Nvidia: The king of AI

Two of the biggest stock winners of the past decade are Nvidia (NVDA 1.86%) and Advanced Micro Devices (AMD +2.18%), which are up over 16,320% and 9,770%, respectively. The two chip rivals have been major beneficiaries of the artificial intelligence (AI) infrastructure boom, and both remain well-positioned for the long term.

However, the question is which AI stock looks like the smarter long-term buy right now. Let’s dig in to find out.

Nvidia: The king of AI

Nvidia was the biggest winner in the initial phase of AI, as its graphics processing units (GPUs) became the primary chips for training large language models (LLMs). The company’s advantage in this area stemmed from its CUDA software platform, which it developed to expand the use of its chips beyond their initial purpose of accelerating graphics rendering in video games.

While it took time to unfold, Nvidia smartly seeded CUDA into universities and research labs that were doing early work on AI. As a result, developers learned to program GPUs using CUDA, and most foundational AI code was written on its software platform and optimized for its chips.

This dynamic continues to give Nvidia a sizable moat with AI model training today. In the interim, the company also built out a powerful data center networking business through its 2020 acquisition of Mellanox, and more recently, it “acquired” Groq and its language processing unit (LPU) technology for inference. It has also developed its own ARM-based central processing units (CPUs). This has helped transform Nvidia from a maker of GPUs into a complete AI infrastructure company.

Nvidia offers its customers end-to-end AI server solutions configured for specific AI tasks, such as training, inference, and agentic AI. Nvidia’s revenue growth has already been eye-popping, with 85% growth in Q1, and these offerings position the company for strong future growth.

Despite its growth and strong positioning, Nvidia’s stock remains attractively priced, trading at a forward price-to-earnings (P/E) ratio of under 16 times fiscal 2028 (ending January 2028) analyst estimates.

AMD: Riding two powerful trends

AMD has largely been in the shadow of Nvidia during the training phase of AI, as it was unable to overcome its larger rival’s CUDA advantage. However, it has greatly improved its ROCm software platform over the past few years, and it is much better positioned for inference, which is much less technically demanding than AI model training. Inference also tends to be less about raw compute power and often more about fast access to memory.

AMD’s chiplet design allows for a larger KV (Key-Value) cache and for more memory to be packaged with its GPUs than those from Nvidia. Meanwhile, the company just announced it was acquiring memory optimization company MEXT, which uses AI-driven software to increase memory capacity while lowering costs without impacting performance. Its technology moves infrequently accessed data from high-cost DRAM (dynamic random-access memory) to unused flash memory, and then uses AI to predict when that data will be needed and transfers it back before it is even requested. Together with its prior acquisition of ZT Systems, this will allow AMD to offer end-to-end servers for inference at an attractive cost.

The company already has two $100 billion GPU inference deals in place, and it looks poised to become a major player in this arena. That’s good news, as the inference market is expected to eventually grow larger than the training market.

In addition to its inference opportunities, AMD also has a major opportunity in agentic AI. The company is a leader in data center CPUs, and this market is set to explode, as CPUs are needed to provide the sequential logic needed to manage AI agents. The GPU-to-CPU ratio is expected to go from 8:1 for training to 1:1 for agentic AI, and AMD sees this as a $120 billion addressable market.

The verdict

With a 39.5x one-year forward P/E, AMD is much more expensive than Nvidia, but with a market cap of less than $900 billion, it is a much smaller company than its $5 trillion rival. With two huge opportunities ahead of it, it has significant upside, while Nvidia could eventually run into the law of large numbers in terms of revenue growth.

I really like both stocks, but I think AMD could have more upside over the next decade, given its smaller size and its positioning with two big emerging trends.