The AMD CEO started the conference call by describing the affect of AI on her firm's cloud computing business. Su explained that "AI was the primary driver of growth in the quarter, as every major cloud provider expanded their Epic footprint to support a broad range of AI workloads, from general-purpose compute and data processing to head nodes for accelerators and emerging agentic applications." Head roles refer to specialized computing clusters in a data center that are responsible for managing resources and conducting other first order tasks to direct the remaining computing resources.

Increasing and agentic AI are increasing the need for server CPU compute, as these workloads require additional CPU processing for orchestration, data movement and parallel execution, in addition to serving as the head nodes for GPUs and accelerators. As a result, we are seeing both stronger near-term demand and deeper engagement with customers on long-term capacity planning.

AMD CEO Is Optimistic About Agentic AI's CPU Demand Complementing GPU TAM Using these, the executive now believes that the total addressable market for server CPUs will now grow at a rate of 35% annually to touch $120 billion by 2020. AMD's previous estimate was an a18% annual growth, and in its recent coverage, investment bank UBS was more optimistic as it outlined a $170 TAM by 2030. At our Financial Analyst Day in November, we outlined a server CPU market growing at approximately 18% annually over the next 3-5 years.

When asked about whether the additional demand for CPUs would end up cannibalizing or complementing the current market for GPUs, Su, naturally, believes the latter to be true. Calling the CPU demand "largely additive to the [GPU] TAM," she added that the additivity stems from the fact that accelerators are needed to run the foundational models and the agents "spawn" CPU tasks.

According to her, the ratio of CPUs to GPUs is important in this scenario. "So if you're installing a gigawatt of compute, you know, the ratio, there's a percentage of CPU as part of that gigawatt will increase. You know, some of the conversation in the industry has been about, you know, CPU to GPU ratios," she said.

However, Su also outlined that eventually the CPU demand could outpace GPU demand if agentic AI becomes more popular:

"And it's very hard to call exactly, but, you know, we certainly see the movement towards, you know, where in the past the CPU to GPU ratio was primarily, you know, just as a host node, you know, in like a one to four or one to eight configuration, you know, now changing and getting closer to a one-to-one configuration or, you know, even, you know, you can even imagine if you get lots and lots of agents that you could have more CPUs than GPUs."

Yet as of now, the key win, according to her, is that "veryone is now planning and thinking about CPUs at the same time that they're thinking about, you know, their accelerator deployment, which is a good thing."

By: DocMemory
Copyright © 2023 CST, Inc. All Rights Reserved