That said, past performance doesn't necessarily guarantee future results. Let's try to read the tea leaves to see if this iconic chipmaker is still a buy.
While generative artificial intelligence (AI) has been Nvidia's latest big break, the opportunity was virtually nonexistent before late 2022, when OpenAI's ChatGPT burst onto the scene. Before that, the company experienced a series of boom-and-bust cycles as its advanced graphics processing units (GPUs) were adapted to various bubble industries.
In the early and mid-2000s, Nvidia specialized in video gaming. However, the launches of Bitcoin and other cryptocurrencies created unprecedented demand for its GPUs for mining: solving complex computations to verify transactions and mint new digital coins.
Eventually, both opportunities (lumped into Nvidia's gaming segment) stagnated and declined, representing only $2.9 billion of the company's fiscal second-quarter sales -- under 10% of the total.
The data center segment now represents around 88% of its sales. While the company doesn't break this down product by product, most of its sales likely come from advanced AI chips like the H100 and H200, used to run and train large language models (LLMs). If LLM demand fades, much of Nvidia's recent growth will evaporate.
In the worst-case scenario, Nvidia's recent AI-led growth is on the cusp of a significant slowdown, if not an all-out crash. Granted, the company's growth looks spectacular -- second-quarter revenue jumped 122% year over year to $30 billion. But this could be based on an unsustainable AI race as capital-rich tech companies overspend on chips to avoid the perception of falling behind, even if they don't expect to profit.
The worst offender might be Meta Platforms, one of Nvidia's top customers. In 2024, the company expects to have $38 billion to $40 billion in capital expenditures, with much of that going to Nvidia GPUs. However, while CEO Mark Zuckerberg is confident in the opportunity, it is still unclear how the company expects to turn his optimism into returns.
Unlike Amazon or Alphabet, Meta doesn't have a cloud platform to rent out its computing power to start-ups. Furthermore, its flagship LLM, Llama, is open source and free, which will make it difficult to monetize.
The situation is reminiscent of Zuckerberg's last big idea: a virtual reality concept called the Metaverse that saw the company burn through around $46.5 billion with very little to show for it.
Meta isn't the only tech giant with an uncertain AI business model. Electric vehicle maker Tesla is also buying massive numbers of Nvidia chips to build its Dojo supercomputer, a project CEO Elon Musk admits is a "long shot."
Many of Nvidia's top clients have highly speculative AI strategies. Eventually, their shareholders could push back on all the capital spending, leading to a drop in chip demand.
While there are good reasons to be cautious about the stock after its big rally, the outlook isn't all grim. Analysts continue to develop spectacular projections for the AI industry, with PwC expecting it to add $15.7 trillion to the global economy by stimulating labor productivity and consumer demand. If this is true, Nvidia's millionaire-making journey is just getting started.
In the best-case scenario, the main obstacle to AI's monetization is technological limitations. And this is something Nvidia itself can overcome by providing better hardware, such as its new Blackwell-based AI chips. As the chips get faster and less energy-demanding, running LLMs will become cheaper, reducing the threshold needed to achieve profitability.
With a market cap of $3.5 trillion, Nvidia is the largest company in the world. And it looks unlikely to deliver multibagger potential from this point. While it is impossible to predict the future, the bear case looks more likely because it relies on fewer assumptions.
Right now, many of the company's clients aren't making enough tangible returns to justify their current AI spending. And while Blackwell chips could unlock a new bull run in the near term, longer-term investors should probably wait for a pullback before considering a position in the stock.