Infrastructure and operations teams must enable two new areas for delivery in order to be in the mainstream of compute infrastructure: high-performance computing and AI. In this podcast, we explore the realities of the AI and generative AI demands on I&O and recommend actions to take.
As we think about the continued fervor over AI and generative AI (GenAI), 2024 is shaping up to be very different from 2023 in three specific ways:
Urgency and action — First, what we are starting to see among our clients is a sense of urgency, where enterprises are shifting from exploration to action. Last year was all about ideation. This year seems to be a lot more about implementation.
Technology stack — The technology stack continues to evolve across multiple layers. At the silicon layer, we are starting to see new AI supercomputing innovations and new technologies from cloud providers. In the application layers, we see application-specific integrated circuits (ASICs). There are also networking innovations as well as shifts at the model layer (from large language models to multimodal models).
Emergence of agents — We are seeing the emergence of a new area of agents as well as “agent-to-agent” ecosystems focused on connecting and planning approaches to reasoning for purposes of taking action. All of this is being combined with context and memories integrated with our systems and software.
In this podcast, our expert analyst Chirag Dekate shares his insight and recommended actions to help I&O leaders deal with the realities of AI and generative AI in all three of those dimensions.
About the Guest:
Host Frances Karamouzis is joined by our expert analyst, Chirag Dekate. Chirag’s research focuses on providing strategic advice on generative AI systems, engineering AI pilots into production across a hybrid and multicloud context with an emphasis on AI (generative AI) infrastructures, quantum technologies (quantum computing, quantum sensing, quantum networking), high-performance computing, and advanced analytics infrastructures (quantum computing, neuromorphic, GPUs and beyond).