Lisa Su just told the world that CPUs are about to have a very good half-decade. The AMD chief executive projected that the server CPU market will grow at more than 35% annually over the next five years, a figure that would push the total addressable market past $120 billion by 2030.

Here’s the thing: this isn’t just optimism from a CEO talking her book. AMD’s previous estimate, shared at the company’s November 2025 analyst event, pegged annual growth at 18-20%. The new number nearly doubles that. Something meaningful shifted in AMD’s calculus, and it has everything to do with how AI workloads are evolving.

Agentic AI is rewriting the hardware playbook

For the past few years, the AI hardware conversation has been almost entirely about GPUs. Nvidia became the most valuable company on the planet largely because training massive language models requires absurd amounts of parallel processing power, which is what GPUs excel at.

But training a model and actually running it in production are two different animals. Inference, the process of deploying trained AI models to generate responses, doesn’t always need the same brute-force GPU clusters. And agentic AI, where autonomous software agents chain together multiple reasoning steps to complete complex tasks, is pushing demand toward more balanced computing architectures.

In English: as AI moves from “build the brain” to “use the brain at scale,” CPUs become a much bigger part of the equation.

Su made these remarks during public appearances on May 22, 2026, including an event in Taipei. The timing matters. Taiwan is the epicenter of semiconductor manufacturing, and signaling this kind of growth projection there sends a message to the entire supply chain: prepare for volume.

The shift toward inference-heavy and agentic workloads means data centers can’t simply stack more GPUs and call it a day. They need sophisticated CPU architectures to manage orchestration, memory handling, and the kind of sequential logic that agents perform constantly. This is precisely where AMD’s EPYC server processors fit into the picture.

AMD’s data center momentum backs up the forecast

Su’s projection would ring hollow if AMD’s own numbers weren’t cooperating. They are.

AMD’s data center segment, which includes its server CPU business, has posted consistent year-over-year revenue growth. The company has been steadily gaining market share against Intel in the server space, a trend that accelerated as EPYC processors earned credibility with hyperscale cloud providers and enterprise customers.

Consider the transformation arc. When Su took over as CEO in 2014, AMD was a company fighting for relevance, burning cash, and losing ground on nearly every front. A decade later, it’s a credible competitor to both Intel in CPUs and Nvidia in accelerators. That turnaround is one of the more remarkable leadership stories in recent tech history.

The previous 18-20% growth estimate was already aggressive by historical standards. Server CPU markets don’t typically grow at that pace. Doubling the projection to 35% suggests AMD is seeing something in its order pipeline and customer conversations that goes beyond incremental demand improvement. It suggests a structural change in how data centers are being built.

A $120 billion TAM by 2030 would represent a massive expansion from current levels. For context, that’s the kind of market size that attracts not just existing players but new entrants, custom silicon efforts from cloud providers, and a whole ecosystem of supporting infrastructure companies.

What this means for investors

Look, projections from a CEO should always be taken with appropriate skepticism. Lisa Su has every incentive to paint a rosy picture of the market her company serves. But the underlying logic here is sound, and it’s corroborated by broader industry trends.

The AI infrastructure buildout has been overwhelmingly GPU-centric. If agentic AI and inference workloads genuinely shift the balance back toward CPUs, that’s a meaningful tailwind for AMD’s highest-margin business. It also introduces a competitive dynamic that investors should monitor carefully.

Intel, despite its struggles in recent years, still holds the majority of the server CPU market. A rising tide of 35% annual growth would lift Intel’s prospects too, at least in theory. But AMD has been the share-gainer in this space, and a rapidly expanding market gives it room to grow revenue without needing to take every point of share from its rival.

Then there’s the Nvidia factor. If CPUs become a more critical component of AI infrastructure, the narrative that Nvidia is the sole gatekeeper to AI computing gets more nuanced. That doesn’t mean GPU demand falls. It means the overall compute pie gets bigger, and CPUs claim a larger slice of it.

The risk, naturally, is execution. A $120 billion market opportunity means nothing if AMD can’t deliver competitive products on schedule. Custom silicon from Amazon, Google, and Microsoft also looms as a wildcard. These hyperscalers have been designing their own chips for years, and a booming CPU market gives them even more reason to accelerate those efforts.

Investors should watch AMD’s data center revenue trajectory over the next two to three quarters for early confirmation of this thesis. If the 35% growth rate is real, it should start showing up in order volumes and revenue guidance well before 2030. The gap between Su’s previous estimate and her current one is enormous, and the market will want to see receipts.