A Strategic Pivot Beyond Graphics Processors
For years, the conversation around artificial intelligence hardware has revolved almost exclusively around graphics processing units. GPUs became the symbol of the AI revolution, the workhorses behind model training and the engines powering the most ambitious machine learning projects. Yet the landscape is now shifting in a way that few would have predicted just a couple of years ago. A new chip called Vera is being pitched as a major growth driver, opening up a brand new $200 billion total addressable market that had previously been untouched by certain dominant players in the AI hardware space.
This is not merely a long-term vision or speculative roadmap. There is already visibility into nearly $20 billion in standalone CPU revenue this year alone, with major hyperscalers and system makers signing on as partners to deploy these chips. The scale of this initial traction signals something deeper than a product launch — it represents a strategic recognition that the future of AI infrastructure will be more heterogeneous than the GPU-centric narrative has suggested.
Why CPUs Matter for Agentic AI
The key driver behind this market expansion is the rise of agentic AI. Unlike earlier paradigms focused on training large models, agentic AI involves systems that act autonomously, reasoning through problems, executing multi-step workflows, and interacting with users and other systems in real time. In this new paradigm, CPUs — not GPUs — handle the orchestration layer. They manage the workflows, coordinate reasoning steps, and enable the real-time interactions that make agentic systems feel responsive and capable.
This division of labor matters enormously for the economics of AI infrastructure. While GPUs excel at parallelized mathematical computation, the coordination logic that ties together complex AI agents requires the kind of general-purpose processing that CPUs deliver. As AI systems graduate from being passive predictors to active agents, the demand for high-performance orchestration hardware grows in lockstep.
Bullish Forecasts From Wall Street
Analysts have taken notice of this structural shift, and their projections reflect a market that is being fundamentally re-rated. Bank of America now sees the server CPU total addressable market reaching approximately $125 billion by 2030. Citi is even more aggressive in its outlook, projecting around $132 billion driven by the fast-growing demand for agentic AI workloads. These are not modest upward revisions — some analysts have noted that CPU market estimates have already been raised sharply as the AI industry transitions from training models to actually deploying them at scale, a phase where CPUs play a critical role.
This shift from training to deployment is particularly important. Training is a finite, intensive process; inference and orchestration, by contrast, are continuous and scale with usage. As more enterprises move AI from experimental pilots to production environments running agentic workloads around the clock, the cumulative demand for CPU capacity compounds rapidly.
Repositioning at the Center of a New Opportunity
What makes this development so notable is the strategic logic behind it. Rather than simply defending an existing position in GPU dominance, a calculated move into the CPU market positions a company at the center of a new and rapidly expanding opportunity. The Vera chip is not a defensive product designed to plug a gap — it is an offensive play to capture share in a market that did not previously contribute to revenue.
The broader lesson is that the AI hardware story is becoming richer and more layered. The future will not be decided by a single chip category but by the ability to deliver integrated systems where GPUs, CPUs, and other specialized processors work in concert. Companies that can credibly offer this full stack stand to benefit from the entire spectrum of AI workloads, from raw model training to the nuanced orchestration that agentic AI demands. In that sense, the CPU is no longer the supporting actor in the AI revolution — it is stepping into a leading role of its own.