Nvidia’s results for the quarter that ended July 31 are with us, and the market didn’t seem impressed. After executives had finished their earnings call, its shares were down more than 8% after-hours, and almost 12% for the week. Bloomberg tells you all you need to know about this fascinating company here and here, while you can see the Bloomberg Opinion livestream on the results as they came in, featuring me, Dave Lee and Jonathan Levin, here, and listen to Cameron Crise’s Macro Man podcast here. In short, investors disliked that forecasts for the current quarter weren’t raised much, and didn’t hear enough reassurance on how quickly the company’s next-generation Blackwell chip will roll out. If an 8% after-hours selloff sounds extreme, put it in the context of Nvidia’s share price since its previous earnings announcement in May: (My apologies that the X-axis doesn’t include the dates for intraday prices; it’s an issue that should be fixed soon.) Over three months, Nvidia gained 50%, lost it all briefly on Aug. 5, and gained nearly all of it back. Even now, it’s up more than 20% in three months. The volatility is exceptional for such a large company, but only to be expected given the scale of what Nvidia is doing, and the degree of uncertainty over the future. Specifically, its extraordinary profit margins had risen every quarter since the launch of ChatGPT. This time, the decline was very slight. It still keeps more than 50% of sales as profit, five times the average margin for the S&P 500, so this shouldn’t worry anyone. Such strong profitability could not possibly have been maintained. But it does appear that people were disappointed that rising margins weren’t continuing in perpetuity: That might say something about exactly how realistic the hopes for the company had become entering the results. And if there is dissatisfaction with profit margins above 50%, it also suggests that the AI excitement may be past its peak. News stories about AI on the terminal have been declining slowly for a few months. As this chart shows, Chat GPT ignited the commotion, which peaked last November when Nvidia results combined with the drama of Sam Altman’s brief dismissal as OpenAI’s CEO: Another indicator, suggested by Peter Atwater of Financial Insyghts, is to look at the more speculative plays on AI, such as Super Micro Computer Inc., Michael Saylor’s MicroStrategy and even Bitcoin, which tends to move on the same swings of emotion. All of these reached a peak in the spring and have since sold off by more than 20% (far more than that for SMCI, which has featured in a negative report by the short seller Hindenburg): To quote Atwater: “Short of a resurrection in irrational retail investor behavior, the course ahead is set.” THE peak in the AI frenzy is behind us. Unappreciated by the crowd, it occurred in late February/early March. As they always do, the worst left the party first — and dramatically so.
This doesn’t mean Nvidia is going to burst, as it is still enjoying phenomenal growth. But if the excitement fizzles and speculative froth is removed just in time for rate cuts to come to the rescue, that’s probably healthy. Ahead of Nvidia, there was another landmark: Berkshire Hathaway topped $1 trillion in market value. It’s the eighth US company to do so, behind the Magnificent Seven. Warren Buffett took over Berkshire back in 1967, long before any of the Magnificents had even been founded, and the companies he’s owned since then have not innovated or disrupted on anything like that scale. There’s also a belief that it’s buoyed by the widespread veneration for its CEO. These criticisms are unfair. Even at a trillion dollars, Berkshire Hathaway is remarkably cheap. The presence of Buffett, now 91, has been a drag on the share price for the last quarter-century as investors ponder the impossibility of replacing him. The result is that the conglomerate trades very cheaply. Taking the classic value investor’s metric of comparing the company’s market cap to the total equity on its balance sheet (the book value that results from subtracting liabilities from assets), Berkshire trades at a 60% premium to its book value. Banks, still humbled 16 years after the Global Financial Crisis, are a bit cheaper. Apart from that, very little out there is better value than Berkshire. Dominated by insurance, Berkshire is cheaper than the average US insurance company, and much cheaper than the S&P 500 Value index; it’s better value than value. As for the Magnificent Seven, they trade at 13 times book: Nvidia traded at 63 times its book value before its results; it’s excluded to make the chart legible. Trusting balance sheets (and it’s much easier to put a price on Berkshire’s assets than on the products of the Magnificent Seven) reveals that years of compounding have left its equity worth more than $600 billion. That is more than double any of the Magnificent Seven. Indeed, Berkshire’s book value is greater than that of Nvidia, Apple Inc., Tesla Inc., Meta Platforms Inc. and Amazon.com Inc. combined: There is a good reason why Berkshire looks so much cheaper; book value probably understates the worth of the Magnificents, whose growth prospects are far better. You buy a company’s future earnings flow when you buy its stock. Nvidia’s earnings should grow far faster than Berkshire’s in the decades ahead. There’s also a warning. Buffett has always bought companies with a “wide economic moat” — a euphemism for well-entrenched monopolists. He’s great at finding them, but he’s needed to keep on the lookout. Three decades ago, the rap against him was that Berkshire was all about his successful bet on Coca-Cola Co. Now people say the same thing about Apple. The Magnificent Seven (particularly Nvidia) trade at such high multiples because they’re perceived to be monopolies. They might sustain this for decades. It’s more likely that the next generation of Buffett hopefuls will have to look around for new moated companies before long. | 
