Hyperscale Cloud Data Centers in the Era of Specialization

Global reliance on cloud services, powered by transformative technologies such as generative AI, machine learning, and big-data analytics, is driving exponential growth in demand for hyperscale cloud compute infrastructure. Meanwhile, the breakdown of classical hardware scaling (e.g., Moore’s Law) is hampering growth in compute supply. Building domain-specific hardware can address this supply-demand gap, but catching up with exponential demand requires developing new hardware rapidly and with confidence that performance/efficiency gains will compound in the context of a complete scale-out system. This course will build up to the latest developments in hardware-software co-design in hyperscale cloud data centers, focusing on four broad stages: (1) The homogeneous data centers of yesteryear, (2) Early avenues for specialization (e.g., networking, ML accelerators), (3) Scaling specialization (e.g., ML supercomputers), and (4) A view into contemporary cross-stack specialization efforts in data centers.