AWS

Amazon Web Services is the world's largest cloud provider and one of the largest buyers of AI GPU infrastructure. AWS purchases NVIDIA H100/H200 GPUs in volume and provides them as EC2 GPU instances (p4, p5 family) to enterprise customers. Simultaneously, AWS designs its own custom AI chips — Trainium (for training) and Inferentia (for inference) — to reduce dependence on NVIDIA and offer lower-cost alternatives. AWS's scale makes it a major demand driver for every upstream supplier in the AI hardware supply chain, from TSMC foundry capacity to SK Hynix HBM memory. The hyperscaler's procurement decisions ripple through the entire supply chain, as multi-year GPU purchase agreements directly determine allocation priorities at TSMC and NVIDIA.