Alex Salkever

The runaway success of ChatGPT, Stabile Diffusion, Midjourney and other generative AI systems has taken the new technology genre mainstream at lightspeed. But even though generative AI might seem like magic, there is a lot that has to happen behind the curtain for the Wizard to work. Already, the ChatGPT busy message is a common occurrence. Generative AI is incredibly compute intensive and combines heavy requirements for both massive data crunching in the cloud and localized, specialized data for specific contexts and use cases. Increasingly, as it becomes a part of every application — which is clearly the intent of Microsoft, the part owner of OpenAI — answers will need to be delivered quickly and crisply — as if we were actually talking to another person. To make this possible, we will need a total overhaul of our technology infrastructure. This talk will cover how the insatiable demand for AI will impact data centers, wireless networks, servers, devices, cloud computing, networking, and many other related aspects of infrastructure.