Advanced Micro Devices (AMD) is raising its expectations for the server processor market as the development of AI agents reshapes data center infrastructure. CEO Lisa Su now expects the market to exceed $200 billion by 2030, growing at an annual compound rate of more than 50% over the coming years. The latest forecast underscores a broader shift in AI investment: demand is expanding beyond graphics processing units (GPUs) used to train and run models toward the central processing units (CPUs) needed to coordinate increasingly complex AI workloads.

AI Agents Change the Economics of Server Computing

The rapid adoption of AI agents is changing how computing resources are deployed. Traditional generative AI applications often rely heavily on GPUs to process model requests, with CPUs managing supporting tasks. Agentic AI systems, by contrast, can break objectives into multiple steps, retrieve information from databases, call external tools, execute software and coordinate several processes before delivering a result. Each additional task creates demand for computing resources beyond the model itself.

AMD estimates that this shift could substantially expand the market for server CPUs. Su has repeatedly raised the company’s outlook as customer discussions and emerging workloads have revealed a larger opportunity than previously anticipated. The company estimated the market at approximately $60 billion in November 2025, raised the projection to more than $120 billion in May 2026, and subsequently lifted it above $200 billion. The revisions illustrate how quickly expectations for AI infrastructure demand are evolving, although they remain forecasts rather than guaranteed market outcomes.

AMD Sees an Opportunity Beyond GPUs

The expanding CPU market could broaden AMD’s position in AI infrastructure. The company competes with Intel in server processors and with Nvidia and other chipmakers across the wider AI computing ecosystem. Its EPYC server CPU family is central to its efforts to capture demand from cloud providers and enterprise customers, while its broader portfolio includes AI accelerators designed to compete in a market dominated by Nvidia.

Su has also emphasized AMD’s ambition to capture more than half of the projected server CPU market. Achieving that target would depend on customers adopting its processors at scale, continued product development and the company’s ability to secure manufacturing capacity. A larger addressable market does not automatically translate into proportional revenue or market share, particularly as rivals respond to changing workloads and data center designs.

Supply Constraints and Forecast Risks Remain

The opportunity comes with operational challenges. AI chip demand is placing pressure on semiconductor manufacturing capacity, advanced packaging and high-bandwidth memory supplies. AMD has indicated that it plans to increase chip supply substantially in 2027, while working with manufacturing and technology partners to prepare for longer-term demand. The pace of that expansion will matter if customers accelerate investment faster than suppliers can respond.

The forecast also depends on the economics of agentic AI. Enterprises must demonstrate that autonomous systems deliver enough productivity gains to justify their infrastructure and operating costs. If adoption slows or spending plans are revised, demand could fall short of current projections. Conversely, broader deployment of agents across business software, cloud platforms and data centers could create a larger and more persistent need for CPUs than earlier AI investment cycles implied.

Investors will be watching AMD’s data center revenue, server CPU orders, production capacity and customer commitments for evidence that the revised market outlook is translating into commercial growth. The key question is whether AI agents become a sustained source of computing demand—or whether infrastructure spending ultimately expands more slowly than the industry’s increasingly ambitious forecasts suggest.