This article was originally published on Fool.com. All figures quoted in US dollars unless otherwise stated.
Key Points
- Nvidia's valuation remains surprisingly fair compared to its growth and profitability metrics.
- The company's great fundamentals may be concealing a hidden risk.
Three years ago, OpenAI's ChatGPT hadn't even launched. Today, it is the leading software service in the multibillion-dollar generative artificial intelligence (AI) industry.
Nvidia (NASDAQ: NVDA) also plays a massive role in that arena. But with its shares up by more than 1,000% over the last three years, investors have to wonder how much growth potential is left for the chipmaker. Could it still be a compelling long-term buy?
Nvidia's stock is surprisingly reasonable
With a market cap of $4.63 trillion, Nvidia is the largest company in the world. And investors could be forgiven for assuming it's trading at sky-high multiples that are detached from its fundamentals.
But that isn't the case. With a forward price-to-earnings (P/E) multiple of just 28, Nvidia stock has a surprisingly reasonable valuation compared to the Nasdaq-100's average of 26, particularly when considering its explosive top- and bottom-line growth rates.
In the second quarter, its revenue soared 56% year over year to $46.7 billion, driven by strong sales in its data center segment, where the company designs and sells the cutting-edge AI chips most commonly used to train and power large language models (LLMs). Despite selling physical products, it boasts a software-esque gross margin of 72.2%, which shows that its chips continue to be well differentiated from the competition because of factors like CUDA, the proprietary software platform that helps developers program its chips for specific tasks.
Investors can compare this to the way Apple synergizes its iOS with its iPhones to make them work better together, and tries to create a walled garden of services and apps that makes it harder for users to leave the ecosystem for alternative hardware, even if rival devices have comparable raw technical stats.
Nvidia's economic moat gives it immense pricing power, which flows straight to its bottom line. Second-quarter earnings per share jumped 61% year over year to $1.08, and analysts expect more growth in the future.
What could go wrong?
Nvidia's CUDA platform and its cutting-edge chip design have combined to secure it a dominant position in the market for generative AI hardware. And its customers continue to spend eye-popping sums on its wares. According to The New York Times, big tech's data center buildout is actually accelerating, with Alphabet, Microsoft, and Amazon spending a combined $112 billion on capital expenditures over the last three months alone. Much of that cash went to Nvidia's high-priced AI chips for data centers.
On the surface, it looks like everything is great. Nvidia has a strong economic moat, its customers remain willing to buy its products, and its valuation remains incredibly low compared to its earnings growth. What could go wrong?
The short answer is the technology itself. Right now, generative AI doesn't seem to be commercially viable on a large scale because of challenges with LLM accuracy and the immense costs needed to run and train the algorithms. (Nvidia's high prices play a role in inflating costs throughout the industry.)
The losses are getting hard to ignore. For example, ChatGPT creator OpenAI expects to lose $9 billion this year on $13 billion in revenue, a burn rate of around 70% of sales. Losses are expected to increase as it scales up its operations.
It is normal for growth companies (and by extension, industries) to generate losses when they are in their early expansion phases, but the AI buildout seems to be based on some core assumptions that may not materialize. There is actually no guarantee that today's generative AI systems will evolve into more useful forms like artificial general intelligence (AGI), which is a currently theoretical technology that would be able to solve problems and learn without human input.
Many analysts are already arguing that LLMs' development may be reaching a point of diminishing returns. And if the technology doesn't become dramatically more useful, Nvidia's customers may begin to rethink the vast amount of resources they are committing to their infrastructure budgets.
What do the next three years have in store?
With its relatively low valuation and strong customer demand, Nvidia stock looks unlikely to crash anytime soon. But it's also hard to get excited about it as an investment now, considering the extreme hype and lack of profitability on the consumer side of the AI industry.
In light of all that, Nvidia stock looks to me like a hold for now. And investors who already own shares should consider taking some profits.
This article was originally published on Fool.com. All figures quoted in US dollars unless otherwise stated.
