Beyond Nvidia: Why the Next Phase of AI Could Crown a New Market Leader

Beyond Nvidia, the next phase of AI could shift market leadership as demand moves from training frontier large language models toward inference and agentic AI. Nvidia has led earlier workloads using GPUs and its CUDA software ecosystem, but the article argues the center of gravity is changing as cost-per-inference becomes more important than raw compute. It highlights Advanced Micro Devices (AMD) as a potential beneficiary, citing its chiplet approach that can package more memory for inference. AMD also acquired MEXT, described as using AI to offload infrequently used data to flash memory and restore it to DRAM just before it is needed, aiming to expand memory capacity while reducing costs. For agentic AI, the piece notes data centers may require more CPUs relative to GPUs, and it references Intel projections of a 1 CPU to 1 GPU ratio. The article frames AMD’s opportunity through these two workload shifts.