Nvidia Stock: High Risk, Higher Return?

-32.67%
Downside
138
Market
92.84
Trefis
NVDA: NVIDIA logo
NVDA
NVIDIA

Nvidia stock (NASDAQ:NVDA) alone has created more value in the last three years than the combined market cap of 250 companies in the S&P 500. Talk about the 1% – specifically, $3 trillion! But everyone knows that. The bigger question is: what’s next? Should you buy, sell, or hold Nvidia at this point? Or should you look at other alternatives such as robotics surgery company Intuitive Surgery which looks poised to revolutionize the healthcare sector – potentially driving ISRG stock up 10x?

In our recent analyses, we argued for a range of outcomes: Nvidia’s road to $300 and the counterpoint Nvidia’s downside to $40.

Nvidia Stock Journey So Far

Relevant Articles
  1. Could Nvidia Stock Crash 50%?
  2. Trump Helps AI Stocks Like Nvidia. Here’s How
  3. Why Did Nvidia Stock Fall 5%?
  4. Can Fed Rate Cut Take Nvidia Stock To $200?
  5. Here’s A Better AI Pick Than Nvidia
  6. Could Nvidia Stock Crash To $40?

NVDA stock swelled over 10x from levels of $13 in early January 2021 to around $140 now, vs. an increase of about 50% for the S&P 500 over this roughly 4-year period. It was a bumpy ride, with returns for the stock being 125% in 2021, -50% in 2022, and 239% in 2023. The underperformance in NVDA’s stock vs. the S&P 500 in 2022 stands out in particular – as the benchmark index had returns of -19% that year. Notably, 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.

Now, below are the facts, and a concrete view of how you might want to think of Nvidia in your portfolio.

Three undeniable facts about Nvidia

  1. Nvidia is an arms supplier and the AI war is on
  2. AI isn’t just an input: it’s a strategic input. Companies from apparel to the energy sector to pharma giants don’t just need it as a hygiene factor – they need it to win vs. the competition. Meaning? These Nvidia’s customers cannot afford to be price-sensitive
  3. Heady growth comes with massive swings. Froth always includes water: the core, and the air. It won’t be smooth to watch Nvidia swing in your portfolio

So what should you do?

Answer: Add Nvidia to your portfolio in the amount such that while you sit envisioning a 5x gain in the long term, you’re also willing to stomach 50% loss at any time during the ride. Yep, be willing to bear short-term pain – the price you’ll pay – to be rewarded in the long run. This sounds simple in theory but is never easy to live by in practice.

Below are more specifics on the pros and cons.

The bad first

Here’s the thing, competition is heating up. AMD isn’t sitting idle – the company has doubled down on AI-focused GPUs and claims its new Instinct MI300X chip outperforms Nvidia’s in several areas. Intel is also emerging as a competitor, gaining traction with value-priced AI chips. Moreover, big tech players like Google – one of Nvidia’s biggest customers – are developing their own AI and machine learning silicon. With these rivals, the rapid growth in Nvidia’s revenues and the unusually high levels of Nvidia’s cash-flow margins (over 50% in recent quarters) may not be sustainable.

The underlying economics of the AI GPU market remain weak, with most of Nvidia’s customers still unprofitable. Why is that? Large language models are expensive to build and train and the payoff can take a long time. VC firm Sequoia estimates that the AI industry spent $50 billion on Nvidia chips last year, possibly exceeding $100 billion when including other costs. Yet, these investments have only generated about $3 billion in revenue, with few services besides ChatGPT gaining large, paying user bases. We may well be in an AI “FOMO phase,” where companies feel compelled to invest in AI simply because their competitors are. As shareholders seek better returns, capital spending may cool off, impacting companies like Nvidia.

Separately, large AI deployments follow two stages. The first is the compute-intensive training phase, during which AI models are built from large datasets with GPUs being used to process data and adjust the model’s parameters. The second stage is the deployment phase, where trained models run in real-world applications and typically require lower processing power, allowing companies to use less expensive hardware to run models for daily tasks. As the AI revolution began just about two years ago, most companies are in the training phase and there is a possibility that demand could ease as we move toward lower power inferencing.

The good and the great

Nvidia has had a head start in the AI market. And it has the foresight to not let its early mover advantage slip away. The company has built a comprehensive ecosystem around its AI processors, including programming languages and software, which can lock customers into its products as they scale up their AI investments. Nvidia’s extensive software stack – including CUDA, cuDNN, and TensorRT – not only enhances performance but also simplifies AI development, making it increasingly difficult and costly for companies to switch to alternative platforms. Look at it this way. As companies deploy their AI bets they will be spending time and resources on training, tuning their AI models, and integrating these tools into their broader IT systems and workflows. This makes it more difficult and costly to shift to competitors. The ecosystem could help to protect Nvidia’s margins, with software-related sales also likely expanding in the long run.

While the initial AI models deployed by the likes of OpenAI in 2022 were largely text-based, models are increasingly multimodal meaning that they work with speech, images, video, and 3D content – calling for higher computing power and a larger number of GPU shipments. Moreover, unlike a decade or so ago when advancements in computing power – particularly with processors – outpaced the development software that could fully utilize these capabilities, in the AI era, the demand for computing power has skyrocketed due to the intense computational requirements of machine learning models. This trend could result in sustained strong demand for Nvidia.

The shift in U.S. monetary policy could also give Nvidia an extra lift. The Fed’s 50 basis point cut – the first in nearly four years – brings the federal funds rate to 4.75%-5%, with room for further reductions. Lower rates boost growth sectors like tech by increasing the present value of future earnings. The rate cuts are particularly beneficial to Nvidia. Why?  Lower interest rates would reduce financing costs for builders of large data centers, potentially driving up capital spending in the space, and helping players like Nvidia which sells GPUs for servers. Moreover, the economics of the AI revolution remain mixed, given the high costs of model training and inferencing. A drop in interest rates could improve the financial feasibility of these investments. Check out our analysis of other ways to profit from the Fed’s next move? 

 Returns Oct 2024
MTD [1]
2024
YTD [1]
2017-24
Total [2]
 NVDA Return 15% 183% 5225%
 S&P 500 Return 1% 22% 159%
 Trefis Reinforced Value Portfolio 1% 16% 768%

[1] Returns as of 10/25/2024
[2] Cumulative total returns since the end of 2016

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