AI Compute

AI compute encompasses chips, memory, cloud services, data centers and power, providing the foundation for model training and inference. Recent developments include Apple’s launch of the M6 and Meta’s procurement of Azure services, while companies including NVIDIA and Anthropic continue to expand their financing and infrastructure strategies. At the same time, HBM bandwidth, energy supply and China’s chip software ecosystem remain bottlenecks. From national computing-power policies and competition among GPU cloud providers to the emergence of compute futures, the cost of supply and ways of accessing capacity are changing. These shifts will affect AI service pricing, semiconductor demand and industry competition, making the sector one to watch.