Over the last few years, Nvidia (NVDA +0.30%) benefited enormously from the increased investment in chips for artificial intelligence (AI), but news out of China is causing investors to wonder if the stock's high valuation is worth it ahead of potential risks to the massive investment big tech companies have put into data center infrastructure.
We'll look at potential pitfalls for Nvidia in the new year, before considering why Advanced Micro Devices (AMD +1.60%) might be a better buy right now.
NASDAQ: AMD
Key Data Points
Growth is starting to slow
When OpenAI released ChatGPT in 2022, companies started spending billions on GPUs to train AI models. Nvidia led the market for graphics processing units (GPUs), so it was the right company at the right time to deliver monster returns to investors.
In 2023, Nvidia's revenue and share price soared. In two years, grew into a very large company with trailing-12-month revenue of $113 billion. It's naturally going to get more challenging to keep growing at high rates with such large revenue figures.
Analysts still expect revenue to grow 52% in calendar 2025, according to Yahoo! Finance -- about half the rate it delivered over the past year. One risk that could cause the company to fail to meet those estimates is supply chain constraints limiting the production of its new Blackwell computing platform, which starts shipping this quarter.
Nvidia reported robust demand for its H200, which could pick up the slack until Blackwell revenue kicks in. But investors should be aware that Nvidia needs operators of data centers to keep spending more money every year in order to sell more chips, and there are questions about how sustainable this spending will be in the next few years.
Could China's DeepSeek be the catalyst for less spending?
China's DeepSeek says its AI model can perform on par with the best models from American AI companies, including OpenAI's ChatGPT, but the catch is that it cost less than $6 million to develop, which is incredibly cheap in the world of AI research. While there are analysts raising doubts about the start-up's claims, investors worry that big tech companies may start looking for ways to do more with less, as DeepSeek claimed it did. Since Nvidia is the leading GPU supplier for data centers, lower spending would have a direct impact on Nvidia's growth.
The data center market previously experienced cycles in growth that hurt Nvidia's business and sent the stock down a few times in the last 10 years. Big tech companies spent billions on data centers over the last few years, and there's a risk of too much investment leading to a surplus of computing capacity. That boom could lead to a corresponding downturn in data center spending, which would hurt Nvidia's bottom line.
All said, everything has to go right to justify the stock's valuation, which even after the recent dip on the DeepSeek news is still expensive at a forward price-to-earnings multiple (P/E) of 42. However, investors don't have to take on that risk when Nvidia's GPU competitor is trading at a much lower valuation.
More growth for half the price
AMD stock returned nearly 5,000% over the last 10 years and has more than doubled in the last five. The stock is trading at a lower P/E than Nvidia but offers comparable growth potential based on analyst estimates. It faces the same risk as Nvidia, but AMD's revenue from data center GPUs is expected to total just 20% of its total revenue this year. What's more, the stock's lower valuation offers a better risk-to-reward ratio.
Data by YCharts; PE = price to earnings, EPS = earnings per share.
There's one advantage for AMD that might give it an edge over Nvidia, and that's the company's focus on designing its GPUs for AI inferencing. AI training helps models digest mountains of data; inferencing is the process that allows AI models to process new information and make decisions in real time, which is essential for self-driving cars and AI agents.
AMD's MI300X chip has been used by Microsoft for its Azure cloud business, but the chipmaker says its new MI325X version delivers up to 20% higher performance for AI inference than Nvidia's H200.
AMD could be in a solid position to see tremendous growth for its data center GPUs in the coming years, and investors can buy the stock at just 24 times this year's earnings estimate. That is almost half the earnings multiple investors are paying for Nvidia shares, yet analysts expect AMD's earnings to grow over 44% annually over the next few years -- higher than the 38% they expect Nvidia's earnings to grow per year.
Given the high expectations built into Nvidia's share price ahead of near-term risks that could send the stock down, AMD stock seems a better bet. If data centers continue their rapid pace of investment in AI hardware, AMD could outperform its top rival. But if data center spending on GPUs slows down, AMD's lower P/E multiple will soften the blow, while the company could still see strong growth in other chip products, including central processing units (CPUs) for PCs.
