It’s nearly impossible to read or listen to anything even remotely related to artificial intelligence (AI) and not find a reference to Nvidia. The company’s graphics processing unit (GPU) chipsets are perhaps the single most important piece of architecture used in generative AI.
Don’t believe me? Industry research suggests that Nvidia held 98% of GPU shipments over the last two years; meanwhile, Jon Peddie Research estimates that Nvidia owns 88% of the GPU market. With a stat line like this, is it fair to say that Nvidia is the best AI opportunity out there? Maybe.
Start Your Mornings Smarter! Wake up with Breakfast news in your inbox every market day. Sign Up For Free »
But considering shares of Nvidia have gained more than 800% over the last two years, I’m inclined to think the music is going to slow down at some point. Below, I’m going to outline two AI opportunities that I think are poised to break out over the next several years and give Nvidia a run for its money. Let’s dig in!
The first company on my list of top AI stocks is Advanced Micro Devices (NASDAQ: AMD). Over the last couple of years, AMD has been frequently benchmarked against Nvidia — a comparison that I don’t find to be particularly apples-to-apples.
Since the dawn of the AI boom, Nvidia’s primary source of growth has come from its H100 and H200 GPUs. As I alluded to above, Nvidia’s one-two punch GPU architecture helped the company acquire nearly the entire market. While Nvidia’s compute and networking products are indeed quite powerful, one thing that also helped the company gain such an enormous lead in the marketplace was a lack of competition.
AMD has been quietly building its own GPU empire over the last year or so, but it’s nowhere near the size of that of Nvidia. In my eyes, that could soon change. AMD’s answer to Nvidia’s H100 and H200 GPU combo is its own chip accelerator dubbed MI300. When the MI300 launched earlier this year, AMD’s management was guiding for revenue around $2 billion. But during the company’s third-quarter earnings call a few weeks ago, AMD CEO Lisa Su hinted that the MI300 is scaling so quickly that the company’s data center GPU business is now on pace for $5 billion in sales this year.
The best part about this is that many of AMD’s major customers adopting the MI300 architecture are also customers of Nvidia. When you layer on top that there could be more than $1 trillion of AI infrastructure spend over the next few years, AMD looks well-positioned to continue capturing incremental market share as it scales its data center GPU operation.
