Will AI’s Tough Economics Pop Nvidia’s Bubble?

Nvidia (NASDAQ: NVDA) stock has seen a 120% plus rally this year. With a $2.8 trillion market cap, Nvidia’s valuation is in the vicinity of tech titans Microsoft (NASDAQ:MSFT) and Apple (NASDAQ: AAPL).  Nvidia has a lot going for it as companies are looking to shift their sizable installed base of traditional data centers to accelerated computing using graphics processing units or GPUs to perform more artificial intelligence tasks. The stock currently trades at about $113 per share or roughly 41x forward earnings. While this multiple may seem high, it is not unreasonable given Nvidia’s recent growth. Nvidia’s financial performance has been solid, and the company is on track to grow revenues by almost double this year, reaching approximately $120 billion, with earnings per share poised to double. However, there are significant questions about the underlying economics of the end market for GPU chips and the broader AI ecosystem, as most of Nvidia’s customers remain loss-making.

Setting up AI systems and training large language models is an expensive affair. This raises concerns about how the massive investments companies are making in artificial intelligence infrastructure will ultimately pay off. To provide context, venture capital firm Sequoia recently estimated that the AI industry spent approximately $50 billion on Nvidia chips to train advanced AI models over the last year. Considering other ancillary investments such as buildings and power generation, it is probably safe to assume that the total amount required to set up AI data systems could be about double this figure or about $100 billion spent over the last year. However, the VC firm estimates that these investments generated a mere $3 billion in revenue.

Although this is still early days for the AI industry, we have yet to see many AI services, besides ChatGPT, that have a reasonably large base of paying customers. For instance, customers readily pay $10 or more per month for music or video streaming services such as Spotify and Netflix, but we have yet to see AI services that offer similar utility to a broad customer base. In fact, OpenAI itself could incur losses amounting to $5 billion in 2024, according to a report in The Information, meaning that the company will have to raise even more funds beyond the seven rounds of funding that raised over $11 billion.

Considering that markets project Nvidia’s revenues to grow to almost $160 billion in FY’26, with the bulk of this growth coming from AI-focused GPUs, it remains to be seen whether the revenues of its end customers will grow in tandem to justify their big spending. While AI appears to be a generational opportunity in the computing space, it will take time for clarity to emerge regarding the use cases and what customers will be willing to pay for these services. If Nvidia’s customers can’t make money, how can Nvidia sustain its high growth rates and thick margins?  Investments into AI data centers at this juncture appear to be driven, in part, by panic or the fear of missing out on the next big thing in tech, rather than by careful consideration of potential returns. As a result, customers may begin seeking cheaper alternatives for chipsets beyond Nvidia’s offerings, which typically cost over $25,000 each. This could also open the door for other chipmakers, such as AMD (NASDAQ: AMD) and Intel (NASDAQ: INTC). Although Nvidia’s chips are currently superior in terms of overall performance, with the company also using proprietary software and programming languages to better lock in customers, increased competition may render Nvidia’s current revenue growth rates and abnormally high margins unsustainable. For perspective, Nvidia posted net margins of 57% in Q1.

NVDA stock has seen extremely strong gains of 785% from levels of $13 in early January 2021 to around $115 now, vs. an increase of about 45% for the S&P 500 over this roughly 3-year period. However, the increase in NVDA stock has been far from consistent. Returns for the stock were 125% in 2021, -50% in 2022, and 239% in 2023. In comparison, returns for the S&P 500 have been 27% in 2021, -19% in 2022, and 24% in 2023 – indicating that NVDA underperformed the S&P in 2022.
In fact, consistently beating the S&P 500 – in good times and bad – has been difficult over recent years for individual stocks; for other heavyweights in the Information Technology sector including AAPL, MSFT, and ORCL, and even for the megacap stars GOOG, TSLA, and AMZN. In contrast, the Trefis High Quality (HQ) Portfolio, with a collection of 30 stocks, has outperformed the S&P 500 each year over the same period. Why is that? As a group, HQ Portfolio stocks provided better returns with less risk versus the benchmark index; less of a roller-coaster ride as evident in HQ Portfolio performance metrics. Given the current uncertain macroeconomic environment with high oil prices and elevated interest rates, could NVDA face a similar situation as it did in 2022 and underperform the S&P over the next 12 months – or will it see a strong jump?

We value Nvidia stock at $89 per share, which is approximately 21% below the current market price. See our analysis on Nvidia Valuation: Is NVDA Stock Expensive Or Cheap? for more details on what’s driving our price estimate for NVDA stock. Also, see our analysis of Nvidia Revenue for a closer look at the company’s key revenue streams.

[1] Returns as of 7/29/2024
[2] Cumulative total returns since the end of 2016

Invest with Trefis Market-Beating Portfolios
See all Trefis Price Estimates