I work on systems that deal with PBs of data every day, so we'd stand to benefit significantly from .NET Core improvements for numerous reasons. At that level of scale, reducing memory copying would be the biggest win for us.

For the vectorization space specifically, .NET Core string operations like Equals and IndexOf are (now) vectorized, looks like Bing took advantage of that: https://devblogs.microsoft.com/dotnet/bing-com-runs-on-net-c...

Never was big into ML, but I know people that do a lot of it in Python or R. If those mathematical libraries could be ported to .NET, I wonder how big the improvements would be.