Nvidia Already is the GPU Giant. Now It's Aiming to Dominate in a New $200 Billion Market.

The market is still treating Nvidia primarily as a GPU monopoly, but the more important shift is that it is turning into the default compute layer for AI workloads across price/performance bands. If the stand-alone CPU push gains traction, the implication is not just incremental revenue — it is a broader capture of workload architecture decisions, which raises switching costs and compresses the addressable space for Intel and, more selectively, AMD. The second-order effect is that Nvidia can monetize more of the stack at the moment customers are redesigning infrastructure for agentic AI, when procurement standards are least entrenched.

The key catalyst is not a single product launch but a change in mix: CPUs are lower ASP than top-end GPUs, yet they can be a wedge into deployments that do not justify accelerator-heavy spend. That matters because AI capex is likely to broaden from training-dominated clusters to inference-heavy, always-on agent systems over the next 6-18 months. In that transition, energy efficiency and total cost of ownership will matter more than peak throughput, which favors a supplier that can bundle CPU, GPU, networking, and software into one purchasing decision.

The market may be underpricing the competitive damage to Intel, where this is another proof point that AI does not require a historic CPU incumbent to own the architecture. For AMD, the risk is more subtle: even if it wins share in traditional server CPUs, Nvidia’s ecosystem can lock the highest-value AI deployments before AMD gets a chance to participate. The upside for Nvidia is that this creates a larger TAM, but the risk is execution — if Rubin slips or the stand-alone CPU product disappoints on software enablement, the stock could de-rate on the grounds that the narrative outran the shipment curve.