The 'Next Nvidia' Trade? Why Investors Are Suddenly Watching Advanced Micro Devices, Arm Holdings, and Marvell Technology - The Globe and Mail

Recent investor attention has shifted from NVIDIA's dominance in the AI chip market to emerging players like AMD, ARM, and Marvell. While NVIDIA controls about 80% of the AI chip market through its GP... Read original →

The AI chip market is undergoing a structural shift from training to inference, eroding NVIDIA’s dominance. Inference workloads prioritize energy efficiency and cost-per-inference, creating strategic openings for ARM-based CPUs, AMD’s MI300 series, and Marvell’s custom ASICs. This shift redirects upstream demand toward heterogeneous integration in packaging and EDA tools, while hyperscalers like AWS and Azure accelerate in-house chip adoption, reducing reliance on GPU generality. Geopolitically, U.S. export controls are accelerating RISC-V development in Taiwan, China and mainland China, indirectly benefiting ARM’s licensing model but complicating its compliance burden. Intel may counter with aggressive Gaudi 4 pricing, while NVIDIA could leverage software moats like TensorRT. Over the next 18 months, inference chip shipments will grow twice as fast as training chips; if ARM and Marvell secure deep integrations with top cloud providers, they could capture over 25% of new AI data center deployments by 2027.