It's a bit over my head but the gist I'm getting is that you can do more interesting things when you can go from simply predicting the heat dissipation of a component, to controlling that heat dissipation.

While there have been efforts in tuning thermal conductivity, their performances have suffered due to reliance on moving parts, ionic motions, or liquid solution components. This has resulted in slow switching speeds for heat movement on the order of minutes or far slower, creating issues in performance reliability as well as incompatibility with semiconductor manufacturing.