How Lisa Su Has Positioned AMD to Be a Big AI Winner and Why the Stock Is a Buy

In the early days of artificial intelligence (AI), Advanced Micro Devices (NASDAQ: AMD) looked like it would be nothing more than an afterthought to the dominant Nvidia. However, AMD CEO Lisa Su has managed to position the company to be a big AI winner in the coming years.

This started with the bold move of striking a deal with OpenAI, where Su was willing to give the frontier AI model company an up to 10% stake in AMD. In exchange, OpenAI committed to buying 10 gigawatts of AMD's graphics processing units (GPUs) to run inference for its models. It was a deal worth in excess of $100 billion, but more importantly, the size of the commitment would require OpenAI to incorporate its ROCm software platform into its systems. Meta Platforms would sign a similar deal shortly afterward.

At the same time, Su bet that AMD's innovative chiplet design, which is devised to function as part of a larger interconnected system and could be packaged with more memory, would become ideal for inference, a market expected to grow larger than training. She also went out and acquired ZT Systems so the company could start selling complete rack systems.

Fast-forward to today, and AMD recently signed up two new large GPU partners, Anthropic and Microsoft, for its next-generation GPUs. However, AMD no longer had to include equity stakes as part of these arrangements and will instead invest up to $5 billion in Anthropic.

As part of the deal, the maker of the Claude large language model (LLM) will deploy 2 gigawatts of AMD's MI450 GPUs housed within its Helios rack-scale solutions. The Helios racks also include the company's server central processing units (CPUs), networking gear, and ROCm software. The first delivery is expected in the first half of next year. Microsoft will also begin to use the Helios system within its data centers for inference, although terms were not disclosed.

AMD CEO Lisa Su at an event. Image source: Getty Images.

In addition to GPUs, Su has also positioned AMD as a leader in the server CPU market. With the rise of agentic AI, the ratio of GPUs to CPUs is expected to shrink drastically, with projections that it will go from 8:1 for training to 1:1 for agentic AI. At the same time, AMD is making high-core CPUs designed specifically for agentic AI and inference that will come at higher prices. It recently upped its industry CPU forecast, predicting this will become a $220 billion market in the next few years, with a goal of taking 50% market share.

Between its GPU and CPU opportunities, Su has positioned AMD for massive growth in the coming years, making it a top AI stock to buy.

Geoffrey Seiler has positions in Advanced Micro Devices and Meta Platforms. The Motley Fool has positions in and recommends Advanced Micro Devices, Meta Platforms, Microsoft, and Nvidia. The Motley Fool has a disclosure policy.