It's been an incredible run. Nvidia is now the second-best performing stock in the S&P 500 -- up 148% this year, just behind Super Micro Computer, another name that readers of this newsletter know very well.
In terms of when to sell a stock, I often draw on the investing wisdom of Jeff Vinik, a one-time hedge fund manager and former portfolio manager at Fidelity. Vinik has said in prior interviews that he sells stocks when fundamentals deteriorate, valuations get high, or he finds better ideas.
None of those things seem applicable to Nvidia today.
Over the last few months, the fundamentals behind the AI boom have only gotten better. In May, Nvidia posted eye-watering sales growth of 262% for its fiscal first quarter and earnings growth of 461% versus the prior year.
Nvidia's valuation is about 35 times 2025 consensus earnings, according to FactSet. That's not crazy given that sales are forecast to grow 37% for the time period.
Meanwhile, the industry backdrop for Nvidia keeps improving. This past week, JP Morgan published survey results from 166 chief information officers, or CIOs, who are responsible for $123 billion in annual enterprise tech spending. The survey suggested that generative AI hardware spending could rise by more than 40% annually over the next three years, going from 5% of IT budgets this year to 14.5% in 2027.
PiperSandler recently issued a similar CIO report showing that nearly half of enterprises shifted from an AI testing phase to an implementation phase over the past year.
While we spend a lot of time talking about AI doomsday narratives, the real spending is focused on simpler, less controversial goals: making companies more efficient. Most of the spending on graphics processors from Nvidia is to help derive insights and analysis from the pools of unstructured data inside corporations.
Companies are reworking their computing infrastructure from a traditional model of information and file retrieval to the new generative AI approach where answers and insights are generated on demand. In a competitive landscape, companies have no choice but to make that move.
Investors have driven Nvidia higher, but they may still be under appreciating the opportunity. That was a key point from Coatue Management founder Philippe Laffont during a Bloomberg investing event last week.
Laffont noted that some $100 trillion, in today's dollars, has been invested in CPU infrastructure since the PC was invented. All of that will need to be replaced with GPU-focused equipment over time. There are trillions in spending still to come.
Coatue has been a big investor in Nvidia. The company's latest 13F filings shows a stake worth $1.25 billion, as of March 31. A Coatue spokesperson declined to provide an update on the firm's current Nvidia holdings.
Even Nvidia bulls may be asking the question of whether it's time to trim their positions in the stock. This is the hardest decision in investing.
Here too, Laffont has some advice. He cautioned against being "too cute."
"Stan Druckenmiller once told me he made 120% of his money in obvious ideas, and he lost 20% everywhere else. I feel that's very true," Laffont said. "By far the biggest mistakes that I've made is selling stock early."
Laffont said that Coatue missed out on $20 billion in additional gains in Tencent, because it sold stock too soon two decades ago.
To be sure, the opportunity in AI is no longer a secret. But that doesn't mean the stock gains are over. Laffont, for one, remains bullish. He points to Apple's huge decade of gains, even as the iPhone dominated the conversation in tech.
A similar multiyear opportunity lies ahead for Nvidia. Until the facts change or valuations get crazy, it's best not to get too cute.