products (currently about 15% of total sales).
Going forward, the graphics processor market is expected to increase at roughly 28% annually through 2030, while the AI market is projected to increase at 37% annually during the same period. That gives Nvidia a good shot at annual sales growth exceeding 25% through the end of the decade.
In that context, its current valuation of 31 times sales is tolerable, despite being a substantial premium to the three-year average of 23.6 times sales. To be clear, Nvidia is not without risk, and shares are not cheap. But patient investors should consider buying a small position in Nvidia stock today, especially as part of a larger basket of AI stocks.
