ML Inferencing At The Edge

[@jaffarbh]

I think we will start to see embedded AI accelerators on DPUs (Data Processing Units). This way it's possible to do inferencing at the network level without relying on host CPUs or GPUs. This means higher throughput and lower latency.