In our latest Tech Talks Daily Podcast episode, I sit down with Seamus Dunne from Digital Realty to explore the accelerating demand for compute power driven by AI and generative AI technologies. As these innovations push the boundaries of what's possible, the infrastructure supporting them must evolve at a matching pace—or risk being left behind.
Seamus delves into the heart of the matter, explaining how Digital Realty responds to the rising demand for high-performance infrastructure like GPUs and AI chips. He shares insights on how AI drives the need for robust data storage and efficient data access solutions. The discussion touches on the strategic importance of local investment in infrastructure to maintain competitive advantage, especially in regions like the UK.
We also explore the types of AI workloads becoming prevalent across various sectors. Currently, a significant portion of these workloads is dedicated to training models. Still, Seamus anticipates a shift towards inferencing as AI models mature and become integral to advanced products and services. This shift necessitates a robust and adaptable IT infrastructure strategy to deploy AI models alongside cloud services effectively.
Seamus highlights how data centers are pivotal in our daily digital activities—from streaming Netflix and conducting Zoom calls to online shopping and planning holidays. Beyond everyday conveniences, data centers are the powerhouse behind generative AI and other innovations driving economic growth.
Seamus discusses the evolving needs of AI infrastructure, emphasizing the importance of proximity to data centers for low-latency performance, particularly for AI inferencing. He also touches on the geographical advantages and challenges, noting how regions like the UK are well-positioned for AI growth while others, like Dublin, may need more support due to regulatory constraints.