This episode is available to stream on-demand.

AI demands vast amounts of compute and swathes of power, with GPUs consuming 10-15 times more energy than their CPU counterparts. To meet these demands, specialized 'AI data centers' have evolved, designed to support high density workloads efficiently. However, these operations face numerous design and construction challenges, and must adapt in line with the constraints of their physical location as well as the workloads they are supporting. This begs the question: how can we build for high density sustainably?

• How can we leverage GPUs and CPUs in the most efficient way?
• Deploying AI at speed and scale without compromising on ESG principles
• Understanding the resource intensiveness of different workloads
• What does efficiency look like to an AI data center?