Companies are spending big to develop their own AI chips so they don't have to keep paying Nvidia
Within a few years, AI chips will sell at a small fraction of Nvidia's 'only game in town' price.
Artificial intelligence and seven (supposedly magnificent) stocks have been driving the U.S. market's performance for a while. I agree with the market - these companies are magnificent. I'd want my kids to work for any of them.
But there's a magnificent difference between a magnificent company and a magnificent stock - the difference lies in magnificent valuation. The price you pay for the shares of a company, even the magnificent ones, matters. Of course, I could have written these same words about Nvidia $(NVDA)$ when its shares were much lower. I'll always look dumb questioning apostasy until I'm not. I'm used to it.
I'm fairly certain what I'm about to say will age well, though not in a linear way. Let me tell you a story.
I was talking to a friend. He told me a tree had fallen on his almost-new car. His car was totaled, and he got a check from insurance. He still had the memory of buying a car during the pandemic, how difficult it was and how expensive cars were. He said, to his shock, when he went to buy a replacement, he found car dealers fighting for his business. He got a car at a huge discount to the sticker price, because the market is oversupplied with cars. Predictably, high prices a few years ago led to higher supply today, and thus lower prices.
Then the conversation (as you would expect) shifted to the stock market and, of course, Nvidia. He asked me what I thought about it.
I told him that the market is now flooded with cars, despite them being difficult to get a few years ago. And the same is happening with GPU chips.
A few years ago, some people thought high prices would persist in the car market indefinitely. Though I imagine most people thought that at some point it would end in lower prices. Here's all you need to know: The laws of economics work the same way with microchips.
Today, Alphabet $(GOOGL)$, Apple $(AAPL)$, Meta Platforms, Tesla $(TSLA)$ and Amazon $(AMZN)$, for example, are buying a very large number of chips from Nvidia, as it is the only game in town. Nvidia's skyrocketing profitability is at their expense. Actually, it's a capital expenditure - only a portion of their spending that shows up in Nvidia's revenue shows up in Alphabet's earnings (income statement expenses), for example.
As competing products hit the market, Nvidia's sales will falter, and so will its margins and earnings.
Let me explain, using Alphabet as an example: Nvidia is currently selling some of its AI microprocessors for $40,000 a pop. Microprocessors are capital expenditures, thus they are depreciated over five years or so. Per accounting rules, only $8,000 of the $40,000 Alphabet pays to Nvidia shows up in Alphabet's income statement in the form of depreciation, while the full $40,000 shows up in Nvidia's revenues. That's why these companies' free cash flow often is much lower than their income.
These tech giants aren't pleased to be paying Nvidia for their needs. So, in addition to Intel $(INTC)$ and AMD $(AMD)$, these companies are spending big to develop their own AI chips. Moreover, dozens of other companies we haven't heard of are working on AI chips.
Fast-forward a few years, and AI chips will sell at a small fraction of Nvidia's "only game in town" price. The cure for high prices is high prices. This is what I love about capitalism.
The argument I hear about Nvidia stock is that it's not insanely expensive, as it is trading at somewhere around 30-40 times future earnings. True. Though it's a high valuation for a $3 trillion company, it's not "insane."
But, and this is a huge but, the "E" in this P/E calculation is a bit misleading and provides a false sense of security. As competing products hit the market, Nvidia's sales will falter, and so will its margins and earnings.