AMD hits $649.42 high as Meta Muse, OpenAI Dots run on its chips

The personal AI agent boom is giving AMD and Intel new life after years in which Nvidia's GPUs took center stage with the launch of ChatGPT and the generative AI era. CPUs were once the primary workhorse of a computer server, and now they are back, powering long-running agents that can work for hours on their own. Meta debuted Muse in early September, and it topped the Apple App Store in less than two weeks; OpenAI released its agent Dots last week.

The mechanics matter here. GPUs still handle inference, or processing an AI model, but an agent also needs a CPU, because CPUs perform the workflows while GPUs do the thinking. Meta and OpenAI use virtual machines, which partition a server into lighter virtual computers, so one server can support scores of users. Some AMD EPYC processors have up to 192 CPU cores, and CPUs are less expensive than GPUs — Nvidia GPUs can cost more than 10 times a single EPYC chip and are usually sold in clusters of hundreds or thousands.

What the outcome hinges on is whether that cost advantage translates into durable share. AMD's data center revenue more than doubled to $6.7 billion in the quarter ended in June, accounting for almost 60% of total sales, and AMD said in July it expects to take more than half of the CPU market. But Meta describes itself as largely CPU-agnostic by design, OpenAI uses multiple CPU providers, and cloud providers are building custom Arm-based chips — so AMD's current edge looks less like a lock than a lead it has to defend.