Nvidia drives a new wave of infrastructure investment
When we first bought Nvidia in 2017, the catalyst was that large companies started buying Nvidia Graphics Processing Units (GPUs) for processing data for artificial intelligence (AI) calculations (see more here in our 2017 letter). Since the company’s products had traditionally been used for video games, the addition of enterprise customers hinted at a new growth opportunity.
But back in 2017, we couldn’t foresee just how central Nvidia products would be for the massive data centers that do AI calculations. Over the last few years that has become more apparent as Amazon, Microsoft, Google, and other large companies have been spending billions of dollars on Nvidia GPUs. In 2022, their data center business ($15B in Sales) became larger than the video game business ($9B in Sales).
Source: FactSet. The gray bars are estimated Data Center sales for the company for 2023, 2024, and 2025.
Now in 2023, the data center business is taking a massive leap forward to $27B in annual sales according to current estimates. Much of this increased demand is due to the latest waves of AI models like ChatGPT. But it’s more than just ChatGPT. The world’s data centers are reorganizing around GPUs to do work that was traditionally done on Intel Central Processing Units (CPUs). About 25% of the Fortune 100 are currently making this shift. Nvidia has built a lead in both software and hardware that means they are the best performing solution for these workloads.
The cloud computing giants (Amazon, Microsoft, Google) are also beneficiaries of this renewed interest in artificial intelligence, fresh off the heels of recent concerns about a slowdown. Just a few short weeks ago Wall Street analysts were spooked that Amazon Web Services (AWS) growth had slowed to 16% from 40% a year ago. But now AWS and the other cloud computing platforms should see a dramatic about-face, as their customers want to rent a lot more Nvidia GPUs from them.
As long term investors in Nvidia, we are humbled by how well it has worked for our clients. We’re up 15x from the initial purchase price in 2017, and we have to admit that we didn’t exactly see this coming. It wasn’t easy to hang on either—we experienced two painful 50% declines during this period.
The lesson is clear: long term investors can succeed even without perfect foresight if they remain patient with a highly profitable business that exhibits strong competitive advantages. Only in hindsight can we see that even the optimists weren’t optimistic enough.
Best regards,
Evan McGoff
P.S. The name of the company is pronounced EN-vid-e-ah. As in, they are the envy of other chip makers.
Glossary of terms:
CPU – Central Processing Units. The primary chips that do the math calculations that do most computing work. They are versatile, but in recent years they haven’t gotten much faster. Intel has dominated this market for decades.
GPU – Graphics Processing Units. Primarily used for video games, these more specialized chips are good at doing many small calculations at once. While initially made to draw 3D scenes for video games, these chips happen to be well suited for performing the calculations that more recent kinds of artificial intelligence require.
Cloud computing – Large data centers where companies rent out computing capacity to other businesses. Amazon and Microsoft are the largest players in this market. They buy massive amounts of hardware and rent them to businesses—who find this to be more cost effective than buying and managing the hardware themselves.
Artificial Intelligence – Using large amounts of data to make predictions, and to generate new output. Amazon uses AI to guess what products you’ll want to buy based on past purchases. Newer kinds of AI generate audio, images and text.
ChatGPT – A “chatbot” created by OpenAI that will generate text in response to a prompt. It can generate whole paragraphs of text based on your instructions in a variety of languages, including computer programming languages.
Disclosure: Dock Street Asset Management, Inc. and our clients own Nvidia (NVDA), Amazon (AMZN), Microsoft (MSFT), and Google (GOOG). This article is not intended to be used as investment advice.