As AI workloads increase, there's an intense demand and need for high capacity memory chips presenting a technical challenge known as the memory wall. Memory is also a battle ground for the top makers of AI accelerators. Professor Arijit Raychowdhury of the School of Electrical and Computer Engineering, Georgia Institute of Technology speaks to Bloomberg's Ed Ludlow on "Bloomberg Technology." Raychowdhury weighs in on Nvidia's H100 and AMD's MI300X.