Thread by @PatrickMoorhead

I think @awscloud has the clearest compute product positioning:
-"fastest"
-"largest"
-"most"
-"only"
-"best"
#reinvent

Graviton2 growing up a bit today with C6gn. Much higher performance workloads are possible like data analytics, nw appliances, load balancers, some HPC. #reinvent

My recent writeup of @awscloud Inferentia and Inf1. forbes.com/sites/patrickm…

BOOM #1: @awscloud goes with @intel Habana. #reInvent

KABOOM! @awscloud gets into the ML training game with its own silicon. @nvidia finally gets some training competition. #reinvent

The multi-cloud control plane game gets even more fun with AWS ECS and EKS Anywhere. #reinvent

There is a need for different kinds of databases or database optimizations for what you want to accomplish. *But* it's also untenable to manage 12 different databases and easily take advantage of *all* the data. AWS Glue Elastic Views should help. #reinvent

AWS wants to improve ML data prep process with Sagemaker "Data Wrangler." This process is very time consuming and labor-intensive. This could make my top 5. #reInvent

For being so down on hybrid 5 years ago, @awscloud sure does has a wide variety of hybrid options now. #reInvent