Category Archives: James’s Analyses

The risk of NOT investing in Nvidia

November 19, 2024
James Stevens

Over the past year, I’ve repeatedly encountered articles and individuals expressing concern about buying Nvidia stock at its current price. Much of the concern seems to stem simply from the fact that the price of NVDA has risen so rapidly over the past year. Today, it closed at $147.01, near its all-time high, and about 190% higher than its price one year ago. The phrase “AI bubble” has been thrown around quite a bit in the context of Nvidia’s stock price. However, this growth alone does not necessarily mean the stock is overpriced. Compared to earnings, it’s about as “cheap” as it has consistently been over the past several years. In fact, there is potential risk in not investing in Nvidia that I haven’t seen anybody focusing on.

Here’s a chart showing Nvidia’s stock price, trailing 12-month EPS (earnings per share), and trailing P/E ratio (price-to-earnings ratio) over the past 7 years.

Nvidia’s stock price, TTM EPS, and P/E. Source: macrotrends.net.

You can see that the trailing P/E ratio has remained relatively consistent over the past 5 years (aside from a brief spike in early 2023, which turned out to be investors correctly predicting forthcoming massive growth). The stock price has grown rapidly because earnings have grown rapidly, and the growth in stock price has remained roughly proportional to the earnings growth. So, if you think Nvidia is overpriced right now, then you should have also considered it overpriced in November of 2017, and if you’d avoided investing in it at that time, you would have missed out on its subsequent 2600% gain.

Moreover, Nvidia’s current forward P/E of 35.71 is not much higher than that of other stocks that you probably consider quite stable and reliable. Walmart’s is 30.86. Apple’s is 30.67. Procter & Gamble’s is 24.51. But, you might ask, does Nvidia’s assume unbelievably fast growth over the next year? Well, if you were to instead calculate a forward P/E without any assumed growth by multiplying their earnings from last quarter (Q2 2024, $16.6B) by four, you’d get a “forward” P/E of 55. Considering that Nvidia’s Q2 net income was 2.7 times its Q2 net income one year ago, its P/E ratio seems quite reasonable to me, no matter how you look at it.

Over many years, Nvidia has developed, and continually improved, an unmatched, powerful computing platform and associated software stack. They have about 90% of the GPU market share, and that share has been increasing. Businesses have recognized the value of generative AI and how essential Nvidia’s chips are to powering the machine learning algorithms involved; as a result, Nvidia’s Blackwell superchips are sold out for the next 12 months. Skeptics keep asking, what happens if generative AI doesn’t turn out to be as valuable as businesses hope? Yet the trend we see is that these AI tools have been continually improving. So my question for the skeptics is: what happens if generative AI continues to increase in value?

There is risk in investing in any company. Right now, Nvidia doesn’t seem like a highly risky investment to me.

In fact, I see risk in not investing in Nvidia. Generative AI is already making many workers more productive and decreasing the differences between better workers and less productive ones. To create their AI models, tech companies are hoovering up data, possibly your data. If you’ve ever posted anything on the internet, it may very well be used to train an AI tool to perform a task that you’re good at, and you likely won’t be compensated. As a simple example, an AI model can use text you’ve created to learn how to write, answer questions, and carry on a conversation. The one job that you’d think would be reliable during an AI-driven industrial revolution, software development, is already being augmented by AI coding assistants.

Exactly how generative AI will affect worker productivity, salaries, and unemployment is highly uncertain. Right now, many C-suite executives think generative AI will shrink workforces. Others are quick to remind us that, in the past, many new innovations have created more jobs than they destroyed. Yet, even if this happens, generative AI advancements will lead to disruption in the job market and changes to the skills required to remain a valuable worker. And as for hypothetical job creation, the invention of the automobile may have created more new jobs than it destroyed, but in this analogy are we actually the human workers? Because it also led to a lot of unemployed horses.

We cannot predict the future, but in any scenario, tech companies will be using your own marketable expertise to train models to do what you do. Without laws ensuring that you’re compensated, the only way for you to get a cut from the businesses that will create uncertainty and instability in your employment may be to become a partial owner of said companies, i.e., to invest. If 10 years from now, you’re desperately trying to re-train in the midst of a surprisingly challenging job search, you may wish you had invested in the businesses powering the disruption. At the very least, you could consider it a hedge against one possible outcome.

While it may not yet be clear which software companies will dominate the AI space, it has been evident for a long time that one hardware company will power the revolution for the foreseeable future: Nvidia.

Disclosure: I am long NVDA, obviously.

Disclaimer: Investing involves risk. Do your own research. Before making any investment decisions, seek advice from a registered investment advisor, which I am not. By accessing this site, you acknowledge that the author holds no liability for any profits or losses you incur.

What happens to the stock market when interest rates drop? Historical data provides some insight.

January 30, 2024
James Stevens

The Federal Open Market Committee (FOMC) is expected to cut interest rates multiple times in 2024. Given the conventional wisdom that the Fed funds rate affects stock market performance, many may wonder how such cuts might affect their investments.

To determine how the stock market has reacted in the past to rate cuts, I gathered and analyzed historical market and interest rate data going back to 1955. As you might expect, the results indicate that, on average, the market performs significantly better when interest rates are decreasing.

The following chart illustrates the phenomenon. You may be able to observe it in a few spots on the chart, and with a bit of mathematical analysis, we can quantify these observations.

Monthly interest rate and S&P 500 data since 1955. Blue: Annualized, cumulative trailing-6-month Fed funds rate (T6MFR). Green, red, grey: Future-12-month S&P 500 return when T6MFR is decreasing (green), increasing (red), or flat (grey).

I’ll provide additional mathy details below, but the high-level takeaway is that, on average, the S&P 500 gained 10.40% over 12-month periods that follow a declining Fed funds rate, compared to 5.98% when interest rates had increased. That’s a big difference, and keep in mind that these averages include periods following small, incremental interest rate changes, which we are likely to see multiple times this year, and which may continue on an ongoing basis until the Fed decides to stop cutting rates.

The difference between a 5.98% return and 10.40% return is substantial, especially if compounded over a longer time period. The fact that these averages represent almost 70 years of data, and that this period includes many distinct instances of increasing and decreasing interest rates, gives me some confidence that this result is meaningful. At the same time, as you can see in the chart, the S&P returns over 12-month periods have varied wildly, including several sub-20% losses, as well as many super-30% gains. Most years are not “average,” and countless (unpredictable) factors affect market movements. We’ll just have to see how “average” this period of decreasing interest rates is.

Finally, here are the answers to the questions that the mathematician and scientist readers have been muttering at their screens up to this point: S&P data set was not adjusted for inflation. For each month, I calculated a “current” Fed funds rate as the annualized, cumulative rate over the preceding 6 months. I considered this rate “increasing” or “decreasing” if it changed over those 6 months by more than +/- 0.1 percentage points, and “flat” if it changed by less. If I adjust either of these 6-month periods to 3 months, the resulting averages are similar. If I adjust the 0.1% delta to 0.2%, the resulting averages are similar. I used the arithmetic mean to compute these averages; if you’d like to see averages computed using the geometric mean, check out my spreadsheet.

Please send me a message if you have any feedback, or even, heaven forbid, find a bug in the spreadsheet!