Firstly, it’s worth noting I haven’t published in over two years. These pieces take a lot of research, which detracted from MBA homework, job recruiting, and networking. However, it was nice writing about tech/healthcare private equity and VC investment ideas and targets. Since I last posted on Pins and Needles, I had a stint at Akuna Capital in a venture capacity where I looked at a number of tech startups— including one starting the NASDAQ of music and another applying machine learning to the commodities trading space. PE/VC has eluded me since but I’d like to stay close to my passion, investing.
For that reason, I’m looking to the public markets, where I can take more immediate action with capital I’ve saved over time (rather than just hanging around the hoop for a PE/VC role). What are my criteria? I’m looking for a tech opportunity that (1) benefits from strong secular tailwinds (drivers which are irrefutable such as the aging population), (2) demonstrates customer stickiness, (3) exhibits market dominance, and (4) is growing at a rocketship clip. I also want to see double digit returns in 12 months. Enter NVIDIA.
There’s a lot of hype around the graphics processing unit (GPU) and chip manufacturer NVIDIA right now. The company fits the criteria above. Tech companies and consumer internet companies are putting all their chips (no pun intended) on LLMs and generative AI research. They are purchasing GPUs from NVIDIA in droves. You don’t need to believe we’re going to see a breakthrough from scores of companies; you just have to believe all these companies will continue to chase gold for 12-24 months. In the early days of the internet, we saw a similar gold rush. That lasted a few years. Even if you’re a cynic and believe only OpenAI and Gemini will succeed (based on their ability to amass talent), we’re still in early-to-mid stages of this gold rush. Tech companies continue to flourish even after the early 2000s bust, as do crypto ventures. For that reason, I’d call LLMs a secular tailwind.
NVIDIA and its underlying CUDA API is the foundation upon which software plugs in and plays. It’s difficult to switch to a competitor like AMD (Advanced Micro Devices) once you’ve built on top of CUDA. As one Quora user says, the machine learning community develops software for CUDA first and foremost. Of course, some say open source frameworks like Jax and Pytorch are becoming more popular, loosening the hold of CUDA (and stickiness of NVIDIA). But I bet the largest software purveyors and game developers will want to use the best graphics cards in business, and NVIDIA continues to innovate even as recently as a couple months ago with the GeForce RTX 40 Super Series. They are as blue chip (once again, pun not intended) as it gets.
NVIDIA’s the market dominant player. Their main competitor is AMD. But AMD’s financials reveal $850m in net income in 2023 (against $22.7Bn in revenue) - I question the company’s operational efficiency with those stats. Furthermore, they’re a much smaller player - NVIDIA saw revenue of $26.9Bn in 2021. Today, NVIDIA realized revenue of $61Bn in FY2024 (which is basically close to our CY2023). Its share of the graphics card market has been estimated at 80% in 2023. That figure has only increased in recent months, reflecting the choice of large tech buyers. NVIDIA has the weight to invest in R&D ($8.7Bn in FY2024) and stay ahead of the curve.
NVIDIA is growing quickly. Revenue has grown at 50% CAGR over past two fiscal years, most of which has been in the past 6 months. Gross margins have steadily improved as well from 65% to 73%.
This isn’t a Seeking Alpha or Reuters article. My goal isn’t to create a watertight argument for NVIDIA or introduce you to its various divisions (Data Center, Gaming, etc.). You can peruse any number of online sources for that. My goal is to assess the downside case and whether NVIDIA is still a strong investment opportunity despite the stock price increase past $800. I am after all investing myself, so want to protect the investment.
The back-of-the-envelope financials at the end here have most inputs in blue and calculations in black. Assumptions for FY2025 are in purple font, yellow highlight off to the right. The stated price is the price as of early Feb in the year the 10-K is released to reflect public reactions to the SEC filings. We calculate P/E and AV/Adj. EBITDA ratios for NVIDIA in the recent few years as well as the projected year. Our goal is to see whether AV/EBITDA in 2025 is low enough that it presents a cheap buying opportunity, exerting upward price pressure on the stock. How did we project FY2025 (YE Jan 2025)? Quarterly and conservatively:
First of all, we saw low revenue in FY2023 - this was due to a slowed macro environment and tempered gaming demand. We similarly assume a sequential 44% dip in Q2’25 and 23% dip in Q3’25 in gaming revenue, in case the same thing happens again in FY2025.
For Data Center GPU revenue in FY2025, we assume quarterly revenue increases of 10%, 20%, 27%, 1%. The historical increase is shown to the right, and these assumptions are in line and reflect slower growth than FY2024.
NVIDIA has a customer concentration problem - where 13% of its revenue comes from one direct customer. What happens if this customer switches vendors? A doomsday scenario. We bake that into the 2H’25 revenue projection in the attached model. We still assume R&D and SG&A (particularly sales) stay in line with revenue if that customer loss happens.
We keep diluted shares outstanding flat.
Also assume increased market competition in 2025 from Intel and AMD as well as relaxed GPU buying. This gets reflected in a lower P/E - 50x vs 67x or 62.9x after FY2022.
The result of all this is net income of $46.5Bn in 2025P, and a price of $933, implying an increase of shareholder value of 17%. Maybe the real answer is a 14%-20% swag on share price appreciation. I’m fine with that error bar. I think a rough approach is arguably better than a 3-statement model given how difficult it is to predict the future. For the level of risk we’re underwriting, 17% is a pretty good increase in one year. Many PE firms deliver 20% gross IRRs and 16% net IRRs, after fees.
Let’s wait a year and see if we’re right.