Compute | New York // Tech Showcase: Differentiated IT architecture in the age of AI

With AI estimated to double the power demand of data centers in the next couple of years as models are put into production, an architecture-first strategy is needed that takes a total system, full stack view, with hardware and software tightly coupled. This approach helps prioritize and optimize overall system performance and resilience to support large-scale, mission-critical applications.

In this session we’ll discuss how today’s enterprise computing requires an architecture that has extensions and accelerators that integrate with the core systems, where the processor units (PUs), memory, input/output (I/O), network communications , and data center infrastructure are uniquely integrated, enabling them to process and scale on demand – while saving data center space and power costs. We’ll highlight how the mainframe is playing a bigger role in the AI story, and how it is evolving to accommodate and accelerate newer innovations, including the next crop of AI accelerators that are expected to be significantly enhanced—with each accelerator designed to deliver 4 times more compute power.