Our application is similar to excel/google spreadsheets. We don't expect the user to scroll through all that data, but we give tools to do sorting, filtering, plotting, etc. over that data.
One of the reasons we decided to maintain this library is that it does virtualize both columns and rows. We use it at Schrodinger for our enterprise platform and have demo'd support of billions of rows. This is possible because we don't use a wrapper DOM element to handle scrolling.
Right now we are migrating all the state to Redux for better maintainability and performance and hope to make extending and optimizing the library more user friendly.
misiti3780 would you be interested in collaborating with us to help us capture your improvements as well?